Done right — like Granola combining human input with AI for superior meeting notes — it makes us more valuable.

This felt like a useful mental model. A comforting one, even.

As an AI strategist, this helped me reimagine value creation with AI instead of just using it to make things faster and cheaper.

Then I listened to Sangeet Paul Choudary on CXO Talk podcast. (Link in the comments)

My certainty evaporated.

The False Dichotomy

Sangeet, who advises Fortune 500 CEOs offers a framework that completely reordered my thinking. His core argument: augmentation can be just as dangerous as automation for workers. Sometimes more so, because it’s slower and less visible.

The automation versus augmentation framing is a false dichotomy.

His key insight: “Augmentation means the human effort gets complemented. But complementarity does not guarantee that you will continue to retain your salary and your agency associated with that work.”

Just because AI helps you do your job doesn’t mean you’ll keep your pay, your power, or the parts of your job that made it meaningful.

The Flattening Effect

When AI augments workers, the gap between highly skilled and less skilled workers compresses dramatically. The premium for expertise collapses.

Sangeet’s example:

Before Google Maps: Experienced London cab drivers spent years learning “The Knowledge” — memorizing 25,000 streets and thousands of routes. They commanded premium wages.

After Google Maps: A driver with one week of experience, augmented by GPS, could perform nearly as well as a 20-year veteran.

The premium for experience vanished. Not because drivers were replaced, but because the differentiating skill became commoditized.

The same pattern shows up in knowledge work: less skilled workers using AI can match the output of highly skilled workers using AI. Everyone gets better, but the skilled premium collapses.

The Hollowing Out Process

AI doesn’t usually replace your job all at once. Instead, it gradually absorbs the parts that make you valuable, leaving you with an increasingly commoditized role.

Sangeet described paid social advertising workers. Five years ago, they were highly valued, working alongside sophisticated algorithms, training and improving them.

As they trained the algorithms, their knowledge gradually transferred into the code. The work that remained became standardized, routine, interchangeable.

They didn’t lose their jobs. They lost their agency. They shifted from decision-makers to algorithm operators.

Two Types of Work

This is the end game of uncritical augmentation:

Strategic roles: People who create and manage the algorithms — data scientists, strategists, system designers. High agency, high compensation, often paid in equity.

Commoditized roles: Workers managed and directed by algorithms. Low agency, compressed wages, algorithmic supervision.

The cruel irony? The very act of being augmented by AI — working alongside it, training it, improving it — is often what pushes you from strategic work to commoditized work.

Two Different Futures

Sangeet’s framework crystallized the problem for me. But I wanted to push further: what would it actually take to design systems differently? I went back to the cab driver example to explore the fork in the road.

What Actually Happened:

Google Maps arrives. The skill premium vanishes — a one-week driver can now perform like a veteran.

Uber launches. The algorithm assigns rides, sets prices, manages drivers, dictates positioning. Navigation knowledge is gone. Route choice is gone. Pricing power is gone. Customer relationships are gone.

Result: The driver becomes an interchangeable unit executing algorithmic decisions.

What Could Have Been Designed:

GPS handles navigation — it becomes infrastructure, not the job itself.

The driver role gets intentionally restructured around what algorithms can’t commoditize:

• Urban expertise (real-time judgment on construction, events, actual fastest routes)
• Concierge knowledge (“best anniversary dinner spot,” “where tourists should actually go”)
• Relationship building (regular clients who request you specifically)
• Experience delivery (conversation, local stories, welcoming people to the city)

Compensation rewards expertise and relationships, not just completed trips.

Result: GPS is a tool. The driver remains the differentiator.

The Hard Truth

This second path exists — it’s what high-end car services do. But it’s a much smaller market.

And this is where Sangeet’s example led me to an uncomfortable conclusion: mass-market economics demand commoditization. Scale pushes people toward commoditized work over time.

The premium market can keep workers in strategic roles. But it’s maybe 5–10% of the market.

The other 90%? Cost matters most. Scale requires standardization. You can’t have a million “expert drivers with personal relationships” — it doesn’t scale. You CAN have a million “follow the app” drivers.

Where I’m Stuck

I’m generally an optimist about technology. But Sangeet’s framework has made me deeply uncomfortable.

If market forces inevitably push toward commoditization at scale, what’s the socially responsible way to design AI systems?

I’ve been looking at attempts to resist this pattern: companies like Costco that invest in workers despite market pressure, platform co-ops where workers own the algorithm, European regulations that limit algorithmic management.

But none have proven they can work at mass-market scale while remaining competitive.

I don’t have answers. What I have is a new lens that’s changed how I see “AI augmentation” — and I can’t unsee it.

How do we design systems that keep humans in strategic roles rather than commoditized ones? Or is that even possible at scale?

This isn’t rhetorical. I’m actually asking.

If you’re wrestling with this, designing products, leading teams, or navigating your own career in an AI world — I’d genuinely like to hear how you’re thinking about it.

‘Actually, I don’t care about the finding so much; I just care that there is a finding.’

Long after the interview, the conversation stayed with me. I was thinking about how I would sum up mine.

To give you some background, I am a customer researcher and product strategist. I also build research programs to inculcate a culture of customer empathy.

Here’s my ethos that has evolved over the 5 years. I used to be keen about the method(s), tools and domain expertise. I realize that I am more focused on enabling and communicating impact, building partnerships and scaling evidence based product decisions.

Learnings over rigor:

I am partial to lean and creative research methods that prioritize what we learn and how fast we can learn to make progress. Procedures can easily become proxy to making decisions and moving forward.

What does this look like?

• When there was absolutely no time to conduct research, my team leveraged Nielsen Norman’s research on e-commerce product comparison.
• When we had to concept test at scale, my team employed the 5 second test that is easier to teach anyone in 5–10 min.
• When we had only 4 hours to research for a chat bot, I employed slack channels to understand the conversation styles between millennial and non-millennials segments.
• Recently I have also been employing projective techniques (withdrawal) and games (buy a feature) from the realm of market research. These methods create a better rapport with customers to make progress.

Having championed lean methods, I recognize that it is important to be clear about when it is time to take a pause and go back to the drawing board. Foundational ethnography can be product life-altering. Usually, these are in service of re-defining value propositions, exploring new business models or revenue streams and not continuous/incremental improvements.

Actionable over Academic:

Most of us depend on research to advance a product, service or society. It is never an academic exercise. Ensure the research is in service of some decision making.

What does this look like?

• Spell out the So What and Now What of research findings. You are not a reporting squirrel but a change agent.
• When reports are not making the necessary impact, attempt a quadrant chart (2X2) that communicates quickly and feels familiar for the leadership
• Before conducting research, ask if the findings will change behaviors; it could be the product’s direction or the feature’s capability. I am aware that it does not always elicit an honest response. But this is more about communicating that research should not be done in vain or check some box.
• Sometimes, people value research but are upfront that the insights may not influence the immediate sprint. This is also good, and it has helped me prioritize research efforts accordingly. Importantly, it helps me understand the active and passive champions in the organization and manage my expectations.

Inclusion vs. Isolation

Research in isolation is never good for the researcher or the team. Before research can influence the customer’s experience, it needs to persuade the minds of our peers and stakeholders. After all, they are the means to the desirable end. It just comes down to taking ownership to make research a team sport.

What does this look like?

• I share the research calendar ahead of time, for people to participate at their convenience. This is an easier way for them to appreciate active listening and not brush off research interviews as banter. Slowly but surely, people warm up to it. You might even get lucky influencing team members to include ‘research participation’ during story point estimation!
• Most of us make research accessible in a centralized location. We can go a step further and consult the team when naming these files. After all, if the names are not intuitive, no one can use it. And I can promise you that this investment draws better engagement.

Usually, it is a combination of the product-name, feature version, and type of research in a desired sequence. E.g Phoenix-Buy-homepage10.2-cardsort-v2

• Opening up analysis and synthesis sessions can be magical. When it works, the teams move beyond understanding and internalize the customer context. And that leads to enormous gumption when solving problems. What can be better than that?
• I am also a strong advocate of hallway evangelism. Pin the customer insights/journey maps in meeting rooms, cafeteria walls and other common spaces. This becomes a conversation piece and gains unpredictable momentum.
• I have learnt that integrating quantitative and competitive insights along with qualitative insights touch more stakeholders. If you are lucky, you end up shifting the one dimensional data-driven mindset into a nuanced data- informed organization.

Collaboration vs Dogmatism:

This is actually about being adaptable for the greater good.

What does this look like?

• Scheduling ‘ask me anything’ sessions to engage skeptics has been a big hit. This helps create a climate to listen to opposing views. Also personally, this is a great exercise to increase our empathy quotient.
• Organizing lunch ‘n learns with product marketing, customer success and support teams to listen to their customer stories creates bonding; over time it breaks the silo boundaries and eases communication.

’ We know our customers and can map the journey ourselves; skip research’

• Sound familiar? I used to debate this with fervent explanations about how customer’s voice should be from customers (duh!). Well, most of my meetings were futile. After a few battle scars, I got wiser. I learnt how to organize a show and tell.
• I take my stakeholders at their word and invite them into a room to create provisional personas and journey maps. Invariably, they have a simplistic notion and/or may not be able to speak for some of the contextual needs across various touch-points. I quantify the risks of incorrect assumptions and gaps and that usually softens their resistance. Sometimes they come around and a few other times, they just accept the risks. Notwithstanding the outcome, I consider it a progress when everyone is aligned on risks and the next steps.
• There’s another surprising benefit to this approach. If I succeed in actually talking to the customers, I can compare the stakeholder version and the customer’s and let the gap sink in.

No persuasion; just immersion. It works wonders.

This session has also surfaced less known business context that never came up before.

Imagine the surprise of cloud architects when they learnt that their leadership was expanding the portfolio to address emerging business challenges. This led to rethinking the architecture and the funding models. And, it goes without saying that the architects were my biggest research champions thereafter!

Finally, I realized that my convictions stem from the kind of leadership I admire and value.

So, that’s me. I’d love to hear yours.

What is your research ethos? was originally published in UX Collective on Medium, where people are continuing the conversation by highlighting and responding to this story.

“We know what our customers want; We want to delight them by building it not just spend time on talking to them”.

I was astonished. Not because I was hearing this for the first time. But because the earnestness in the eyes of the entrepreneur was real. It was not the typical hubris or ignorance I am used to.

Something was different about him. I wanted to find out more. I dropped my judgments, defenses and just listened.

He is right. What he is referring to, is called ‘genius design’. You can get away with external customer research if you are also going to be the primary user of the tool every day. Until, your customer segments comprise of people unlike you.

This was a watershed moment. I decided to continue this exercise for the next dozen clients whose world view I do not share.

What I came away with was a life lesson to empathize better. Not just walk a mile to gather information or sell better but to truly make a connection. For that I needed to become their lawyer defending their positions to get to a better state of clarity.

“Research costs a lot of money and takes too much time. The quote the agency gave was just not affordable”

Mm… he is not wrong. Niche customer base is hard to access and is expensive to recruit. (e.g. A tourist who is also an acrobat). And ‘lean’ is not a concept that most customer researchers are comfortable with. Many have a hard time conducting research that needs to be just enough to make progress.

I remember how nervous and incomplete I felt, when I integrated customer research in agile projects. Estimating the research duration was torture; but it paid off. Research earned its rightful place in the product development process and contract agreements.

“My past experience tells me that this is an opportunity that we need to go after. I can see the patterns in adjacent industries.”

He has a point. He was in the food industry. The disruption he was planning shared similar dynamics in the mobility and health care industry. Intuition has a place that customer research cannot always support or validate.

“This is an incremental change. An obvious problem and a straightforward solution. Let’s move forward and watch how they behave.”

Not entirely wrong. Simple, incremental changes do not always call for formative research.

“One time, the researcher disappeared doing the research and came back with a big binder. Who has the time to go thru all that?”

Totally relatable. Inclusive customer research is still not the norm. Teams that do not get a chance to observe or listen to the customer cannot empathize to create a better experience.

“The cost of building this is very low. We are going to re-purpose our existing capability; we need to test before Thanksgiving and ensure that it does not adversely affect our brand during holidays.”

That’s fair. If the investment and brand affinity risk is low plus a time-to-market contingency, it may be reasonable to trade off customer research. Not as a norm but as an exception.

Well, these conversations are almost five years old but they have changed how I do things.

1. I do not moralize customer research anymore as I find it condescending. I empathize with the client and talk about it in terms in earning customer trust and investment risks

2. I assume the best intentions and ask better questions to understand where they come from and what is valuable to them

3. Calibrate the research methods and right size them; always strive to make the research process more inclusive

4. Set clear expectations about timelines and format based on what works for the team

5. I also do not go out of my way to convince someone. I just find a way to show value in a way that makes sense to them; or respectfully bow out.

I am still customer-obsessed; but a better change agent!

Note: Customer research in this topic refers to the qualitative research. This simply means interviewing and /or observing customers in an environment to understand their motivations, needs, frustrations and aspirations to create innovative solutions. This work establishes the mental model of the targeted customers to build intuitive, immersive experiences. The reference here excludes market research or surveys which are usually filed under quantitative research.

This question is a lot of work. Don’t even try to answer before you ask more questions!

When it was my turn, it was clear that I needed to understand more before trying to come up with clever answers. Also, I believe in the value of abstractions, frameworks and models that can be reused across problems.

Hard questions and difficult conversations helped me to separate the known-knowns and known-unknowns. Of course, I pray for the third kind, unknown-unknowns.

Quite a few articles have been written about this positing extreme views. And I have found myself agreeing with both sides, in parts. But I had to do this exercise for myself just like you have to. This post only intends to save you time in identifying some questions and frameworks.

Step 1: Path to clarity — Reality check

1. What are the customers telling me? Is there an expectation or a latent need for more?

2. Why do I care about distinguishing delight from great customer experience indicated by 5 stars or loyalty indicated by promoters? What is not enough?

3. Am I responding to a stakeholder/leadership pressure? Are they asking for this but worrying about something else? What is bothering them?

4. What is the problem I am trying to solve by answering the question? [growing customers, increasing engagement/retention, removing frustrations to stop churn-usually the last one is a bad candidate for this and related to question. 5]

5. Have I already considered the lower rungs of the ladder; such as usefulness, ease, the value across touch points?

6. What is the betterment that we expect to achieve after I unravel this?

7. Is there a substantial difference between providing seamless experiences and delight? If yes, to whom? [customer/business/partner/user]

8. Are my current KPIs not enabling me to discern this?

9. What level(s) is most meaningful to apply this? [CX level / UX level/ journey/ feature/capability]

10. What measurable business value can I correlate this to?

Step 2: Path to clarity — Shared understanding

Discuss these questions with your stakeholders and arrive at a shared understanding to avoid rabbit-hole issues.

Step 3: An unfailing checklist to unlock what can be delightful and meaningful

1. Leverage the jobs-to-be-done framework and look for:

~Upstream value fulfillment [e.g. A retail app facilitating the upload of your purchase history when you install the app]

~Downstream fulfillment [e.g. Texting customers store locations when they search for store locations on their devices]

~Look for Social, emotional and functional angles of jobs-to-be-done [e.g. Virgin Atlantic facilitating on-demand service for snacks and drinks turning airplane wait-staff into hosts [Jim Kalbach’s presentation lays out the model beautifully]

2. Gamification framework

3. BJ Fogg’s Persuasion model (Motivation, Ability, Triggers)

4. Dana Chisnell’s framework: Pleasure, Flow, Meaning [Check-out Jared spool’s talk on this referring to this model]

5. Other inspirations include: Personalization, planned spontaneity, outcome- driven innovation, anticipatory CX etc.

Step 4: How do I measure Delight?

Typically, the use cases you come up with will make it evident about how to measure success. The following list is meant as an additional reference:

1. Measuring experience as a function of expectations [Source: Snapsurveys blog]

2. Track share of wallet over at least 6 month period, if you have only a few competitors. This is meaningful when the changes are strategic or you are making multiple changes and measuring the accumulated impact.

3. Growth of customers, reduction of churn, especially if you are in the product-market fit phase

3. Facilitate a referral code and measure acquisition via referral; just don’t depend on the aspirational NPS when you can actually measure this.

4. Social conversation and amplification rate / engagement with user generated content. Also email ‘open’ and ‘read’ rates can tell you the story

5. Measuring mini NPS/CES at a feature level and qualitative feedback gives you the much needed feedback.

6. Beyond NPS, think about integrating Gallup CE11’s questions such as,

• I am proud to be a [Brand] customer
• [Brand] is the perfect [company/product] for me
• I can’t imagine life without [Brand]

What is your experience? Love to know what’s worked for you.

User needs to follow 30 people (Twitter)

User needs to send 2000 messages (Slack)

Sounds Familiar? These assertions are called the ‘magic numbers’ in the product world. They are so seductive that my clients want me to unearth these metrics for their products.

The first thing that came to my mind was ‘post-hoc fallacy’, captured memorably in this funny scene from West Wing.

Enjoyed the video? Welcome back.

In my head, this chase for the numbers felt too simplistic. I was thinking, ‘Why won’t 6 friends in 8 days cut it? Why not 12 days?’

The core value of products are elusive and ultimately it is defined by how the customer perceives it. There could be so many variables and complex relationships behind the scenes that are actually indicative of product adoption and growth. This distillation felt more like a dilution to me.

Until…. it hit me!

May be, this framing was all about keeping it simple to remember; helping the teams rally towards a unified purpose. I couldn’t help but recall the countless times I have lost the teams and stakeholders when I stressed the need to track user behavior, the failed attempts to inspire them to care about the user’s progress and not just be swayed by virality and visits in vacuum.

Well, it was time to be proactively skeptical about this; not use my distrust as an excuse for inaction but apply this in an effort to find a more reliable alternative.

As a consultant, I have the opportunity to work with diverse product teams to do this experiment. After many stimulating conversations about engagement and customer focused metrics, I have arrived at a framework that my teams could relate to, that compels action and is easy to remember.

I tend to think of them as ‘tipping point metrics’, as it feels like a customer retention threshold. Plus the moniker, magic numbers, feels flippant and gives the idea that these metrics are not reliable or repeatable.

Here’s the “Tipping Point Metrics” framework. Due to NDA constraints, I have illustrated this 4-step framework with Mint as an example. Let’s get started.

Step 1:

You begin by answering a few fundamental questions about your product and customers.

Who is your product helping and what do you know about them?

Not having answers to these questions is a red flag that you need to pay attention to. One exception is the customer lifetime value (CLV); you can proceed and as you complete this exercise, you will be better informed to establish the CLV.

Step 2:

What are the key activities of your customer journey?

You can also think of this as ‘What are the essential chain of actions that your customer needs to complete for you to make money?’ (You can track the channel for additional specificity)

Step 3:

Identify the loyal customer cohort and measure the duration between key tasks

To identify your loyal customer cohort, ask yourself,

Who are my engaged customers that I already consider retained; that I don’t bother with retention/win-back campaigns?

Too early and unsure? Many startups can provide the necessary cue here. If a customer is engaged for a 3–6 month period, they usually consider those cohorts dependable, retained customers. It can vary for each business. You gotta start somewhere right?

Now measure the duration between key tasks of this loyal customer cohort.

Step 4:

You are ready to establish the ‘Tipping point Metrics’.

Patterns will emerge and you’ll be ready to establish the tipping point metrics based on the data you have.

Remember that this is just the beginning and avoid the post-hoc fallacy. These relationships could be a strong correlation that you may want to dig deeper or establish causation. You are informed enough; now be prudent and run experiments to reaffirm.

Sometimes, your data could be so reliable they could all be true; but experiments can still clarify the priority as some user behaviors are more influential than others.

You can stop here. Or go the extra mile and give step 5 and 6 a try.

Step 5 & 6:

Armed with these patterns, you can ask more questions and validate more hypotheses. Talk and observe your users to surface unmet needs and unspoken gaps.

One valuable takeaway here. Tipping point metrics evolve and you may surface more than one. As we make progress we may refine our business model, redefine engagement and raise the bar on delight. Embracing this reality keeps you honest and encouraged.

SO…the next time, you hear about these magic numbers, resist the urge to dismiss them. Think of them as a useful illusion and get busy making them reliable.

Still here? Awesome! Love to hear your thoughts.

Magic numbers: the useful illusion was originally published in UX Collective on Medium, where people are continuing the conversation by highlighting and responding to this story.

How do I become a successful design leader?

A question that good designers never fail to ask. In the past few months, I have been trying to answer this in a more meaningful and actionable way.

Well, there are the obvious few: strive to be a ‘T’ shaped expert, a generous coach, a passionate learner and so forth. Somehow, it did not seem enough. I agonized.

What is the one lesson that I bring to work every day?

What can I share about my screw-ups that could save them future face-offs?

Is there an innate strength that they can leverage to be more successful?

Well, one precious nugget of wisdom stood taller than the others.

Knowing that empathy is not a zero-sum game. Extending our non-judgmental understanding to teammates,stakeholders and executives does not leave the users deprived; if anything, it can create more champions for them.

Allow me to expand. As designers, we consider understanding our users and their needs as sacred. We become their advocates. During this process, something more happens…

In our zeal for the user, when stakeholders veto our recommendations, we accuse them for not caring enough about the users. When product managers de-prioritize a feature, we frown upon their decisions. When developers raise feasibility issues, we complain that they are unable to appreciate how design can be game-changing. Inadvertently, we turn conversations into crusades.

Never pausing to realize that we may be guilty of the same; blind to their needs and tone deaf to their interests.

Let’s reflect for a moment. What makes us great designers? We listen to users’ needs, align solutions to their metal model and speak their language to craft successful experiences.

How about showing the same human-centered approach towards business stakeholders and our project teams?

It looks like this. We take time to understand how different stakeholders define the project’s success and why. We show consideration to their concerns and constraints. We are attentive to the project/product metrics that may impact their growth or image within the organization.

Slowly but surely, we get a handle on their priorities and why they are not able to get behind our recommendations. Let me be absolutely clear. This does not always make them right; just understandable.

What changes now?

Well, we become fluent in stakeholder-speak. We learn to focus on profitability and opportunity cost when talking to business stakeholders. We correlate design decisions to shorter conversion cycles when the sales team is in the room. We emphasize on productivity savings when we talk to colleagues from operations.

We don’t stop there — We begin to include peer teams in ways we could not imagine before.

We collaborate early with developers to avoid rework. We initiate conversations with product mangers to understand competition and market forces to adapt our design proactively.

We internalize that empathy is not a zero-sum game

What else happens?

• When we show that we care about the business and team needs, we become a problem solver
• When we exhibit our ability to see the big picture, we earn credibility and elevate the conversation
• When we renounce the us vs. them mentality, we become one tribe

So, when anyone asks about becoming a successful design leader, I say,

Leadership is about building trust and driving influence. One sure way to that is to internalize that empathy is not a zero-sum game.

Empathy is not a zero-sum game was originally published in UX Collective on Medium, where people are continuing the conversation by highlighting and responding to this story.