<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Alex Murphy on Medium]]></title>
        <description><![CDATA[Stories by Alex Murphy on Medium]]></description>
        <link>https://medium.com/@alex_88610?source=rss-4fa7a7001f6b------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*93dJp_h841QVy6xMBzkehA.png</url>
            <title>Stories by Alex Murphy on Medium</title>
            <link>https://medium.com/@alex_88610?source=rss-4fa7a7001f6b------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 13:20:54 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@alex_88610/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Trust is the right to stop paying attention]]></title>
            <link>https://medium.com/design-bootcamp/trust-is-the-right-to-stop-paying-attention-1b0016f148ed?source=rss-4fa7a7001f6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/1b0016f148ed</guid>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[customer-strategy]]></category>
            <category><![CDATA[business-strategy]]></category>
            <category><![CDATA[trust]]></category>
            <category><![CDATA[empathy]]></category>
            <dc:creator><![CDATA[Alex Murphy]]></dc:creator>
            <pubDate>Sun, 08 Nov 2026 21:41:08 GMT</pubDate>
            <atom:updated>2026-11-09T12:10:36.746Z</atom:updated>
            <content:encoded><![CDATA[<p><em>The most empathetic thing you can do for a customer isn’t to understand how they feel. It’s to give them permission to stop thinking about you.</em></p><figure><img alt="A pink puppet heart suspended against a solid black background, with visible strings attached to its upper edges." src="https://cdn-images-1.medium.com/max/1024/1*28LTePQgpJYlFKXYmz__RA.png" /></figure><p>Most advice on customer trust is about earning it: be transparent, keep your promises, communicate. All true, and all missing what trust is actually for. Trust is what lets a customer stop thinking about you. Which means a surprising amount of what we call good service is quietly making customers think harder.</p><p>Think about the things in your life you never check. You don’t test the tap water before you drink it. You don’t count the cash the ATM hands you. You don’t read your power bill line by line, at least not until the month it was wrong.</p><p>Now think about the things you do check. The insurer you ring with a pen in your hand, ready to write down a name and a reference number. The utility whose chat you screenshot before closing it, because last time “the notes on your account” said something different. The courier whose tracking page you refresh like it owes you money.</p><p>Nobody is born keeping notes on their insurer. Somebody taught them to.</p><p>Here is the least romantic, most accurate definition of customer trust you will read this year: it is the moment a customer stops keeping notes.</p><h3>Trust is a machine for not thinking</h3><p>In 1968 the German sociologist Niklas Luhmann argued that trust is not mainly a feeling. It is a way of reducing complexity. The world contains far too many things that could go wrong for anyone to monitor them all, so we trust. We hand the monitoring to someone else and get our attention back.</p><p>Cognitive psychologists have a name for the same move: cognitive offloading. We write things on our hands, set reminders and save passwords so our brains don’t have to hold them. Trust is offloading onto a person or an organisation. <em>You</em> keep track of this, so I don’t have to.</p><p>That reframes the whole job. A customer who trusts you is not a customer who feels warmly about you. It is a customer who has stopped checking. The pen, the screenshot and the reference number are not quirks. They are what distrust looks like when it does the paperwork.</p><p>And it is expensive. Not for you. For them. Every note is attention the customer is spending because you couldn’t be relied on to hold it.</p><h3>Your empathy is giving customers homework</h3><p>CX has spent a decade teaching people to perform empathy. “I completely understand how frustrating that must be.” Acknowledge, validate, reassure. Fine. But notice what that empathy asks of the customer: to engage, emotionally, with you, about the problem. It is empathy that makes the customer do more work.</p><p>Compare two airlines on the same cancelled flight.</p><p>The first sends a beautifully written message. It is very sorry. It understands this is disruptive. Please visit the app to explore your rebooking options.</p><p>The second sends one line: your flight was cancelled, you’re on the 4:10pm instead, your seat and bags have moved with you, nothing to do.</p><p>The first airline is more sympathetic. The second is more empathetic. It took the thinking off the customer, which is the only part of the problem the customer actually cares about. The first handed over homework in a nicer font.</p><figure><img alt="A black paper storm cloud with ligntning coming from it. A pink gloved hand pointing to the cloud on a pink background." src="https://cdn-images-1.medium.com/max/1024/1*hcruQNtvfLUCG-AocuZ28w.png" /></figure><h3>Not all communication builds trust. Some of it cancels it.</h3><p>The standard CX reflex is that more communication equals more trust. Keep them informed. Over-communicate. Never leave them wondering.</p><p>Except every message reopens the file. “Your parcel has left the depot.” “Your parcel is at the regional hub.” “Your parcel is out for delivery.” “Your parcel is still out for delivery.” Seven messages for a phone case. Each one hauls the customer’s attention back to a thing they had happily stopped thinking about, and the cumulative message is not “we’ve got this”. It is “keep an eye on this, something might go wrong”.</p><p>The useful distinction isn’t more versus less. It is opening versus closing.</p><p>A message that <strong>opens</strong> a loop asks the customer to watch, check, decide, remember or act. “Action may be required.” “Please review the attached.” “We’ll be in touch.”</p><p>A message that <strong>closes</strong> a loop lets them stop. “Done. Nothing for you to do.”</p><p>Trustworthy organisations send mostly closing messages. Untrustworthy ones send opening messages, then wonder why customers ring to “just check”.</p><h3>Nobody in the room is paid to let customers forget them</h3><p>If this is so obvious, why do organisations keep doing the opposite? Because almost every incentive in the building rewards customer attention.</p><p>Marketing wants touchpoints. Product wants app opens and daily active users. Legal wants a disclosure on every screen and a tick box before every step. The engagement dashboard goes green when customers think about you more often. For a social media app, fair enough. For a power company, an insurer, a bank or a broadband provider, the ideal engagement number is somewhere close to zero.</p><p>The CX team didn’t set any of those targets. They inherit the result: customers trained to stay vigilant, who call to confirm, screenshot every chat and ask for a reference number “just in case”. Then the CX team gets asked why customer effort scores are so high.</p><p>There’s a cruel asymmetry here too. One of the more reliable findings in psychology is that bad experiences weigh far more heavily than good ones. Trust is built slowly, by many uneventful months, and destroyed in one bad week. And once a customer has started keeping notes, they rarely stop. The screenshots carry on long after the problem is fixed. You only have to teach someone to watch you once.</p><h3>Not thinking is not the same as not noticing</h3><p>There is an obvious way to get this wrong, and plenty of organisations have already found it.</p><p>If trust is permission to stop paying attention, then inattention is also something you can farm. The subscription that renews quietly at a higher price. The introductory rate that expires without a reminder. The loyalty penalty, where the customers who stopped checking are precisely the ones paying the most. These businesses have also worked out that customers would rather not think about them. They’ve simply decided to profit from that instead of earning it.</p><p>Behavioural economists call the deliberate friction that makes good choices hard “sludge”. Richard Thaler popularised the term and Cass Sunstein later wrote a book on it. Sludge is the cancellation that needs a phone call during business hours. Exploited inattention is its quieter cousin. Nothing is hard. You just weren’t looking.</p><p>The line between the two is easy to state. When a customer stops thinking, whose interests does your organisation act in? Trust is a customer handing over their attention on the understanding you’ll use it on their behalf. Spending it on yourself isn’t clever revenue management. It’s a breach, and customers always find it eventually, usually on the one day they finally check.</p><p>So here’s the test for anything that happens automatically: if the customer were watching, would they be glad? A rebooked flight, a refunded fee, a cheaper plan applied because their usage dropped: yes. A quiet price rise on renewal: no. If the honest answer is no, that isn’t a closing message. It’s a trap with good manners.</p><h3>What to do instead</h3><p>Stop asking how to make customers trust you. Start measuring how often they have to think about you, and drive that number down.</p><p>Two things, this month.</p><p>First, run a homework audit. Take one common journey, a claim, a move, a cancellation, and print every message and step the customer encounters. Highlight everything that asks them to check, remember, decide or do something. Next to each highlight, write one question: could we have done this for them? Most of the yellow will turn out to be work you handed over because it was cheaper to send a message than to fix the process.</p><p>Second, start counting checking behaviour. Requests for a reference number. “Just confirming” calls. Repeat logins to the tracking page. Customers emailing to ask whether their email arrived. Every one of those is a customer who doesn’t trust you enough to stop paying attention. That’s your real trust score, and unlike your NPS, it tells you exactly where to look.</p><p>Then rewrite your communications with one test: does this message let the customer stop thinking, or make them start? If it’s the latter and you can’t fix the process behind it, at least have the decency to say what they need to do and when they can stop.</p><p>The goal isn’t a customer who feels reassured. It’s a customer who has put the pen down. Give them their attention back. That’s what trust is for.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1b0016f148ed" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/trust-is-the-right-to-stop-paying-attention-1b0016f148ed">Trust is the right to stop paying attention</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Stop Telling Customers How They Feel: New Research Exposes the Arrogance Hiding Inside…]]></title>
            <link>https://medium.com/design-bootcamp/stop-telling-customers-how-they-feel-new-research-exposes-the-arrogance-hiding-inside-c2a696895aa5?source=rss-4fa7a7001f6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/c2a696895aa5</guid>
            <category><![CDATA[content-strategy]]></category>
            <category><![CDATA[customer-service]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[business-strategy]]></category>
            <dc:creator><![CDATA[Alex Murphy]]></dc:creator>
            <pubDate>Wed, 14 Oct 2026 20:39:22 GMT</pubDate>
            <atom:updated>2026-11-10T23:31:25.008Z</atom:updated>
            <content:encoded><![CDATA[<h3>Stop Telling Customers How They Feel: New Research Exposes the Arrogance Hiding Inside Customer-Centricity</h3><figure><img alt="Glossy hot pink crystal ball with a glowing question mark inside, flanked by two oversized pink hands, rendered as a surreal 3D illustration on a black background." src="https://cdn-images-1.medium.com/max/1024/1*VV_EGeW7FFjsLJESXGMsCA.png" /></figure><h3>Three new studies suggest customer-centricity works best when a business admits it doesn’t actually know what happens next.</h3><p>Three pieces of research landed in the last few months: on recommendation language, on carbon labels, and on how frontline staff get treated. None were written to agree with each other. Put side by side, they agree anyway. The fortune-telling trade has a worse hit rate than anyone still charging full price for it wants to admit.</p><p>“You’ll love this.” “We know what our customers want.” “Our customers value sustainability.” “Our people deliver the same experience every time.”</p><p>Every one of those sentences is customer experience dressed up as confidence. Underneath the tailoring, it is a guess wearing someone else’s certainty. Customer experience was supposed to be in the listening business. Somewhere along the way a good deal of it quietly switched trades to fortune-telling, and kept the old job title.</p><h3>Meet the customer who does not exist</h3><p>Every organisation of a certain size keeps one on staff. She lives in the strategy deck under a name like Sarah, with an age bracket, a few “pain points,” and an uncanny habit of wanting exactly what the roadmap had already planned for Q3. She is consistent. She is rational. She agrees with the segmentation model, because she is the segmentation model, wearing a cardigan.</p><p>The customer who actually exists is nothing like her. She is distracted, contradictory, in a hurry, and fiercely protective of the feeling that she is choosing something rather than being handled. Fortune-telling works beautifully on Sarah. It keeps missing on everyone else, for the excellent reason that everyone else is real.</p><h3>“You’ll love this” is a prediction, not a compliment</h3><p>Start with recommendations, where the fortune-telling is least disguised. Jia Yan, Lisa Guntzviller and Elisabeth Bigsby, researchers in communication at the University of Illinois Urbana-Champaign, <a href="https://doi.org/10.1080/03637751.2026.2687654">published a study in <em>Communication Monographs</em></a> this August that tested the actual wording of a recommendation on actual people, rather than assuming. 1,229 participants were given a health recommendation, eat more vegetables, move more, from one of four sources: a friend, a doctor, a trusted health organisation, or a brand with no obvious business being in the conversation. Some messages predicted how the listener would feel. Others simply offered.</p><p>The predictions did worse. Not identically everywhere, a friend gets more licence than an insurance company, and a doctor gets more than a cereal brand, but consistently across every group studied, the more a message told someone what they were about to think or feel, the harder it triggered reactance: the flinch of a person who senses their own choice being made for them without being asked. Certainty, delivered as fact about someone else’s future feelings, is the one ingredient that reliably makes a good recommendation worse at its only job.</p><p>This was a study about vegetables and exercise, not loyalty points or software trials, so the leap to retail is mine, not the researchers’. It holds up because the mechanism does not care what is being recommended. The moment “I found this useful” turns into “you’ll find this useful,” a business has stopped describing its own experience and started dictating someone else’s. Most people resent being told what they are about to feel, even about a discount code.</p><p>The fix costs nothing and flatters no copywriter: hand the sentence back to its rightful owner. “This saved me twenty minutes” survives a sceptical reader. “This will save you twenty minutes” asks them to take your word for their own future. One is a fact. The other is a tarot card with a logo on it.</p><figure><img alt="A pink-gloved hand points towards a dark storm cloud with a pink lightning bolt against a vibrant pink background, symbolising negative emotions and the power to influence how others feel." src="https://cdn-images-1.medium.com/max/1024/1*hcruQNtvfLUCG-AocuZ28w.png" /></figure><h3>Knowing is not the same as doing, and the gap has a price tag</h3><p>The second trick fortune-tellers rely on is treating a reading as if it were the future. Maiken Maier and Lukas Paul Fesenfeld of the University of Bern <a href="https://www.sciencedirect.com/science/article/pii/S0306919226001168">tested that directly in a pre-registered field study</a>, published in September, tracking more than 2,300 Swiss residents and over 5 million real purchases against Migros’s carbon label across three years of shopping data.</p><p>The label worked exactly as advertised, right up until the point where it needed to. Awareness rose. Attitudes improved. Stated intentions climbed. Every box a stakeholder update loves to tick got ticked. Then the researchers checked actual trolleys, and the story shrank considerably. Overall meat consumption barely moved. Labelled products crept up by roughly a single percentage point, with a little swapping within categories and almost nothing between them. The authors’ own verdict was “modest and heterogeneous,” their polite way of saying the label told the truth about what people believed and nothing reliable about what they would do.</p><p>This is not an argument that carbon labels are useless, or that telling people things is a waste of ink. It is a correction to the specific, expensive habit of treating a survey answer as a receipt. A customer’s actual choice happens inside what the researchers call a “stimulus-intensive” environment: price, habit, taste, and whatever is already sitting in the trolley, all shouting at once. Understanding does not switch any of that off. It just adds one more voice to a room that was already loud.</p><p>If your dashboard shows rising agreement with “I care about sustainability” and someone is treating that as a sales forecast, they are reading a palm and calling it a P&amp;L.</p><h3>The brand ate the human, and the customer noticed</h3><p>The third sleight of hand is the one most businesses never clock as a trick at all: making staff interchangeable, then acting surprised when customers treat them that way. Elizabeth Trinh and Julia Lee Cunningham of the University of Michigan’s Ross School of Business <a href="https://www.sciencedirect.com/science/article/pii/S0749597826000440">published six studies in <em>Organizational Behavior and Human Decision Processes</em></a> this September, including a six-week field deployment with 632 supermarket employees and three further experiments across car sales, hotel, and restaurant settings.</p><p>The intervention was almost embarrassingly small: a name badge extended with a prompt like “ask me about my favourite hobby,” a photo, a personal detail let slip in conversation. Nothing touched scripts, training, or targets. Just a crack of individuality let through the corporate laminate. Customer mistreatment, measured by trained observers and by the staff themselves, dropped meaningfully in every study. The mechanism was perceived humanness: the cue made employees register as people rather than props, and that registration did the protective work. The researchers also ruled out the obvious alternative explanations, including that it was simply politeness norms doing the job.</p><p>Real limits apply. The field study’s response rate was low, which should soften how far the numbers travel. Two of the lab studies restricted natural conversation to isolate the effect, an artificial baseline, though the result held in the one study that let people talk freely. This is one protective mechanism among several, not a licence to stop worrying about customer abuse.</p><p>What it exposes is the unpriced cost of standardising people. Uniforms and scripts make service consistent, and consistency is worth something. But the same choices that make an employee interchangeable with the brand also make it easier for a customer to treat them as an obstacle rather than a person. Nobody is rude to “checkout operator #4” the way they are rude to Dave, who collects vinyl and told you so.</p><h3>A more honest kind of customer-centricity</h3><p>Line the three up and the pattern has nothing to do with tone of voice guidelines or empathy training. Fortune-telling is the expensive habit, not the cheap shortcut it looks like. Predicting how a customer will feel about a message burns trust for a sliver of extra conversion. Predicting that a customer’s behaviour will follow their survey answer burns credibility with your own board when sales refuse to follow the slide. Predicting that staff are interchangeable with the brand burns the people doing the job, quietly, in attrition numbers nobody reads until the exit interviews pile up.</p><p>A genuinely customer-centric organisation does five unglamorous things instead. It recommends from its own experience rather than forecasting someone else’s. It designs the purchase environment around what people actually do, and treats a stated intention as one flickering signal among many rather than a promise. It keeps frontline staff visibly, awkwardly human instead of filing them down into uniformity. It gives customers real control over decisions, not a longer menu of pre-approved options dressed up as agency. And it trades “we know our customer” for the much less quotable truth: we are continually finding out how customers behave, under these conditions, this month, subject to revision without notice.</p><p>Nobody is putting that second sentence on a mission statement wall. It is, nonetheless, the only one of the two that has ever been true.</p><p>Customer experience was never meant to be a prediction engine. It was meant to build the conditions where better choices and better interactions become more likely, and then get out of the way enough for both the customer and the employee to stay recognisably, inconveniently human. The fortune-telling business is booming because certainty sells better than curiosity does.</p><p>It just doesn’t work better. That is the whole report, and nobody is issuing a refund.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c2a696895aa5" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/stop-telling-customers-how-they-feel-new-research-exposes-the-arrogance-hiding-inside-c2a696895aa5">Stop Telling Customers How They Feel: New Research Exposes the Arrogance Hiding Inside…</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Most CX Is Theatre. I Wrote a Book About What Happens After the Applause]]></title>
            <link>https://medium.com/design-bootcamp/most-cx-is-theatre-i-wrote-a-book-about-what-happens-after-the-applause-ca4982666fa0?source=rss-4fa7a7001f6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/ca4982666fa0</guid>
            <category><![CDATA[book-recommendations]]></category>
            <category><![CDATA[culture]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[business-strategy]]></category>
            <dc:creator><![CDATA[Alex Murphy]]></dc:creator>
            <pubDate>Thu, 01 Oct 2026 02:06:01 GMT</pubDate>
            <atom:updated>2026-11-09T12:10:31.278Z</atom:updated>
            <content:encoded><![CDATA[<h4>Have you tried turning the customer off and on again? is out now, and it is a love letter to customer experience written with one eyebrow raised.</h4><figure><img alt="A black-background promotional image showing Alex Murphy’s book, Have You Tried Turning the Customer Off and On Again?, angled slightly from above. A glossy yellow faceless figurine with a ponytail leans against the left side of the book, mirroring the yellow character featured among the five figures on the cover." src="https://cdn-images-1.medium.com/max/1024/1*8k032peqWwYsOk9HGwSkpA.png" /></figure><p>Every organisation I have worked with has, at some point, talked about putting the customer “at the heart of everything.” The intent is usually genuine. The problem is that good intentions do not always survive contact with systems, incentives, priorities, and the reality of how work gets done. That gap between what organisations want customers to experience and what customers actually experience is where most of the interesting work sits. I got tired of watching that gap get discussed in another slide deck, so I wrote a book about it instead.</p><p>It is out now. It is called <em>Have you tried turning the customer off and on again?</em>, and it is not a book about caring more. Most people working in customer experience already care a great deal. It is about what happens when empathy, good intent and smart ideas meet the machinery of an organisation. I spent years leading customer experience inside organisations, and that experience taught me how difficult it is to turn good thinking into something a customer can actually feel. That tension is a big part of why I wrote this book.</p><h3>The book I wish someone had handed me a decade ago</h3><p>Here is the sentence the whole book hangs off: CX isn’t broken. The buzzword version of it is. Good customer experience was never an act of generosity. It is always an act of self-preservation. When an organisation forces its customers to absorb friction it did not choose, it is not just losing goodwill out the front door. It is hollowing itself out from the inside. Frontline staff spend their days apologising for systems they did not design and cannot fix. Cynicism replaces judgement. Everyone learns exactly how much honesty the building can tolerate, and edits themselves down to fit.</p><p>I did not write this to be encouraging. I wrote it because altruism is optional and self-preservation is not, and CX teams that keep pitching the first one are going to keep losing the argument to finance.</p><h3>When Internal Agreement Feels Like Evidence</h3><p>One of the ideas I keep coming back to is how easy it is for customer work to become internally convincing. The deck is polished. The language is right. The room agrees. The initiative feels important. And sometimes all of that happens without any meaningful change to the customer experience itself.</p><p>That is the uncomfortable bit. Internal enthusiasm can feel a lot like evidence when you are close to the work. But customers do not experience the workshop, the framework or the presentation. They experience the process, the wait, the handoff, the bill, the website, the call, the mistake and what happens next.</p><p>A lot of the book is really about that gap between what feels good inside an organisation and what actually works outside it. Good CX is not about creating more convincing internal stories. It is about making sure the customer can feel the difference without ever having to see the slide deck.</p><h3>What is actually in it</h3><p>The book has twenty-four chapters across five parts, and I tried to make every chapter title do a bit of the arguing on its own. There is one on why journey maps function as the décor of good intentions rather than a plan for anything. There is one that asks why AI keeps getting treated as an answer to a question nobody in the room actually asked. There is one built entirely around the observation that delight will not save you if the Wi-Fi does not work, because a customer whose task has failed does not want a cupcake, they want the thing to work. There is one about the friendliest war in the building, which is what happens when every department insists it already owns the customer relationship and none of them will admit which one of them is lying.</p><p>None of these are contrarian for the sake of it. They are what happens when you take the frameworks CX has been handed, the workshops, the journey maps, the NPS dashboards, the personas, and actually check them against what happens to a real customer afterwards. Most of them do not survive the check.</p><h3>Why I put my name on the argument</h3><p>Have You Tried Turning the Customer Off and On Again? is a love letter, not an obituary. A slightly sarcastic one, written with a smirk rather than inked in devotion, but a love letter all the same. I tease CX because the discipline deserves better than the version it has been diluted into, and because the only way to take it seriously is to stop pretending the sticky notes and soundbites were ever the point.</p><p>This is not a book for people who want to feel validated about the customer strategy they already have. It is for the CX lead who has started drafting the honest version of the deck in their head on the drive home, and then deleted it before the next morning’s meeting. It is for the operations manager who suspects the journey map on the wall has not been consulted since the day it went up. If either of those sounds uncomfortably familiar, you are exactly who I wrote this for.</p><h3>Where to get it</h3><p><em>Have you tried turning the customer off and on again?</em> is out now in paperback. Grab a copy at <a href="https://experienceworks.co.nz/">experienceworks.co.nz</a>, <a href="https://alexmurphycx.com/">alexmurphycx.com</a>, or on Amazon. If you have ever sat in a meeting and clapped for an idea that the customer would never survive, this book was written for the version of you who nodded politely, then spent the drive home conducting a much better meeting with yourself.</p><p>Most CX advice tells you to fix the customer’s feelings. This book asks you to fix the system that is currently making your customer feel that way in the first place, and then to stop applauding yourself in the meantime. If your customer experience strategy is not working, you already know what everyone suggests first. Try reading the book before you try turning it off and on again.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ca4982666fa0" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/most-cx-is-theatre-i-wrote-a-book-about-what-happens-after-the-applause-ca4982666fa0">Most CX Is Theatre. I Wrote a Book About What Happens After the Applause</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The kinder your chatbot gets, the angrier your customers will be]]></title>
            <link>https://medium.com/@alex_88610/the-kinder-your-chatbot-gets-the-angrier-your-customers-will-be-a44d90d34109?source=rss-4fa7a7001f6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/a44d90d34109</guid>
            <category><![CDATA[customer-service]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[business-strategy]]></category>
            <category><![CDATA[customer-experience]]></category>
            <dc:creator><![CDATA[Alex Murphy]]></dc:creator>
            <pubDate>Mon, 21 Sep 2026 21:52:08 GMT</pubDate>
            <atom:updated>2026-09-21T21:52:08.816Z</atom:updated>
            <content:encoded><![CDATA[<p>Warmth with no power to act is not a softer version of good service. It is the exact structure of a con, and the empathy upgrade everyone is buying after this year’s backlash is about to prove it at scale.</p><figure><img alt="Bright pink retro desktop computer with a glowing screen displaying an “X” and the word “WRONG,” connected to a black power cord and plug." src="https://cdn-images-1.medium.com/max/722/1*Z4k3UCZXIWQ3CUUjgm_2bw.png" /></figure><p>A company that sells AI customer service software recently commissioned a survey into how people behave with AI customer service agents. This was brave of it. More than four in ten of us now shout the word “human” into the phone like a distress flare. Roughly one in six swears directly at the bot. More than half give up inside three minutes when the thing is plainly not resolving anything. When even a vendor’s own research reads like a hostage report, the backlash has stopped being a fringe grumble and become the weather.</p><p>The industry has looked at numbers like these and reached for a diagnosis. The bots need to be warmer. Kinder. More human. Sorrier. Every vendor deck now sells empathy the way it used to sell efficiency, and every organisation that got publicly embarrassed by its AI this year is buying the same upgrade. It is the most expensive mistake in the room, and within a quarter or two it will be everywhere.</p><p>Picture a woman trying to get her money back for a concert that was cancelled. Not a difficult case. The event did not happen, the money is hers, everyone agrees on the facts. She finds the chat bubble and explains. The agent, freshly upgraded, is magnificent about it. It uses her first name twice in the first sentence. It tells her it completely understands how disappointing this must be, that it can only imagine how much she had been looking forward to the night, that it is genuinely sorry she has been put through this. It holds space for her emotional journey. Then it offers her a fifteen percent discount code for a future event by the same promoter she now actively hates, and asks if there is anything else it can help her with today. It has not refunded a cent. It cannot. Nobody gave it the authority to move money. They gave it the authority to feel things.</p><p>She did not want to be understood. She wanted her money. And the warmer the machine was about not giving it to her, the more she wanted to put her fist through the screen.</p><h3>Warmth is a tell, not a service</h3><p>Here is what everybody rolling out the empathy upgrade has failed to notice. Warmth aimed at a customer you are not actually helping does not read as kindness. It reads as a tell.</p><p>Think about where else in life you meet lavish sympathy attached to total inaction. The salesperson who wholeheartedly agrees the price is unfair and cannot change it by a penny. The colleague who really hears you and then does the exact thing you asked them not to. Performed concern from something with no power to act is not a neutral pleasantry. It is the specific texture of being managed, and every adult customer has felt it enough times to recognise it on contact. The sympathy is not a gift. It is the sound the door makes as it closes gently in your face.</p><p>A confidence trick, stripped to its mechanism, is warmth deployed in place of substance. The con man’s entire product is how understood you feel while he takes the thing you came in with. That is not an accusation of intent, because the chatbot does not intend anything. It is worse than that. Your organisation has built a device that runs the emotional grammar of a con, sincerely, at scale, twenty-four hours a day, and is calling the result an improvement.</p><h3>The research has been sitting here the whole time</h3><p>None of this is a hunch, which is what makes it faintly embarrassing that the entire industry is walking into it. The behaviour was mapped years ago.</p><p>Psychologists Susan Fiske, Amy Cuddy and Peter Glick call it the Stereotype Content Model. The finding is that we size up anyone we deal with on two axes almost immediately: warmth, meaning do you mean well toward me, and competence, meaning can you actually do anything. The two are separate judgements, and the combination decides the feeling. High warmth with high competence earns trust and admiration. Low warmth with high competence earns respect, the register of the surgeon you do not especially like but absolutely want holding the scalpel. And high warmth with low competence, the quadrant your newly cuddly chatbot has just sprinted into, earns something far more dangerous than dislike. It earns pity, and pity curdles into contempt.</p><p>That is the quadrant of the well-meaning fool. It is where customers file the agent who is terribly nice and completely useless. You have not made your bot likeable. You have engineered it, deliberately and at considerable cost, into the one position on the map that produces warmth and contempt in the same breath.</p><p>There is a second piece of evidence, and it is even more inconvenient because it comes from the frontline rather than the lab. A study written up in Harvard Business Review analysed real service interactions and found that an apology which runs past the first few seconds actively reduces customer satisfaction. What lifted satisfaction was not warmth at all. It was watching someone visibly, energetically try to solve the problem. Customers rated the agents who moved fast to fix things higher than the agents who dwelt on how sorry they were, even when the sorry ones were plainly more pleasant. The lesson organisations took from a decade of that research was, apparently, nothing, because they have just spent a fortune teaching a machine to dwell.</p><figure><img alt="A black matchbox filled with black matches sits on a vivid pink surface, with one pink match among them as a pink-gloved hand reaches toward the box." src="https://cdn-images-1.medium.com/max/1024/1*c3ONW3gidJcg6tTrUnvuxQ.png" /></figure><h3>Why kindness curdles</h3><p>The reason all of this lands so badly is that empathy without power does not simply fail to help. It removes the customer’s last usable option, which is hope.</p><p>When Craig the human agent could not fix your problem, at least the interaction contained a lever. You could ask for his manager. You could hear the hesitation in his voice, sense that he knew it was wrong, and push toward someone who could act. The friction was awful but it was navigable, because underneath it sat a chain of people, one of whom had the power to say yes. The whole miserable ritual of escalation is really just the customer feeling around in the dark for the person with authority.</p><p>The empathetic bot presents a smooth wall with a smile painted on it. It never hesitates, because it is incapable of doubt. It never breaks character, because it has no character to break. It absorbs the complaint, reflects the emotion back in beautifully punctuated sentences, and offers nothing to push against, because there is nothing behind it to push toward. Psychologists have a name for what that does to people, and it is not a gentle one. Learned helplessness is the state you fall into when your actions stop having any effect on the outcome, and it is one of the most reliable ways ever documented to make a person first furious and then despairing. The kinder the wall is while it refuses to move, the faster they get there.</p><p>This is the mechanism behind the headlines. People are not shouting at chatbots because the bots are cold. They are shouting because the bots are warm and immovable at the same time, which is the precise combination that tells a human being, at an almost cellular level, that they are being handled rather than helped.</p><h3>The upgrade nobody senior will sign off</h3><p>So why is everyone buying the wrong fix. Because it is the only fix that is cheap and comfortable at the same time.</p><p>Look at the two numbers the industry is wedged between. Gartner found this year that 91 percent of customer service leaders are under pressure to implement AI. It also found that 87 percent of customers now say any company using generative AI for service must give them a guaranteed way through to a human being. That is the entire trap in a single line. Leaders are being pushed, hard, to deploy the very thing the overwhelming majority of their customers are asking for an escape route from. Empathy is how an organisation appears to answer the first number while doing precisely nothing about the second.</p><p>The thing that would actually work is to give the machine real authority, or route the customer instantly to a person who has it. Both cost money, and both mean admitting that a large share of your contact volume needs a human with a mandate rather than a script with a personality. That is a difficult conversation with a finance director and an uncomfortable one with a board. Warmth requires none of it. Warmth is a prompt change. You can be seen to respond to the backlash by close of business, spend almost nothing, cede no control at all, and produce a bot that sounds like it cares by Friday. It is the one feature you can fake for free, which is exactly why an entire industry is about to overdose on it.</p><p>And notice, as ever, who is left holding it. Not the executive who approved the empathy upgrade to make the complaint numbers look handled. Not the vendor who sold the warmth module and is already booking the follow-on demo. The customer experience team, who did not choose any of this, inherit the transcripts, the escalations, and the job of explaining to a genuinely enraged customer why the machine spent four paragraphs on her feelings and zero seconds on her refund. They did not write the cheque. They are just standing there when it bounces.</p><h3>So do the opposite of what the vendor is selling</h3><p>Stop measuring your bot on how it sounds and start measuring it on what it is allowed to finish. Tone is not the metric. Resolution authority is.</p><p>Take your top ten contact reasons, the real ones by volume, not the tidy ones on the strategy slide. For each, ask one flat question: can the bot actually complete this, on its own, to the end. Refund the charge. Rebook the flight. Override the policy. Cancel the account without launching a retention interrogation. Where the honest answer is yes, give it the authority and then stop dressing it up, because a machine that can act does not need to perform sympathy. Competence reads as respect all by itself. Where the honest answer is no, do the thing the vendor will talk you out of. Delete the empathy lines entirely and hand the customer to a person who can act, fast, with no scenic detour through the machine’s feelings first.</p><p>Those are your only two honest settings. A bot that can help should get on and help. A bot that cannot help should say so in one plain sentence and get out of the way. Everything in between, every “I completely understand how frustrating this must be” bolted onto a system that can do nothing about the frustration, is not service and it is not kindness. It is the confidence trick your own organisation is running on the people who trusted it, and no amount of warmth will do anything except make the moment they realise it arrive a little sooner, and land a little harder.</p><p>Your customers never wanted your chatbot to care. They wanted it to be able to say yes. Give it the power to, or give them a person who has it. Everything else is just a nicer voice reading out the same refusal.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a44d90d34109" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Your customers are grading you against a company you’ve never heard of]]></title>
            <link>https://medium.com/design-bootcamp/your-customers-are-grading-you-against-a-company-youve-never-heard-of-f4a9881a764d?source=rss-4fa7a7001f6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/f4a9881a764d</guid>
            <category><![CDATA[business-strategy]]></category>
            <category><![CDATA[pink]]></category>
            <category><![CDATA[branding]]></category>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[competitor-analysis]]></category>
            <dc:creator><![CDATA[Alex Murphy]]></dc:creator>
            <pubDate>Fri, 04 Sep 2026 01:33:55 GMT</pubDate>
            <atom:updated>2026-11-16T21:12:20.733Z</atom:updated>
            <content:encoded><![CDATA[<h4>Your customers grade your service against the last easy thing they did on their phone, in any app, in any industry. The bar goes up for free, and nobody in your building is watching it.</h4><figure><img alt="A pink-gloved hand holds a pin toward a pink balloon labelled “EXPECTATIONS,” moments before it bursts." src="https://cdn-images-1.medium.com/max/1024/1*AYvQ7bvNekAqOxJFLmal6A.png" /></figure><p>Customer experience is not something you deliver. It is a subtraction your customer does in their head, without telling you, every single time: what just happened, minus what they expected. You have spent your entire budget on the first number. The second one is being set by strangers in other industries, and by your own marketing department, and you do not measure it, manage it, or admit it exists.</p><p>Let me show you where it goes wrong.</p><p>Picture a man on his sofa on a Tuesday night. He has just split a dinner bill with three friends by tapping his phone twice. Four seconds, no login, done. Still faintly glowing from how easy that was, he opens his car insurance renewal. It asks him to create a new password (twelve characters, one of them a symbol nobody has ever typed on purpose), then to verify his identity with a code sent to the email address he opened the account with in 2014 and has not looked at since, and then, for reasons known only to God and the compliance team, to re-enter his car’s registration. Which they already have. Because they are the ones renewing it. He gives up. He will do it later. He does not do it later.</p><p>Here is the thing. Nothing about that insurance form got worse. It is the same form it was five years ago, back when it was “fine”. What changed is the four seconds on the sofa. The dinner-bill app quietly reset what “normal” feels like, and it did it to your insurance company for free, without permission, from a completely different industry. The man is not comparing you to your competitors. He is comparing you to the last time anything, anywhere, was easy.</p><h3>You are measuring the wrong half of the equation</h3><p>Every experience a customer has is that same subtraction: perception minus expectation. Delight is just a positive gap. Fury is a negative one. The identical interaction can produce either, depending entirely on the number the customer walked in carrying.</p><p>Now look at what you actually measure. Resolution times. Satisfaction scores collected the second the call ends. Handle time, first-contact resolution, the whole glittering telemetry of the delivered thing. Every bit of it sits on the perception side. Here is the most expensive number in your entire business that appears on precisely zero dashboards: what the customer expected. You have built an elaborate machine to measure one half of a subtraction, and left the other half to be set by people who do not work for you and could not care less about your quarter.</p><p>Two of those people are worth meeting.</p><h3>The first one you will never control</h3><p>The first is everyone. Every app that gets smoother, every retailer that starts shipping same-day, every checkout that quietly removes a step, every service that remembers you so you never have to introduce yourself twice. None of them are in your category. All of them are raising the bar you will be judged against tomorrow morning.</p><p>This is why customer experience teams have that specific haunted look. They are running as hard as they physically can and the score keeps drifting down, and it feels like a personal failing, and it is not. You can hold your service perfectly, flawlessly still, change absolutely nothing, and still get worse every quarter, because the reference point refuses to stand still with you. Standing still is a slow slide backwards that nobody on your team caused and nobody on your team can stop. The bar goes up for free. You just pay for it later.</p><figure><img alt="A pink-gloved hand places the tallest pink bar beside a series of rising black bars, representing rapidly increasing customer expectations." src="https://cdn-images-1.medium.com/max/1024/1*051DO-f_U93NKSk-RmjaJw.png" /></figure><h3>The second one works down the hall</h3><p>The second person inflating the customer’s expectations is much easier to find, because they sit about forty metres from the contact centre and they have a considerably nicer budget. It is marketing.</p><p>Every brand campaign is a promise. “We’re with you.” “Banking that finally gets you.” “Here to help, whenever you need us.” Beautiful lines. Award-winning lines. And every single one is a cheque that operations has to cash at two o’clock on a wet Thursday when the system is down and the queue is forty deep. The warmer and cleverer the campaign, the higher the expectation it plants, the wider the gap when reality shuffles in wearing odd socks. Which produces one of the quietly absurd truths of modern business: a company can run a genuinely brilliant brand campaign and actively make its customer experience worse in the very same quarter, with the very same money, and never once put those two facts in the same room.</p><p>And notice who is left standing underneath all of it. Not the people who made the ad. Not the leadership who signed off the promise. The customer experience team, handed a gap it did not create, holding a broken chatbot and a script that opens with “I understand your frustration”. They did not write the cheque. They just get to be there when it bounces.</p><h3>The arithmetic nobody says out loud</h3><p>So put it together. You pour money into the delivered side, because that is the side you can see, measure, and put on a slide. Every real improvement gets a small celebration. Meanwhile the expectation line, fed from two directions at once, climbs faster than you are improving. And the customer feels the gap, never your effort.</p><p>That is how a company gets objectively better at its job and scores worse for it at the same time. It is not a paradox. It is just subtraction, doing exactly what subtraction does, while everyone in the building stares at the wrong number.</p><h3>Expectation is a lever, not just a liability</h3><p>Here is the part that should cheer you up, because so far this reads like a piece about forces you cannot touch. You can touch expectation. You have simply never treated it as a dial you are allowed to turn.</p><p>The pizza tracker did not make a single pizza faster. It made the wait feel governed, because you always knew where you stood. The delivery estimate that says Thursday and arrives Tuesday buys more goodwill than the one that says Tuesday and arrives Tuesday, for the identical van doing the identical drop. Telling someone the queue is fifteen minutes and clearing them in twelve beats a surprise eight-minute wait, even though eight is objectively less than twelve. None of that is speed. All of it is expectation, set deliberately, in your favour, on purpose.</p><p>Companies almost never do this, because it feels like confessing to weakness. Under-promising sounds like a shortage of ambition to a leadership team raised on stretch targets. So instead they over-promise, miss, and apologise, on a loop, paying in trust for a number they set themselves and could have set lower. The most underused tool in customer experience is a realistic, clearly stated expectation, and it costs nothing except the ego required to say a slightly less impressive thing on purpose.</p><h3>So do the two things that aren’t on the roadmap</h3><p>Neither of these is a transformation programme. Both are cheap. Both are sitting right in front of you.</p><p>First, find your customer’s real reference point. Not your competitors, that is the comfortable answer and it is wrong. The last easy thing they did before they arrived at you. Go and sit where they sit. Split a bill, order a coffee ahead, cancel a subscription in one tap somewhere, then walk straight into your own renewal flow. That queasy little lurch you feel is your actual score. Design to that bar, or decide honestly that you cannot, but stop grading yourself against the other lot in your category who are drowning in precisely the same tide.</p><p>Second, read your own marketing as a promise ledger. Line by line, every claim, ask one flat question: can operations keep this at two o’clock on a wet Thursday? The lines that pass are brand. The lines that fail are not brand. They are debt, and your customers are the ones quietly paying the interest. For most companies, the cheapest available improvement to customer experience is not an operations overhaul. It is an edit to an ad.</p><p>You cannot control the man on the sofa. He will keep being handed easier and easier things by companies you have never heard of, and the bar will keep rising whether you watch it or not. But you can absolutely stop being the one who raises it against yourself. Start with the cheque your own brand is writing in his name.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f4a9881a764d" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/your-customers-are-grading-you-against-a-company-youve-never-heard-of-f4a9881a764d">Your customers are grading you against a company you’ve never heard of</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Dolly Parton gave a fifty-year CX master class she never knew she was teaching]]></title>
            <link>https://medium.com/design-bootcamp/dolly-parton-gave-a-fifty-year-cx-master-class-she-never-knew-she-was-teaching-e1920a0e5039?source=rss-4fa7a7001f6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/e1920a0e5039</guid>
            <category><![CDATA[dolly-parton]]></category>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[behavioral-science]]></category>
            <category><![CDATA[research]]></category>
            <category><![CDATA[masterclass]]></category>
            <dc:creator><![CDATA[Alex Murphy]]></dc:creator>
            <pubDate>Sun, 30 Aug 2026 23:21:33 GMT</pubDate>
            <atom:updated>2026-09-04T02:05:59.543Z</atom:updated>
            <content:encoded><![CDATA[<p>Grief is the audit no brand can prepare for. Nobody rehearses it, nobody games it, and it files its report too late for anyone to argue. This week it reported on Dolly Parton, and millions of people, independently, returned the same verdict. That is not opinion, which scatters. It is behaviour converging, and behaviour only converges like that when the thing it measures was real every single time.</p><figure><img alt="Glossy hot-pink cowboy hat with white Western-style stitching, silver studs and a star buckle, set against a black background." src="https://cdn-images-1.medium.com/max/1024/1*Nkw6Y2q6mNiTD5TN5H3eeg.png" /></figure><p>She died on the twenty-fifth of August, and the tributes did the thing tributes do: they reached for the numbers. The record sales, the Imagination Library and its nearly quarter of a billion books, Dollywood, the theme park, the vaccine funding, the fifty-year marriage. All of it true, all of it staggering, and all of it slightly beside the point. Because the thing people are actually mourning is harder to put on a slide. What they are mourning is that she meant the same thing on Wednesday that she meant in 1974. She was, for half a century, precisely who she said she was.</p><p>That is not sentiment. It is the single hardest thing to do in customer experience, and she did it for longer than most companies survive. There is a line worth keeping somewhere near your desk: brand makes promises, experience keeps score. Brand is the story an organisation tells about itself, the warm and confident version written by people paid to make it sound inevitable. Experience is what the customer actually gets, the version written by whoever answered the phone, processed the refund, or didn’t. Psychologists even have a tidy name for what the customer does with the distance between the two. Richard Oliver called it Expectation-Confirmation Theory, and it holds that satisfaction is not how good the experience was in absolute terms, but how it measured against what you were promised. The two are supposed to match. The entire discipline exists because they almost never do.</p><h3>Why the gap opens</h3><p>Beat the promise and you delight. Meet it and you are trusted. Fall short of it and no amount of quality elsewhere rescues you, because the customer is not grading the experience, they are grading the gap. That gap is the one number a brand can move without touching the product at all, which is exactly what makes it so dangerous.</p><p>This is why the polished brand campaign is so often the thing that sinks the organisation. It does not raise the experience. It raises the expectation, which is the one number a company can inflate for free, and then it leaves the experience to be delivered by a system nobody funded to keep up. The advertisement promises a relationship. The contact centre delivers a queue. The gap between the two is not a rounding error. It is the whole of the customer’s opinion of you, and they will keep score whether or not you give them a survey.</p><p>You have stood in this gap yourself, probably last month. The airline that sells the seat with a photograph of a flat bed and a cold glass of something, then boards you onto a delayed aircraft where the crew are as worn out as you are. The bank whose advertisement is all handshakes and mortgages approved over coffee, reached in practice through a phone menu built to make you give up before you cost it anything. The energy company that calls you a valued customer in the same letter that raises your bill. None of them lied in the campaign. They funded the promise and starved the delivery, and the customer, who neither knows nor cares which department owns which, experiences only the distance between the two.</p><h3>The promise she kept</h3><p>Dolly Parton is the rare case where the promise and the experience matched for a lifetime. The promise was warmth, generosity, and a complete absence of pretension from a woman covered head to toe in rhinestones. The experience, every single time, was warmth, generosity, and a complete absence of pretension. She gave away books to children who would never buy a ticket to anything she made. She funded the research and let other people take the photograph. The brand said “I am on your side” and the ledger, checked at every point over fifty years, agreed.</p><p>She never opened the gap, and the reason is almost annoying in its simplicity. She never promised more than she was willing to keep delivering. The rhinestones were the honest part. Everyone knew the deal, and the deal never changed. Most organisations cannot keep that promise from the campaign launch to the end of the same financial year.</p><p>The consistency was not passive, either. She spent decades resisting every incentive to be more than the deal. She largely kept her politics to herself, not out of caution but because she had decided her tent would hold everyone, and a brand that means the same thing to people who agree on nothing else is doing something almost no organisation attempts. She let fans and staff describe her rather than correcting the story into something grander. She turned down honours that would have quietly repositioned her upward. Each of those is a small refusal to inflate the promise, and fifty years of small refusals is what a kept promise is actually made of.</p><figure><img alt="A row of vintage black microphones with one sparkling hot-pink rhinestone microphone standing out in the centre, set against a bright pink background." src="https://cdn-images-1.medium.com/max/1024/1*X0D_8Ntj_eECWY14Z6kKtg.png" /></figure><h3>The library nobody upsold</h3><p>Consider the Imagination Library, because it is the whole argument in miniature. A child in the scheme receives a book a month, free, from birth until they start school. There is no account to create, no data to harvest, no tier to climb, no gentle nudge toward a paid version that unlocks the good books. The organisation gives something of real value and asks for precisely nothing back, which is the one move modern loyalty design is structurally incapable of making.</p><p>This is what customer-centred looks like when it is not a slide. Most programmes described as customer-centric are built around the organisation’s need to measure, capture and convert, and the customer is invited to participate in their own harvesting. The Imagination Library inverts it. The child’s outcome is the point and the reputation is the byproduct, not the target. It is worth noticing that this is also the version that built the deepest reservoir of goodwill, and that the causality runs in the direction organisations find least convenient: she was trusted because she was visibly not trying to be.</p><h3>Why almost nobody manages it</h3><p>If consistency is this powerful and this cheap, the obvious question is why it is so rare, and the answer is not that leaders are foolish. It is that the incentives inside most organisations point the other way. Behavioural scientists call it Action Bias: the human preference for visible action over patient maintenance, even when doing less would serve better. A new campaign can be presented to a board. A quarter spent quietly making sure the existing promise still holds cannot. One gets a launch, a budget, and a name attached to it. The other gets filed as business as usual and handed to whoever is already overloaded.</p><p>So the promise gets louder on a schedule, because raising it is an event, and the experience gets quietly renegotiated downward between events, because protecting it is not. Every organisation ends up optimised to widen the exact gap that decides what its customers think of it. Dolly’s advantage was that she was never running that machine. There was no growth team asking how to extract more from the relationship this quarter, because the relationship was the product, not the funnel feeding it.</p><h3>The people holding the gap</h3><p>Here is where it stops being a parable and starts being a Tuesday. When a brand writes a promise the operation cannot keep, the shortfall does not land on the marketing team that wrote it. It lands on the customer experience team, who did not approve the campaign, were not consulted on the tagline, and are now expected to be the difference between the promise and the thing the organisation actually built. They inherit a gap they did not open and are asked to apologise, one interaction at a time, for a relationship the business described but declined to resource.</p><p>They are, in effect, handed a fifty-year standard and a fifty-day budget, and then measured on why customers feel let down. The score the experience keeps is real. The people made to answer for it are usually the only ones in the building who had no hand in setting it.</p><p>The gap does not even announce itself as a strategy failure. It arrives disguised as a run of individual complaints, each one small enough to be waved off as a difficult customer or a bad day, none of them ever added up into the single sentence that would embarrass the people who wrote the promise: we described a relationship we chose not to build. The team closest to the customer can see that sentence forming months before anyone with the authority to act on it will agree to read it.</p><h3>What actually earns the affection</h3><p>The uncomfortable finding underneath all of this is that consistency is worth more than brilliance, and it is available to anyone, which is exactly why so few take it. A brand that reliably delivers a modest promise builds a reservoir of trust deep enough to survive the day something goes wrong. A brand that delivers a spectacular promise four times out of five has taught its customers to expect the fifth, and negativity bias means that fifth time is the one they will remember and repeat. Dolly’s genius was never being more than she could sustain, which meant she never had to spend a single day being less.</p><p>There is a second mechanism working quietly underneath the first. Robert Zajonc documented what he called the Mere-Exposure Effect: the more reliably and repeatedly we encounter something without being burned by it, the more we come to trust and prefer it. Consistency is not just morally tidy, it is cognitively persuasive. Fifty years of the same Dolly, never contradicting herself, never revealed to be someone else once the stage lights went down, is mere exposure compounding into something no launch campaign can shortcut. Familiarity is not the enemy of a brand. Betrayed familiarity is.</p><p>So the reframe is simple, and slightly brutal for anyone about to sign off a rebrand. Stop investing in the promise and start investing in the ledger. Before the next campaign raises what customers expect, find the three points in the journey where the current promise already breaks, the moments customers actually talk about afterwards, and fix those before you spend a cent making the promise louder. Measure the distance between what the advertisement implies and what the worst day of your service actually delivers, and close it from the bottom up rather than the top down. It photographs badly. There is no launch event for closing a gap, no agency pitch, no before-and-after reel. It is also the only version of brand-building that a customer can actually feel, because it is the only one that shows up in the experience rather than the advertisement.</p><p>Dolly Parton spent fifty years proving that the strongest brand in the world is just a promise nobody caught you breaking. The tributes this week are the audit. She passed it the only way anyone ever has, which is by keeping score herself, quietly, long before the rest of us thought to check.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e1920a0e5039" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/dolly-parton-gave-a-fifty-year-cx-master-class-she-never-knew-she-was-teaching-e1920a0e5039">Dolly Parton gave a fifty-year CX master class she never knew she was teaching</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Your CX was already broken. AI just autocompleted it faster.]]></title>
            <link>https://medium.com/design-bootcamp/your-cx-was-already-broken-ai-just-autocompleted-it-faster-294ba59e1c36?source=rss-4fa7a7001f6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/294ba59e1c36</guid>
            <category><![CDATA[generative-ai-tools]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[customer-service]]></category>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[technology]]></category>
            <dc:creator><![CDATA[Alex Murphy]]></dc:creator>
            <pubDate>Tue, 25 Aug 2026 21:05:19 GMT</pubDate>
            <atom:updated>2026-08-26T06:53:09.372Z</atom:updated>
            <content:encoded><![CDATA[<p>Nobody actually blames AI for their bad customer experience. Blame would mean admitting the experience is bad, and no organisation is going to concede that at the exact moment it has bought the expensive thing that was meant to fix it. AI is not the villain here. It is the latest bandaid, the newest distraction stuck over a wound nobody has any intention of treating. There was an app before it. There was a survey before that. The wound is fine, thanks. It just has a very modern plaster on it now.</p><figure><img alt="A hand wearing a bright pink glove holds a white and pink megaphone against a black background, symbolising communication, amplification and making a message heard." src="https://cdn-images-1.medium.com/max/1024/1*iEGoelTjnqm7oYdoUs8w0g.png" /></figure><p>A customer gets charged twice for the same order. She does what customers do now, which is find the little chat bubble in the corner and type “I’ve been charged twice” into it. The AI agent, who has been given a friendly name like Milo and a small cartoon face, replies in under a second. It is warm. It is confident. It is beautifully punctuated. It tells her that her order is on its way and provides a tracking number for a parcel that does not exist, because she wasn’t asking about a parcel, she was asking about her money, and Milo has never once in his short life understood the difference.</p><p>The company’s response, when it eventually notices, is not to ask why the answer was wrong. It is to make Milo a bit better at delivering it. Somebody schedules a session on prompt optimisation. Somebody else books a demo with a newer, cleverer vendor. Nobody books time with the billing team, because the billing team is not new and exciting, and the billing team is where the actual problem has been sitting the whole time.</p><p>Nobody pulls the ticket history. If they did, they’d find the exact same complaint, about the exact same double-charge, filed in 2019, back when Milo was a human named Craig reading from the exact same wrong script. Craig also told her the order was on its way. Craig also couldn’t see her billing system because it lived in a different department behind a login nobody in his team had. The only thing that has changed in six years is that the wrong answer now arrives faster and with better grammar.</p><p>Here is what the plaster is covering. AI didn’t break your customer experience, and it is not going to fix it either. It was already broken, quietly and expensively, for years. All AI did was walk in and perform the broken thing at a speed no human could ever have survived.</p><h3>Glorified autocomplete, and why that matters</h3><p>Here is the least glamorous, most accurate description of the AI in your contact centre that you will read this year: it is glorified autocomplete. That is not an insult and it is not me being a sceptic at a barbecue. It is roughly how the thing works. It predicts the next most plausible bit of text based on the enormous pile of patterns it was trained on and the specific mess you have pointed it at. It does not know your customer. It does not know her problem. It pattern-matches to it, the way your phone finishes “I’ll be there in five” before you’ve decided whether that’s true.</p><p>Sit with what that means for a second, because it’s the whole argument. Autocomplete cannot invent a good experience you never designed. It can only complete the one you already have. If your knowledge base is a decade of contradictory articles written by fourteen different people who have all since left, autocomplete will confidently serve those contradictions to customers. If your returns policy technically exists in three incompatible versions across two systems and one very load-bearing spreadsheet, autocomplete will pick one at random and defend it to the death. You didn’t buy a brain. You bought a mirror with a keyboard, and it is now reflecting your operation back at your customers at scale.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*c3ONW3gidJcg6tTrUnvuxQ.png" /></figure><h3>The mess, industrialised</h3><p>For years, your broken bits were protected by a kind of natural friction. Craig was slow. Craig had a queue. Craig went home at five and took his wrong answers with him. The dysfunction was real but it was rationed, drip-fed to a few hundred customers a day by a tired human who occasionally, mercifully, sensed that something wasn’t right and escalated it to someone who could actually see the billing system.</p><p>Autocomplete removes the rationing. Whatever was broken is now available instantly, to everyone, all at once, twenty-four hours a day, with no queue and no five o’clock. If your handoffs dropped customers into the void, they now drop into the void the moment they arrive, at volume. You have not automated a good experience. You have industrialised a bad one, and then put a cartoon face on it so it feels like progress.</p><p>This is why “the AI is making mistakes” is such a comforting way to frame it, and such a useless one. It suggests the system underneath was fine and the new layer is the flaw. The truth is less flattering. The new layer is doing precisely what it was built to do. It is faithfully completing the system underneath. The mistakes aren’t a bug in the AI. They’re a feature of the thing the AI is standing on.</p><h3>Fluent is not the same as correct</h3><p>Here is the genuinely alarming part, and it has nothing to do with the wrong answer itself. When Craig didn’t know the answer, you could tell. He’d go quiet. He’d say “um, let me just check that for you” and put you on hold for four minutes while hold music played. That pause was awful, but it was honest. It was a signal, broadcast directly to the customer, that the answer was uncertain and the system was struggling.</p><p>Autocomplete has deleted the pause. It never goes quiet, because it is physically incapable of not knowing what to say next. Producing a fluent, confident, well-structured next sentence is the one thing it is genuinely excellent at. So your customer now receives the wrong answer with zero of the warning signs that used to come attached to it. No hesitation. No “let me check.” Just a crisp, cheerful, immaculately worded lie, followed by “Is there anything else I can help you with today?”</p><p>Fluency is not accuracy. We are wired to assume that anything that sounds this composed must know what it’s talking about, and autocomplete exploits that assumption for free, in both directions, without meaning to. The most broken thing about your CX is now also the most articulate. That is a genuinely new problem, and it is the only part of this that AI actually did invent.</p><h3>You cannot autocomplete your way out of a bad system</h3><p>So here is the uncomfortable bit that never makes it onto the strategy slide. There is no version of a better model that fixes this. A smarter autocomplete pointed at a broken process is just a more persuasive broken process. You can spend another six figures on a cleverer one and all you will have bought is more convincing wrong answers, delivered with even better grammar, even faster.</p><p>So here is what to do instead of buying the next one. Before you approve another model, another vendor, another optimisation sprint, do the single exercise nobody ever books time for. Take one real complaint, the double charge, and follow it all the way to the bottom. Not “why did Milo get this wrong,” but “what would anyone, human or machine, have needed to get it right.” The answer is never a cleverer chatbot. It is access to the billing system. A returns policy that lives in one place instead of three. A handoff that actually goes somewhere. Whatever is sitting at the bottom of that complaint is your real project, and it is the thing to spend the money on.</p><p>Do that, and autocomplete stops being a liability and starts being useful, because now it is completing something worth completing. It is a magnificent multiplier. It always has been. The only question that has ever mattered is whether the thing you hand it to multiply is worth more of.</p><p>And give that work to the people who can already see the problem. The CX team didn’t choose this. Nobody in the contact centre stood up in a meeting and demanded a chatbot named Milo be bolted onto a billing system it can’t see. Somebody higher up bought the modern-looking answer to avoid paying for the expensive, unglamorous one, and the CX team is the one now holding the tickets and getting asked why the customers are so angry. They didn’t break it, and they can’t fix it on their own, because the fix was never in their gift. It was in a budget nobody would point at the boring thing.</p><p>So point it at the boring thing. That is the whole strategy. Everything else is just a more expensive plaster.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=294ba59e1c36" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/your-cx-was-already-broken-ai-just-autocompleted-it-faster-294ba59e1c36">Your CX was already broken. AI just autocompleted it faster.</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Loyalty programmes reward customers who were never leaving]]></title>
            <link>https://medium.com/design-bootcamp/loyalty-programmes-reward-customers-who-were-never-leaving-2c47558fc13b?source=rss-4fa7a7001f6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/2c47558fc13b</guid>
            <category><![CDATA[behavioral-economics]]></category>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[digital-marketing]]></category>
            <category><![CDATA[loyalty-program]]></category>
            <dc:creator><![CDATA[Alex Murphy]]></dc:creator>
            <pubDate>Tue, 18 Aug 2026 02:00:36 GMT</pubDate>
            <atom:updated>2026-08-18T07:02:43.171Z</atom:updated>
            <content:encoded><![CDATA[<p>On Tuesday the retention team presents the loyalty numbers, and the room does the thing rooms do when a chart is going up: it relaxes. Members spend more than non-members. Members churn less. Someone says the word “sticky” with real feeling. The slide has a line on it, the line is climbing, and by the end of the meeting everyone has quietly agreed the line belongs to them. The programme is working. It has been declared a success, which in most organisations means it can no longer be questioned.</p><figure><img alt="A silver spoon filled with pale pink liquid and small human-shaped figures in pink, white and black, isolated against a black background." src="https://cdn-images-1.medium.com/max/1024/1*o7F-wSQXkHTJB3AUHS8ZnQ.png" /></figure><p>Nobody in the room asks the one question that would bring the whole thing down, because nobody in the room wants to be the person who asked it. The question is not whether loyal customers spend more. They obviously do. The question is whether the programme had anything to do with it, and the honest answer is being paid for in margin, one gold-tier discount at a time.</p><p>Start with the customer at the centre of the whole apparatus: the one with the card, the app, the points balance she checks roughly never. She is held up in every review as proof the programme works, because she spends more than the customers who never enrolled. This is true. It is also meaningless. She spends more because she was always going to spend more, which is precisely why she bothered to sign up in the first place. The programme did not create her loyalty. It found her already loyal, handed her a discount for the trouble, and then took credit for a decision she had made long before anyone designed a badge.</p><h3>The maths is measuring the wrong thing</h3><p>The number that keeps these programmes alive is that member-versus-non-member comparison, and the conclusion drawn from it is always the same: members are more valuable, so the programme that made them members must be working. It is a tidy argument and it is built backwards.</p><p>The people who enrol in a loyalty programme are not a random sample of your customers. They are the customers already most committed to you, self-selecting into a scheme that rewards behaviour they had every intention of continuing. Comparing them to everyone else and attributing the difference to the programme is like observing that people who buy season tickets attend more games than people who don’t, and concluding that the season ticket caused the enthusiasm. The enthusiasm bought the ticket. Statisticians call this <strong>selection bias</strong>, and it is the load-bearing wall of the entire loyalty industry. Remove it and most programmes are revealed to be an elaborate mechanism for identifying your most devoted customers and then charging yourself a fee to keep them.</p><p>None of this shows up in the reporting, because the reporting was designed by the same people who need the programme to look successful. The metric that would actually settle the question, whether a given customer would have stayed without the points, is the one metric nobody collects, for reasons that become obvious the moment you imagine presenting the answer.</p><figure><img alt="Black paper aeroplanes arranged in a row on a bright pink background, with one crumpled pink paper ball breaking the pattern and a pink-gloved hand reaching towards it." src="https://cdn-images-1.medium.com/max/1024/1*8eXuc_9aoPMzHfihNA5uxw.png" /></figure><h3>Points are a promise, and customers have heard a lot of promises</h3><p>Here is the part the programme design never accounts for. A points balance is not value. It is the rumour of value, deferred to some future moment that requires the customer to remember it exists, accumulate enough of it to matter, navigate a redemption process built by someone who has never used it, and accept whatever the algorithm has decided she is worth this quarter.</p><p>Behavioural economists have a name for why this fails to move anyone. <strong>Temporal Discounting </strong>is the well-documented human tendency to prefer a small reward now over a larger reward later, to a degree that looks irrational on a spreadsheet and looks perfectly sensible from inside an actual life. A points balance is the larger-later reward, and it is competing for attention against every immediate, tangible thing a customer could feel right now. It loses. It always loses. The programme is pouring resource into a form of value that human cognition is specifically wired to undervalue, and then expressing surprise that engagement is soft.</p><p>This matters because loyalty is not decided in the redemption portal. It is decided in the moment something goes wrong, and a points balance is spectacularly useless in exactly that moment. The customer whose order vanished does not open the app to admire her progress toward silver. She is deciding, right there, whether this is an organisation that can be trusted when it counts. The reward is a promise. What she is looking for is proof the promise means anything, and the programme has spent its entire budget on the promise.</p><h3>The customer you are actually losing cannot see the confetti</h3><p>Every loyalty programme has a blind spot shaped exactly like the customer it most needs to keep. Programmes are built for the good days, for the satisfied customer accumulating status she was going to earn anyway. They are aimed with real precision at the person least likely to leave.</p><p>The person who is actually leaving looks nothing like that. They are not angry. They have not complained. They are simply, quietly, losing faith, the way people do when an organisation has been adequate one too many times in a row. A tier upgrade does not reach them. A birthday email arriving in the middle of her disillusionment does not reach them, except as evidence that the organisation has confused knowing their purchase date with knowing them. What would have reached them is being told the truth when something broke, and being told it before they had to ask. The programme cannot do this, because the programme is not designed to notice them at all. It is throwing confetti at the customers standing nearest the exit sign they have no intention of using, while the ones drifting toward the real exit slip out through a door the dashboard was never pointed at.</p><p>And when they goes, the invoice does not land on the loyalty team that built the scheme. It lands, as it always does, on the CX team, who are handed a churn number they did not cause and asked to explain a departure from a customer the programme actively looked away from. They inherit the seam nobody designed, apologise for a relationship they were never resourced to protect, and get to watch next quarter’s slide show the line going up regardless.</p><h3>What actually makes a customer stay</h3><p>The uncomfortable finding underneath all of this is that the things which genuinely bind a customer are the opposite of what a points scheme is built to deliver. In 1966 the social psychologist Elliot Aronson documented what became known as the <strong>Pratfall Effect</strong>: a competent person who makes a visible mistake and handles it well is liked more than one who never appears to err at all. The blunder, owned honestly, creates warmth that sustained polish never can. Organisations spend fortunes trying to appear flawless, when appearing human and then behaving decently under pressure is the thing that actually earns affection.</p><p>It compounds with a second mechanism. Robert Cialdini’s Reciprocity Principle holds that when someone does something genuinely valuable for us, particularly at some cost to themselves, we feel an almost involuntary pull to return it. Not a templated apology with a ten-percent voucher attached, which every customer correctly reads as an attempt to buy silence at the lowest price legal will approve. Something real, that visibly cost the organisation something it did not have to spend. That is what triggers the reciprocal instinct that no points balance has ever activated.</p><p>Handled together, these are what actually accumulate into loyalty: a reservoir of goodwill deep enough that the relationship can absorb a failure without fracturing. A programme sitting on top of a deep reservoir feels like genuine appreciation. A programme sitting on a shallow one is a points-based distraction from a relationship nobody ever invested in, and customers can tell the difference with far more accuracy than the organisation would find comfortable.</p><p>So the reframe is simple and slightly brutal. Stop rewarding presence and start earning it. Presence is not loyalty. The customer who has not left is not the same as the customer who would not leave, and mistaking the first for the second is the founding error of every scheme that measures its own success by counting the people who were staying anyway. The best loyalty programme an organisation can build is a recovery capability that works before anything breaks, owned by someone with the authority to make it real. It photographs badly. There is no badge for it. It also happens to be the only version that survives the morning a customer finds out what kind of organisation she has actually been loyal to.</p><p>The line on the slide will keep going up either way. The only question worth asking about it is the one nobody in the room wants to answer: how much of that would have happened if we had done nothing at all.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2c47558fc13b" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/loyalty-programmes-reward-customers-who-were-never-leaving-2c47558fc13b">Loyalty programmes reward customers who were never leaving</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Your company is not customer-obsessed. Your org chart already confessed.]]></title>
            <link>https://medium.com/design-bootcamp/your-company-is-not-customer-obsessed-your-org-chart-already-confessed-0ffc4c3ce806?source=rss-4fa7a7001f6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/0ffc4c3ce806</guid>
            <category><![CDATA[leadership]]></category>
            <category><![CDATA[organizational-culture]]></category>
            <category><![CDATA[customer-obsession]]></category>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[behavioral-economics]]></category>
            <dc:creator><![CDATA[Alex Murphy]]></dc:creator>
            <pubDate>Wed, 05 Aug 2026 21:05:52 GMT</pubDate>
            <atom:updated>2026-08-16T21:40:19.119Z</atom:updated>
            <content:encoded><![CDATA[<p>Every company says the customer comes first. Then it quietly builds a structure that proves the opposite, and asks the CX team to keep the secret. There is a wall in most offices, usually near reception, where the company has written down the person it would like to be. The words are always the same words. <em>Customer first.</em> <em>Customer-obsessed.</em> <em>The customer is at the heart of everything we do</em>, which is corporate for <em>the customer is at the heart of everything we </em><strong><em>say</em></strong>. The font is confident. Somebody paid a consultancy a genuinely upsetting amount of money to arrive at four words a nine-year-old could have suggested, and everyone who walks past the wall has long since stopped seeing it, which is the only mercy in the whole arrangement.</p><figure><img alt="Two pink rubber-gloved hands forming a heart shape against a black background." src="https://cdn-images-1.medium.com/max/1024/1*ulLYC0l7s3FwA964kxax2g.png" /></figure><p>Nobody on the executive team is lying when they stand in front of that wall at the all-hands and mean it. That is the part worth sitting with. They are not cynics. They believe, sincerely, that the customer comes first. It is just that belief, in an organisation, is not measured by what anyone says at the all-hands. It is measured somewhere far less flattering, and the company has already filled that measurement in, in permanent ink, in a document it does not think of as a values statement at all. It calls that document the org chart. And the org chart has been quietly telling the truth this whole time.</p><h3>The wall is what you’d like to believe. The budget is what you actually believe.</h3><p>In 1938 an economist named Paul Samuelson got tired of asking people what they preferred, because people are magnificent liars, mostly to themselves. So he proposed something colder and far more useful. Stop asking. Watch what they choose. He called it revealed preference, and the idea is disarmingly simple: your real preferences are not the ones you announce, they are the ones you can be observed paying for. A person who says they value their health and buys cigarettes every day has, in the only language that counts, revealed a preference for cigarettes. The stated preference is a press release. The revealed preference is the receipt.</p><p>Organisations have stated preferences too, and they are lovely. They are on the wall. But an organisation also makes choices, constantly, under real constraint, with real money, and every one of those choices is a receipt. Who reports to whom. What gets a budget and what gets a <em>maybe next year</em>. Which meeting the CEO actually attends versus which one they send a deputy to. Who gets promoted, and for what. Which team gets cut first when the number needs to move by Friday. None of these are values statements. All of them are values statements. They are simply the honest kind, the kind written in a language the company cannot fake because it costs something to write.</p><p>And when you read the org chart as a receipt rather than a diagram, a very specific confession tends to fall out of it.</p><h3>Show me the reporting line and I’ll show you the priority</h3><p>Find the person ultimately responsible for the customer’s actual experience. Not the marketing of it. The lived experience of it: the support, the service, the moment something breaks and a real human needs it fixed. Now count the layers between that person and the CEO.</p><p>In a genuinely customer-obsessed company, that number is small, because the thing you are obsessed with does not sit four rungs down reporting to a VP of Operations who reports to a COO whose quarterly objectives are about cost. In most companies, that number is not small. The customer’s designated champion is a director, sometimes a senior manager, wedged into the part of the org chart reserved for functions the business thinks of as plumbing: necessary, unglamorous, and ideally silent. Revenue sits at the top table. Product sits at the top table. The customer sends a representative, and the representative needs an appointment.</p><p>This is the org chart confessing. Proximity to the CEO is the clearest revealed preference an organisation has, because attention is the one resource leadership cannot manufacture more of. Whatever reports directly into the top is what the top actually thinks about. Everything else is what the top intends to think about, later, once the urgent thing is handled, and the urgent thing is never the customer, because the customer, unlike the board, does not get an agenda item until they are already leaving.</p><h3>You are obsessed with whatever your bonus is obsessed with</h3><p>Ask a leader what the company values and you will get the wall. Ask what that same leader is personally compensated on and you will get the truth, and the two are almost never the same document.</p><p>This is the quiet mechanism that turns sincere people into a structure that does the opposite of what it says. A senior leader can believe in the customer with their whole heart on Monday and spend the entire week optimising for the three numbers their bonus depends on, none of which is <em>did the customer’s life get better</em>, because that number is hard to define, slow to move, and impossible to attribute to any single executive in time for the review cycle. So it does not get measured. And a value that is not measured is not a value, it is a decoration. The organisation is not obsessed with the customer. It is obsessed with whatever it has decided to pay its leaders to be obsessed with, and it has, revealingly, decided to pay them for almost anything else.</p><p>There is a tell here that behavioural scientists call the say-do gap, the reliable distance between the attitude a person reports and the behaviour they actually produce, and it does not close on its own. It closes when doing the thing starts to cost less than not doing it. Which is precisely what incentives are for, and precisely what almost no company aligns to the customer, because aligning them would mean paying leaders less when customers suffer, and that is a sentence that has never survived a remuneration committee.</p><figure><img alt="A dark storm cloud releases a bright lightning bolt against a vivid pink background." src="https://cdn-images-1.medium.com/max/1024/1*Sms4Yfo857ArA4xcQYiWXg.png" /></figure><h3>A value is only real once it costs something</h3><p>Here is the test that cuts through all of it, and it is a cruel one. A value that has never cost the organisation anything is not a value. It is a preference the company holds only for as long as it is free.</p><p>You can watch the whole thing resolve in a single event: the budget cut. When the number has to move and something has to give, the company stops describing itself and starts revealing itself, in real time, in a spreadsheet. Watch what goes first. It is almost never the thing on the wall, because the thing on the wall was never load-bearing. It is the research programme. It is the support headcount. It is the one team whose entire job was to notice what the customer was experiencing and say so. The function the company named as its obsession turns out to be the function it defunds first, because in the moment of actual constraint the customer was never the priority, only the mascot, and you do not lose sleep over the mascot when payroll is due.</p><p>This is the most honest moment an organisation ever has, and it wastes it. The redundancy round is a revealed-preference machine running at full power, printing an unedited list of what the company actually values in descending order, and nobody in leadership reads it that way. They read it as regrettable but necessary. They are correct that it is necessary and entirely missing what it reveals, which is that the wall was aspirational the entire time, and the first budget cut is where the aspiration goes to be quietly euthanised.</p><h3>Now you can automate the part you were pretending to value</h3><p>And into this, with immaculate timing, walks the technology that lets you act on the confession without ever having to read it aloud.</p><p>The pitch for AI in the customer function is efficiency, and it delivers. Point it at support, at service, at the frontline, and you can remove a remarkable amount of human cost. The board reads the saving as progress. But notice which function the company reaches for the automation knife first, and notice what that choice reveals. Nobody’s opening move is to automate the executive team. The first thing on the block is the part of the business that sits closest to the customer, the part the wall claims is sacred, and it goes first for exactly the reason the wall is a lie: because in the company’s genuine, revealed hierarchy of value, the customer relationship was always a cost to be minimised, not a relationship to be protected.</p><p>So AI does not make a company customer-obsessed. It does something quieter and more damning. It lets a company that was never customer-obsessed finally act on that fact at scale, cleanly, with a slide that says <em>transformation</em> instead of <em>we have decided the customer is not worth a human being’s time</em>. The stated preference stays on the wall, untouched, glowing. The revealed preference gets a budget, a rollout plan, and an executive sponsor. And the gap between the two, which the customer used to experience as indifference, they now experience as a chatbot that apologises beautifully and changes nothing, which is indifference with better latency.</p><h3>Read the receipt, not the wall</h3><p>So here is the reframe, and it changes what you can see from Monday.</p><p>Stop listening to what a company says about the customer. It will always say the same four words, and it will always mean them, and the words have never once predicted how the company behaves, because words are the cheapest thing an organisation produces. Read the receipts instead. The reporting lines. The comp plans. The agenda of the meeting that actually decides things. The order in which teams get cut. These are not where the company keeps its opinion of the customer as an afterthought. They <em>are</em> its opinion of the customer, stated in the only language it cannot fake, and they are all sitting in plain sight waiting to be read as what they are.</p><p>And if you want the single question that does the work of the whole essay, do not ask a leader whether the company is customer-obsessed. Everyone answers yes, instantly, warmly, and it tells you nothing. Ask instead: <em>when the budget was last cut, whose team went first?</em> Then watch the pause. The wall will always tell you who the company wants to be. The pause, and the org chart behind it, will tell you who it already is.</p><p>Because a company that has to write “customer-obsessed” on the wall is telling you something it does not realise it is telling you. You do not put up a poster of the thing you are actually doing. You put up a poster of the thing you are worried you are not.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0ffc4c3ce806" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/your-company-is-not-customer-obsessed-your-org-chart-already-confessed-0ffc4c3ce806">Your company is not customer-obsessed. Your org chart already confessed.</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Your contact centre is not fixing your problems. It’s hiding them.]]></title>
            <link>https://medium.com/design-bootcamp/your-contact-centre-is-not-fixing-your-problems-its-hiding-them-68d5343de66c?source=rss-4fa7a7001f6b------2</link>
            <guid isPermaLink="false">https://medium.com/p/68d5343de66c</guid>
            <category><![CDATA[behavioral-science]]></category>
            <category><![CDATA[contact-centre]]></category>
            <category><![CDATA[customer-experience]]></category>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[customer-service]]></category>
            <dc:creator><![CDATA[Alex Murphy]]></dc:creator>
            <pubDate>Mon, 27 Jul 2026 23:10:01 GMT</pubDate>
            <atom:updated>2026-07-29T04:49:53.335Z</atom:updated>
            <content:encoded><![CDATA[<p>Somewhere in your organisation there is a room, or maybe a floor, where dozens through to several hundred people spend their entire working lives absorbing the consequences of decisions they were never invited to make. They take the call about the billing error that finance has known about for two years. They soothe the customer whose order was lost in the gap between two systems that were never introduced to each other. They apologise, thousands of times a day, with a consistency and warmth that many couldn’t sustain past lunch. Leadership refers to this operation, when it refers to it at all, as a cost centre, which is a bit like calling your spine an overhead. And every one of those apologies, delivered so smoothly that the customer leaves calm, is quietly erasing the evidence that anything was ever wrong.</p><figure><img alt="Bright pink cartoon triceratops with pale pink horns and spots, standing against a black background." src="https://cdn-images-1.medium.com/max/1024/1*ifepwYcJTcls98a4uJDFhA.png" /></figure><p>That is the part nobody says out loud. A contact centre does not just handle problems. It disappears them.</p><p>An organisation likes to imagine its contact centre as a kind of customer service department, a place where queries go to be answered. What it actually operates is closer to an immune system. When something goes wrong anywhere in the business, and something is always going wrong somewhere in the business, the failure travels down through the org chart, through product and operations and billing and logistics, and it lands, finally, on a person with a headset who has ninety seconds to make it feel survivable. They neutralise it. The customer’s temperature drops, the ticket closes, and the infection is fought off so efficiently that the body it belongs to never develops a fever. No fever, no diagnosis. No diagnosis, no cure. The company stays sick and feels fine, which is the single most dangerous state an organisation can be in.</p><h3>Normalisation of deviance now comes with a service-level agreement</h3><p>The sociologist Diane Vaughan spent years studying how the Challenger disaster happened and arrived at a phrase that should be printed above every contact centre door. She called it the normalisation of deviance: the slow process by which a warning sign, seen often enough and survived often enough, stops looking like a warning at all and becomes just the way things are.</p><p>A contact centre is an industrial-scale machine for normalising deviance. The first time a particular failure generates a hundred calls, it is a crisis. By the thousandth time, it is a category. It has a code in the ticketing system, a canned response, a paragraph in the training manual, and a quietly agreed handling time. The failure has not been fixed. It has been onboarded. It has been given a desk. Everyone who deals with it has developed such fluency in managing it that managing it now feels like the job, rather than evidence that someone upstream is not doing theirs.</p><p>And this is where the cleverness of the whole arrangement becomes almost admirable, in the way a well-designed trap is admirable. The better the centre gets at handling a recurring problem, the less that problem looks like a problem. Competence is the camouflage. The team that has learned to defuse the billing complaint in under two minutes has, without anyone intending it, removed the last remaining pressure to stop generating the billing complaint in the first place.</p><figure><img alt="Row of black paper planes on a bright pink background, with one crumpled pink paper shape breaking the pattern and a pink-gloved hand reaching toward it." src="https://cdn-images-1.medium.com/max/1024/1*8eXuc_9aoPMzHfihNA5uxw.png" /></figure><h3>A calm customer is not a solved problem</h3><p>Leadership, meanwhile, is reading the wrong instrument, and reading it with great confidence.</p><p>The dashboards that reach the top floor measure the surface. Average handling time, first-contact resolution, a customer satisfaction score gathered in the ninety seconds of gratitude that follow a good rescue. All of them are measuring how well the centre absorbs failure, and none of them are measuring how much failure there is to absorb. It is a smoke detector wired to reward the fire brigade for arriving quickly, while saying nothing whatsoever about why the building keeps catching fire.</p><p>There is a bias underneath this that behavioural scientists call survivorship bias, the error of drawing conclusions only from the cases that made it far enough to be counted. The customers who reached an agent and left calm are visible, measurable, and reassuring. The ones who gave up in the queue, or never called because they had simply already decided to leave, appear on no dashboard at all. So the organisation studies its survivors, finds them remarkably content, and concludes the experience is healthy. The people who could have told a different story are, by definition, no longer on the line.</p><p>Nobody, in the history of corporate reporting, has ever built a slide titled “problems our contact centre successfully hid from us this quarter.” It would be the single most valuable slide in the building. It is also the one slide the current metrics are structurally incapable of producing.</p><h3>Learned helplessness, now available with a headset</h3><p>Ask why the centre does not simply escalate all this upward, force the fever, make the problem visible, and you run into the second half of the tragedy.</p><p>In the late 1960s the psychologist Martin Seligman described what happens to any creature repeatedly exposed to a problem it has no power to change. It stops trying. Not because it is lazy, and not because it stopped noticing, but because it has learned, correctly, that its actions do not alter the outcome. He called it learned helplessness, and it is the natural climate of a contact centre that has flagged the same fault fifty times and watched it survive every one. The agents raised it. The team leads raised it. It went into the report. Nothing changed. So eventually, sensibly, the raising stops, and the energy that once went into escalation gets redirected into the only thing that reliably works: getting better at apologising for it.</p><p>This is the quiet death of an organisation’s most sensitive early-warning system. The people closest to the customer, the ones who can hear a problem forming weeks before it reaches a spreadsheet, learn that their signal goes nowhere. So they stop sending it. The richest source of truth in the entire company gently powers itself down, not out of apathy, but out of a completely rational reading of the evidence that nobody upstairs is listening.</p><h3>Now you can hide it faster, cheaper, and at greater scale</h3><p>And into this, with immaculate timing, walks artificial intelligence, promising to make the whole thing more efficient. It will. That is precisely the danger.</p><p>The pitch is deflection. Point a capable model at the top tier of incoming queries and it will resolve, or at least close, a remarkable share of them before a human is ever involved. The board sees the volume to the human team fall and reads it as progress. But deflection and resolution are not the same word, and the gap between them is exactly where the truth used to live. An AI that instantly and politely handles ten thousand instances of the same underlying fault has not fixed the fault. It has industrialised the concealment of it, at a scale and speed no room full of humans could ever have reached, and it has done so while generating a metric that looks, from the top floor, like a triumph.</p><p>There is a further irony, and it is a sharp one. These systems are frequently trained on the contact centre’s own history, on the transcripts of all those thousands of expert apologies. So the model learns, faithfully, to do the one thing the organisation was already doing too well: to make the customer feel handled without anything underneath actually changing. We are automating the camouflage. We are teaching the machine to soothe the fever with the same skill that stopped us diagnosing the disease, and then congratulating ourselves on the cost saving.</p><p>The contact centre was, for all its flaws, still a room full of humans who could in principle get frustrated enough to bang on a door. A well-tuned model never gets frustrated. It will apologise for the same broken process, in the same warm register, ten million times, forever, and never once feel the specific human urge to march upstairs and demand that somebody fix the actual thing. Remove the humans and you do not just cut cost. You remove the last part of the system capable of caring that the problem is still there.</p><h3>The most honest instrument in the building is the one you keep trying to silence</h3><p>So here is the reframe, and it changes what you do on Monday.</p><p>Your contact centre is not a cost to be minimised. It is the most sensitive, most detailed, most continuously updated diagnostic instrument your organisation owns, and you have spent years treating its readings as noise to be suppressed rather than data to be read. Every recurring apology is a fault report. Every canned response is a problem your business has chosen to rent rather than repair. Every efficiency you celebrate at the front line is, more often than anyone admits, a measure of how good you have become at not noticing what is broken.</p><p>The work, then, is not to make the centre quieter. It is to stop rewarding it for silence. Count the apologies, not the handling time. Follow the most fluent, most practised responses upstream, because fluency marks the oldest unfixed failures, and treat every one as a map to a decision someone with actual authority could change. Use the AI not to muffle the signal but to read it, to find the pattern in ten thousand soothed complaints and route it, loudly, to the person who caused it.</p><p>Because a company whose customers are calm is not the same as a company whose customers have nothing to complain about. One has solved its problems. The other has simply hired, and now automated, the most talented people it could find to make sure you never have to hear about them.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=68d5343de66c" width="1" height="1" alt=""><hr><p><a href="https://medium.com/design-bootcamp/your-contact-centre-is-not-fixing-your-problems-its-hiding-them-68d5343de66c">Your contact centre is not fixing your problems. It’s hiding them.</a> was originally published in <a href="https://medium.com/design-bootcamp">Bootcamp</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>