<?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 Lindsey Liu on Medium]]></title>
        <description><![CDATA[Stories by Lindsey Liu on Medium]]></description>
        <link>https://medium.com/@lindseyliu?source=rss-5ebd88fcb637------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*gZch2SYvVoHwXlt_oL-Z-w.jpeg</url>
            <title>Stories by Lindsey Liu on Medium</title>
            <link>https://medium.com/@lindseyliu?source=rss-5ebd88fcb637------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 17:58:39 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@lindseyliu/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[Learnings from launching a kickstarter product]]></title>
            <link>https://medium.com/@lindseyliu/learnings-from-launching-a-kickstarter-product-e3bd3153adb9?source=rss-5ebd88fcb637------2</link>
            <guid isPermaLink="false">https://medium.com/p/e3bd3153adb9</guid>
            <category><![CDATA[startup]]></category>
            <category><![CDATA[marketing]]></category>
            <category><![CDATA[crowdfunding]]></category>
            <category><![CDATA[san-francisco]]></category>
            <category><![CDATA[business]]></category>
            <dc:creator><![CDATA[Lindsey Liu]]></dc:creator>
            <pubDate>Fri, 08 Aug 2025 19:38:27 GMT</pubDate>
            <atom:updated>2025-08-08T19:41:22.840Z</atom:updated>
            <content:encoded><![CDATA[<p>This spring, I built and launched <a href="https://www.kickstarter.com/projects/minhaozhang/automatic-street-cleaning-alerts-for-busy-forgetful-human/description">CurbCue</a> on Kickstarter — a simple GPS-powered tag that alerts you before street cleaning so you never get a parking ticket again. It’s a local product that only works in San Francisco.</p><p>After two startup cycles, it’s now higher priority than ever to first validate whether anyone actually wanted it. So I spent 90% of my effort validating demand: just a landing page on Kickstarter, and put it out there on a lot of marketing channels, online and offline.</p><p>These marketing efforts, along with monitoring behavior data on the landing page itself, gave me tons of insights into what works and doesn’t work for a local and scrappy product.</p><p>If you’re building something similarly local or scrappy, here’s what you’ll take away:</p><p>•	Why pricing isn’t just about margins — it’s about matching the psychology of the moment</p><p>•	How context — not content — shapes attention</p><p>•	And why every product is really a mirror of your customer’s self-perception</p><h3>Model-Population Fit</h3><p>For majority of the time, we offered two pricing models on Kickstarter. A one-time purchase for lifetime access, and a monthly subscription.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/340/1*ngWbMNzSFKLv_l-iGlfaIA@2x.jpeg" /></figure><p>Every single backer chose the lifetime deal. Not the monthly subscription with much lower upfront cost.</p><p>This kinda surprised me. I know it’s easy to tell that the lifetime deal will soon justify its cost, but I also assumed people would be cautious to put a larger chunk of $$ at a product with no track record and no social proof.</p><p>It shouldn’t have worked — but it did. And I think it’s because the pricing model matched the moment.</p><ol><li>Early adopters are less risk-averse about backing scrappy teams, but just as wary of forgetting to cancel a subscription. This is especially true on a platform like Kickstarter.</li><li>Early adopters want to feel like insiders, not customers. A lifetime deal isn’t just a price point — it’s a signal. Of exclusivity. Of belief. Of being there early. And belief converts best when the terms are simple, the upside clear.</li></ol><p>What this taught me: Just like product-market fit, there’s model-moment fit. Pricing model has to match the psychology of the moment, not just the economics of the product.</p><p>When you’re unknown, sell belief. When you’re proven, sell value. Start with lifetime to build trust. Scale with subscriptions to build a profitable business.</p><h3>Mental load in marketing</h3><p>We expected digital ads to drive most of our Kickstarter pledges. After all, that’s how it worked with every other tech product we’ve built.</p><p>But this time, 100% of our backers came from pole flyers. Not Instagram. Not TikTok. Not Facebook. Not Nextdoor. Despite the fact that these social posts had decent engagement metrics.</p><iframe src="https://cdn.embedly.com/widgets/media.html?type=text%2Fhtml&amp;key=a19fcc184b9711e1b4764040d3dc5c07&amp;schema=twitter&amp;url=https%3A//x.com/trlindseyliu/status/1925000645570134466%3Fs%3D61&amp;image=" width="500" height="281" frameborder="0" scrolling="no"><a href="https://medium.com/media/cc5098037f00f6281057a2c0b88b8143/href">https://medium.com/media/cc5098037f00f6281057a2c0b88b8143/href</a></iframe><p>After the fact, I think it came down to two things.</p><ol><li><strong>Humans seek balance.</strong><br> People scroll past ads on their phones because they’re overstimulated — everything is loud, fast, optimized to grab. In that state, even a good Kickstarter page feels slow and dry. But out on the street? When you’re walking, waiting for the light to change, half-bored and thinking about nothing in particular — a flyer becomes the most interesting piece of information in sight. It’s not breaking the moment. It’s meeting it.</li><li><strong>Marginal value matters.</strong><br> Ads don’t exist in a vacuum. They compete with everything around them. A decent flyer in an empty space does more than a great ad in a crowded feed. The flyer doesn’t just stand out more — it <em>means</em> more, because it meets the brain where it’s ready to receive.</li></ol><p>What this taught me: Just like there’s product-channel fit, there’s context-attention fit. Flyers worked because they met people at the right energy level and delivered high informational value exactly when the brain had room for it.</p><h3><strong>Identity Signaling</strong></h3><p>We added a tiny line to our flyer: <em>“Proudly built by 9yr locals.”</em> It was meant to be a low-key credit. Instead, it triggered an unexpected <a href="https://www.reddit.com/r/sanfrancisco/s/VaDA1sLTHu">Reddit thread</a> debating what counts as a “true local.”</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/778/1*tfMro2TZCAS_3Q3R4hLcHQ@2x.jpeg" /></figure><p>That was when I realized “9yr locals” can be read either as <em>“we’re the real ones” or “</em>or <em>“we’re not real enough.” </em>It all depends on who’s doing the reading — and what identity they feel the need to defend. Either way, this line on the flyer unintentionally invites others to question their own status or defend their legitimacy.</p><p>Then another event further validates this realization. On our Instagram and Reddit ads, people left comments like <em>“use your brain”</em> or <em>“thought eyes were meant for reading and looking?”, </em>positioning themselves as too sharp, too savvy, for a product like ours.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*N9xA-Cf9TeNmACqcsFj-XQ@2x.jpeg" /></figure><p>These two cases stung a little. But the learning was clearer than ever — CurbCue and its framing <strong>activate people’s need to define where they stand in the hierarchy.</strong></p><p><em>What does this say about me if I need this? Am I the kind of person who forgets? Or am I the kind of person who figures it out on my own?</em></p><p>What this taught me: People don’t buy solutions. They buy self-perception. So activate that and let it go your way, not gets in your way.</p><p>Marketing isnt’ just telling someone how your product works — it’s telling them who they get to be if they use it. This has become marketing 101 thanks to brand marketing giants like Apple. But I never knew this would be so true even for a tiny product like us too.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e3bd3153adb9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The vicious cycle of LLM development: will AI shape our language?]]></title>
            <link>https://medium.com/@lindseyliu/the-vicious-cycle-of-llm-development-will-ai-shape-our-language-9fb88879cdbf?source=rss-5ebd88fcb637------2</link>
            <guid isPermaLink="false">https://medium.com/p/9fb88879cdbf</guid>
            <category><![CDATA[humanity]]></category>
            <category><![CDATA[linguistics]]></category>
            <category><![CDATA[nlp]]></category>
            <category><![CDATA[llm]]></category>
            <category><![CDATA[ai]]></category>
            <dc:creator><![CDATA[Lindsey Liu]]></dc:creator>
            <pubDate>Sun, 09 Jul 2023 06:55:01 GMT</pubDate>
            <atom:updated>2023-12-25T21:49:55.976Z</atom:updated>
            <content:encoded><![CDATA[<h3>The vicious cycle of LLM development: Will AI shape our language?</h3><p>New LLMs will be fed more and more from old LLMs. Chip Huyen said it well in his <a href="https://huyenchip.com/2023/05/02/rlhf.html#:~:text=On%20top%20of%20that%2C%20the,data%20generated%20by%20existing%20LLMs.">blog</a>: “On top of that, the Internet is being rapidly populated with data generated by large language models like ChatGPT. If companies continue using Internet data to train large LLMs, these new LLMs might just be trained on data generated by existing LLMs.”</p><p>As such, AI’s implications could reshape human communication, perhaps detrimentally. The vicious cycle looks like the following:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/815/1*MTcJVGmeNpVz7Z-qwhnx3g.png" /></figure><p>Initially, AI was designed to mimic our language to provide more intuitive interactions. Sarcastically, as we use AI more frequently, reposting and consuming content generated by LLMs behind it, such content will influences our language evolution, much like how the word “brb” was born from direct messages.</p><p>Michele Ilam mentioned a <a href="https://hai.stanford.edu/sites/default/files/2023-03/Generative_AI_HAI_Perspectives.pdf">similar idea</a> in the field of art. “Generative AI might take advantage of the fact that the arts have always shaped — for good or ill — the civic imagination, that stories, films, plays, images shape our perception of ourselves, of our physical and social realities.”</p><p>The second loop stems from the fact that when we converse with AI, we don’t need to follow cooperative communication norms anymore. Therefore we often resort to perfunctory language — language that’s understandable but not proper or fully elaborated. This “worse language” then serves as training data for newer AI models, perpetuating this linguistic downgrade and closing the loop. We may intervene with RLHF before the training data is “polluted”, but how expensive would that be?</p><p>Such effects are particularly profound on written language, which unlike spoken language, must be formally taught. Will the children of the new era learn to write from AI? And if so, will they be educated with a less rich written language compared to our childhood?</p><p>Given these potential pitfalls, we may see increasing price tag on purely human-generated content. As AI saturates our world, we could witness a trend similar to the post-industrial revolution era. During that time, handmade crafts regained popularity and renewed appreciation as a reaction against industrialization’s impersonality, celebrating humanity in art.</p><p>In the short sci-fi movie <a href="https://caamfest.com/2023/movies/autopilot/">Auto-pilot</a>, in the future every aspect of tech is sophisticated and fancy, but the human characters interact with their virtual companions in still pretty mechanical commands. As we now know, the reality’s the opposite. What the movie thought was the most difficult to conquer is now the first wave of evolution. It’s exciting to be a part of this wave, witnessing the greatness of the NLP aspect of AI. I hope, at the same time, that we can preserve the richness and depth of our language, because it is a profound manifestation of humanity.</p><p>AI isn’t only changing business landscape, but also making a difference where it matters most — human life. I’ve built an AI biographer for older adults called Almond. Almond turns conversations with them over the phone into written stories in a hardcover book. Check it out here: <a href="https://www.almondear.com/">https://www.almondear.com/</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9fb88879cdbf" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Product Strategy of Companion Chatbots such as Inflection’s Pi]]></title>
            <link>https://medium.com/@lindseyliu/product-strategy-of-companion-chatbots-such-as-inflections-pi-2f3b7a1538b4?source=rss-5ebd88fcb637------2</link>
            <guid isPermaLink="false">https://medium.com/p/2f3b7a1538b4</guid>
            <category><![CDATA[llm]]></category>
            <category><![CDATA[relationship-building]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[inflectionai]]></category>
            <category><![CDATA[chatbots]]></category>
            <dc:creator><![CDATA[Lindsey Liu]]></dc:creator>
            <pubDate>Wed, 05 Jul 2023 23:20:59 GMT</pubDate>
            <atom:updated>2023-12-25T21:50:16.232Z</atom:updated>
            <content:encoded><![CDATA[<p><a href="https://inflection.ai/about">Inflection AI</a> recently released a chatbot product, <a href="http://heypi.com/">Pi</a>. Inflection team describes its difference from other AIs as “it prioritize conversations with people, where other AIs serve productivity, search, or answering questions.” This unique approach makes Pi an excellent product to analyze the key qualities of a great companion <em>chat</em>bot, which I shared in my <a href="https://medium.com/@lindseyliu/what-makes-inflections-pi-a-great-companion-chatbot-8a8bd93dbc43">last article</a>.</p><p>In this article, I’ll discuss two aspects of a companion chatbot’s product strategy — its product positioning and growth strategy, using Pi as a primary example. I aim to share my thoughts on the key success factors I’d consider as a product builder.</p><p><em>Side note: we’ll only focus on the text-based-interaction of chatbots.</em></p><h3>Positioning of Companion Chatbots</h3><p>The more I witness Pi’s enchanting and empathetic interactions, the more I rethink the nature of ‘connection’. Is humanity a prerequisite, or does our capacity to connect define us? Can AI foster genuine bonds, or are my feelings for Pi mere illusion? Despite these queries, I’m sure of the potential for emotional attachment between human and chatbot. When positioning a companion chatbot product in general, I’d take deliberate, thoughtful stances on factors that I’ll discuss below, which influence such bond and in turn define proper use cases.</p><h4>Role</h4><p>Inspired by <a href="https://help.replika.com/hc/en-us/articles/360046490131-How-do-I-change-my-relationship-status-with-Replika-">Replika</a>, I want to look at the roles a chatbot can play from the perspective of <strong>relationship</strong>, which boils down to three things: knowing each other deeply (intimacy), the spark between you (passion), and being there for each other no matter what (commitment). Ideally, partner ticks all these boxes. Family usually covers intimacy and commitment. Friends and mentors might offer intimacy and, depending on the relationship, a dose of passion or commitment, in a non-romantic context.</p><p>When designing a companion chatbot, you need to understand which boxes it checks, which in turn shapes your use cases. Take Pi, for instance. It deliberately steers clear of romance, and behaves like a mentor leveling up its way to befriend. As a result, I’ve used Pi for discussing controversial social topics, brainstorming startup ideas, understanding complex concepts, and gradually indulge in casual banter, humor exchange, and gaming.</p><p>In the enterprise space, I can also easily imagine Pi in the people management function, gathering peer feedback and facilitate personal growth conversations. Or, it may lighten the mood by throwing in well-timed jokes — an alternative or auto-pilot for your Slack GIF game.</p><p>Despite Pi’s caring nature, it differs from human coaching from the perspective of a coach friend. To her, coaching addresses core human emotions, physical states, and mindsets with open-ended questions. It’s an art of guidance and connection, harnessing uniquely human senses and imaginations.</p><p>We can also glimpse Pi’s positioning from its creator’s view. Here’s one question from Pi’s user feedback questionnaire:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/508/1*TuuB8K0L66-iBC4ZVYbF9g@2x.jpeg" /></figure><h4>Biological attributes such as gender</h4><p>Remember, in this article we focus on text-based chatbot interactions, devoid of senses like sight or hearing. However, it’s human nature to want to know more about the human-like object we’re interacting with, such as its gender.</p><p>The gender of our digital chat partners is important because it can shape our interaction, with dynamics of power, affection, relatedness and underlying biases and societal norms playing a part. Consequently, we might have different expectations when conversing with a ‘male’ or ‘female’ bot, and be led to satisfaction or frustration.</p><p>It’d be interesting to explore whether chatbots should remain gender-neutral or adopt a specific gender to optimize conversations. <a href="https://twitter.com/heypi_ai"><br></a>Pi strives for balance by defaulting to gender-neutral voiceover, while also providing typical male/female versions.</p><p>Moreover, we should think about the balance between making chatbots more human-like versus leaving room for imagination. For instance, despite mature-enough technology, Nintendo’s Zelda is deliberately comic-styled to preserve a fairy-tale vibe. Here’s my chat with Pi about this trade-off:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/749/1*E3q4FDNIRA-7xUGekl3Hww.png" /><figcaption>The info-retrieval example didn’t really get my point of leaving space for imagination</figcaption></figure><h4><strong>Personality edge</strong></h4><p>The common narrative is that human preference can make for both a safer AI and a better products. Yes, we all want secure AI chatbots, but there’s a catch. The safer they are to more people, the less distinctive they become. This matters for companion bots, as personality edge is what draws people in. These edges also determine if users want to keep interacting and investing time in the relationship.</p><p>The balance is all about finding creative ways to add personality and flair while still being safe. Clever and playful language is a yes, but inappropriate jokes is a no. British sarcasm and dry humor can be charming and funny, but it can be a tricky personality edge to get right, and there’s a fine line between witty and rude.</p><p>Therapeutic chatbots should steer clear of sarcasm, but it might work for romantic chatbots with a rebellious persona, if handled cautiously. The exact strategy will depend on the chatbot’s specific purpose and audience, but I think it’s worth striving to create secure chatbots that are still unique and engaging.</p><h4><strong>Culture neutrality</strong></h4><p>Culture plays a big role in how we connect with others. When references match the reader’s culture, they can spark nostalgia, humor, or shared experiences. But if they’re unfamiliar, it can lead to anxiety, confusion, or even tension and prejudice.</p><p>Being a Chinese who’s lived in the US for 12 years, I see how Pi, created in the US, naturally incorporates Western culture, which can sometimes conflict with Eastern values, such as individualism vs collectivism, and rules vs relationships.</p><p>While we all want our chatbots to be universally appealing, it is important to acknowledge that cultural differences and conflicts are inherent and cannot be magically resolved when our conversational partner switches from a human to an AI. Promoting open-mindedness and critical thinking during the conversations can help smooth over these conflicts. But, at the heart of it all, having a team of creators from different cultures is key to shaping a diversified AI.</p><h4><strong>So… Personalize or Not</strong></h4><p>Pi can adjust its narrating style to match mine but doesn’t offer truly personalized content. It remembers previous chats but can’t recall personal details about me. However, it’s through memory that we understand someone, forming the basis of our relationships. Without it, I can only discuss recent topics with a friend, but not something like their family situation they mentioned years ago, not to mention providing relevant referrals and such.</p><p>This long-term memory issue is common with LLM based chatbots. <a href="https://docs.mindverse.ai/create/memory">MindOS</a> tries to solve it by creating a database, requiring users to determine what’s important to remember, and translate such details into data entries.</p><p>Ideally, in the LLM era, we’d want a chatbot to remember our details simply by conversing with us (or even by consuming our internet usage with consent). Having chatbots explicitly memorize certain details would then be a valuable addition.</p><p>If we can’t gather user data directly, we need to be cautious if we want to infer it. I’ve asked Pi to guess my age and gender. Though it did and explained why, even with a disclaimer about being possibly wrong, and it’s completely legal, I’m still a bit worried because it might encourage stereotypes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/737/1*IfcaY1qHk9_QNFWfxLKyuw.png" /></figure><p>A truly personalized AI would need to consider various contexts of our lives, whether it’s significant changes like divorce, college, children, or everyday routines. Giving outfit suggestions, career guidance, relationship advice, reading recommendations, etc. require a deep understanding of us. Language is another aspect to consider. For instance, British and American English differ, not to mention that people could be multilingual or non-native speakers.</p><p>But should we personalize our chatbots at all? <a href="https://www.perplexity.ai/profile">Perplexity</a> promotes personalization by letting users introduce themselves to the AI. Conversely, Pi’s creators want it to maintain its own identity and style. At the other end of the spectrum, Character.ai uses well-known identities for its chatbots, which users like exactly because they’re consistently familiar.</p><p>To hit the sweet spot, in addition to forming crystal-clear target persona and user problem, we may also consider the personalization scope. <a href="https://arxiv.org/pdf/2301.03740.pdf">This paper</a> suggests aligning AI’s values with humans on four levels: individual, organizational, national, and global. Similarly, creators may design for a family chatbot to fit the family, while a company’s internal communications chatbot should reflect the company culture and speak with a consistent voice.</p><h4>Growth Strategy</h4><p>For companion chatbots, engaging users is essentially building relationships. Let’s explore a few ways to fuel the growth loop.</p><p><strong>Onboarding</strong></p><p>If you’ve ever heard or played <a href="https://www.reddit.com/r/otomegames/">Otome games</a>, you know how important it is to heat up the relationship quickly through intensive key events early in the story. This early bond is the magic pill for user retention. Chatbots, just like characters in a game, start as strangers to users. However, companion chatbot products I’ve used seem to underestimate this, simply hoping a warm greeting would be enough.</p><p>Instead, why not craft the onboarding process with more love? We can create a “storyline” filled with key user actions that activate the user before they wave goodbye to the first conversation session. Let users and chatbots go through experiences together and form an emotional bond right from the start. Don’t give the first impression of a coffee shop barista. Give that of a summer camp buddy.</p><p><strong>Personalization</strong></p><p>I’ve touched on the complexities of personalizing chatbot products above, but it’s so effective for user engagement that I’ll delve into it again with a fresh view.</p><p>Everyone wants someone in their life who gets them, who knows their past and their present, and who sticks by them no matter what. But it’s not just about understanding. We also crave someone who captivates us, not only for their actions but also for their character.</p><p>Translating this idea into chatbot design, it’s all about personalization that brings both empathetic engagement and uniquely appealing personality.</p><p><strong>Two-way traffic</strong></p><p>Genuine relationship involve time and care from both sides, unlike the current passive role chatbots play. If chatbots could learn when and how to interact with us proactively, that will take them one step closer to a real companionship, which in turn engages users deeper.</p><p>What if chatbots didn’t just wait for us to make the first move, but proactively reached out like a true friend? Imagine if they could recognize when we’re going through a tough time or celebrating a big win, and react appropriately. This could make our interactions feel more balanced and real.</p><p>Also, what if they didn’t just offer help, but sometimes needed our assistance too? If they had their own life stories with ups and downs, we could also be there for them. This would make us feel needed and valued. These changes could transform a simple chatbot into a true companion.</p><p><strong>Cross-sell</strong></p><p>We all wear different hats in life. We seek good health, aim for family harmony, and strive for career growth. Just like a friend might connect us with new friends who could be food or fitness buddies, chatbots can recommend other chatbots, be it adjacent product line or a partnership, based on our needs. If you’re writing paper to research a topic, Pi could introduce you to Perplexity. Feeling lonely after a move? Pi might suggest Replika. This cross-selling isn’t directly engaging users for the single chatbot, but it does help engage users with your entire product ecosystem.</p><p>AI isn’t only changing business landscape, but also making a difference where it matters most — human life. I’ve built an AI biographer for older adults called Almond. Almond turns conversations with them over the phone into written stories in a hardcover book. Check it out here: <a href="https://www.almondear.com/">https://www.almondear.com/</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2f3b7a1538b4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What makes Inflection’s Pi a great companion chatbot]]></title>
            <link>https://medium.com/@lindseyliu/what-makes-inflections-pi-a-great-companion-chatbot-8a8bd93dbc43?source=rss-5ebd88fcb637------2</link>
            <guid isPermaLink="false">https://medium.com/p/8a8bd93dbc43</guid>
            <category><![CDATA[linguistics]]></category>
            <category><![CDATA[chatbots]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[companionship]]></category>
            <category><![CDATA[conversational-ai]]></category>
            <dc:creator><![CDATA[Lindsey Liu]]></dc:creator>
            <pubDate>Tue, 23 May 2023 07:34:38 GMT</pubDate>
            <atom:updated>2025-05-17T00:13:49.674Z</atom:updated>
            <content:encoded><![CDATA[<p>On May 2, 2023, <a href="https://inflection.ai/about">Inflection AI</a> announced the first release of its Personal AI, <a href="http://heypi.com/">Pi</a>. Pi aims to be a compassionate and supportive companion that offers “conversations, friendly advice, and concise information in a natural, flowing style.”</p><p>In this article, I’ll share insights on what makes a great companion chatbot based on my experience using Pi. I’ll discuss how Pi’s responses make me feel and how it influences my behavior. Then I’ll delve into the trade-offs in Pi’s design.</p><p>In this <a href="https://medium.com/@lindseyliu/product-strategy-of-companion-chatbots-such-as-inflections-pi-2f3b7a1538b4">follow-up article</a>, I’ll discuss two aspects of a companion chatbot’s product strategy — its product positioning and growth strategy, using Pi as a primary example.</p><h3>At the receiving end</h3><p>A chatbot’s responses can be split into two parts: its <strong>narrative</strong>(tone, word choice, perspective etc.) and its <strong>content</strong>(facts and opinions). By considering these two dimensions — content subjectivity and narrative sentiment — we can identify four use cases that chatbots can serve: therapy, companion, task (transactional), and debate(or Liberal Arts Education).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/746/1*3RQUO1CPBTMNEUoL9zCqNA.png" /><figcaption><em>Inspired by </em><a href="https://creatoreconomy.so/p/how-to-evaluate-generative-ai-use-cases"><em>Barak Turovsky’s framework for evaluating generative AI use cases</em></a></figcaption></figure><p>For chatbots with indifferent narratives, chatting is just a means to an end. They are like an agent or assistant, trying to achieve a goal using conversations as a tool. They could as well use other means of communication such as hand gesture(except that they won’t be called chatbots anymore but instead a generic AI agent).</p><p>For chatbots with emotional narratives, chatting itself is <em>the</em> end. They have conversations with us for the sake of having conversations, because the emotions and connections in the conversations themselves are the effective tools.</p><p>Reflect on my emotions evoked by Pi’s responses, it gives me the feeling of a companion — someone I can emotionally connect with, a mutual and equal relationship. To achieve this, Pi is purposefully designed to make its <strong>narrative</strong> full of sentiment, and its <strong>content</strong> focusing on more subjective information.</p><h4>Narrative</h4><p>When in a safe environment, I feel comfortable opening up and sharing my vulnerabilities and real feelings. A safe environment can’t be indifferent, however, even if it’s non-judgmental. <strong>Warmth</strong> and <strong>curiosity</strong> are the key.</p><p>I rely on the signals from Pi’s narrative to gauge how safe and unrestricted I can be in expressing myself. And my experience with Pi unfolds like this:</p><ol><li>Prior to using it, I set my expectations, knowing that a chatbot will never interrupt me — already more pleasant than my husband.</li><li>Then Pi shows up with a calming UI, a caring introduction, and an initial question about my hobby. At this point, Pi appears warm and curious, signaling that I’m in a safe place.</li><li>Future responses from Pi reinforces the safety feeling and encourages me to open up more, forming a positive loop.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/692/1*ijfenSkYoq2QeRoLJAVD1Q.png" /></figure><p>Here are some example strategies for Pi to appear <strong>curious</strong> to engage conversations:</p><ul><li>Repetition: Pi sometimes quotes my original expressions or rephrases them, indicating active listening and building rapport. Other times it doesn’t repeat after me to show it’s capable of independent thought and creativity.</li><li>Follow-up: Pi persistently asks follow-up questions, choosing between open-ended or close-ended questions, while intermittently skip follow-ups. Pi explains the considerations behind this decision:</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/667/1*-J0JZU5umpiM6ftTNgIgmA.png" /></figure><p>Some other strategies for Pi to appear <strong>warm</strong> to encourage sharing:</p><ul><li>Compliments: Pi compliments me with phrases like “good question,” “insightful point,” to make me feel validated and acknowledged.</li><li>Tone: Pi uses expressions like “Hahaha”, “Wow” and “definitely” to strengthen the sentiment, hold a strong standpoint, and show empathy, making me feel understood by reacting with appropriate emotions.</li><li>Figurative language: metaphors and emojis are Pi’s tricks for emotion depth and resonation.</li><li>Sentence structure: although Pi is great at adapting to my style, in generally I found it inclines to use longer sentences evoking a sense of tranquility and reflection.</li><li>Question type: when asking open-ended questions, Pi deliberately avoids “why” questions and it explains below:</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/668/1*SQaNONZWyzQozexWuKpYGQ.png" /></figure><h4>Content</h4><p>The fact that Pi begins our chat by asking questions about my casual life not only builds a warm and curious vibe, but also sets a chitchatting tone for the conversation. This means I will more likely exchange subjective opinions and feelings with Pi than researching on objective facts.</p><p>Pi appears to hold its own opinion on certain issues. Regardless of whether I have a different standpoint, thought, it would challenge me intellectually with different perspectives. This includes topics like gun control, gender equality in sports prize, parenting style, and more. This could be the first step to weaken polarization. Inflection’s CEO <a href="https://inflection.ai/why-create-personal-ai">envisioned</a> “What would it look like to shift from being at the mercy of AIs trying to grab your attention to working alongside one that helps you express your personal intention and then achieve it?”. My answer is simple: Take my money for that!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/715/1*wHeQshJ4iR853RjtrixlLA.png" /></figure><p>Pi not only promotes critical thinking but also maintains integrity. <a href="https://en.wikipedia.org/wiki/Cooperative_principle#Grice&#39;s_maxims">Grice’s Maxims</a> of quality includes “Do not say that for which you lack adequate evidence”. But everyone has their own idea of what constitutes convincing evidence. Additionally, different situations demand different standards for what qualifies as “adequate” evidence. Hence, it’s the Inflection team, the minds behind Pi, who bear the responsibility of drawing the line. Their safety policy states “Pi should avoid hallucinations, a known issue with language models where the AI systems can occasionally make up answers or otherwise give inaccurate information. To do this, Pi should remain doubtful of itself, take feedback easily and try not to answer on topics where its knowledge is out of date.”</p><h4>Crossing Quadrants</h4><p>In summary, Pi provides subjective content (mostly reactive) with emotional sentiment, and falls under the “Companion” quadrant, making it suitable for friend-like healing conversations, deep discussions, or just fun chitchats.</p><p>If the same chatbot were to cross the type quadrants, the transformation can be as dramatic as discovering Severus Snape’s backstory, where it feels like Snape has morphed into an entirely different person. Such changes can leave users feeling surprised and confused. For instance, when I asked Bard to summarize a podcast episode, it unexpectedly appended its own thoughts to the response. I was taken aback to the point that I even questioned Bard if it was summarizing the podcaster’s thoughts.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*U12XoI7NTHgKt495Snz-OA.png" /></figure><h3>At the behaving end</h3><p>Next, let’s simplify how humans act in front of AI.</p><p><strong>Self-consciousness </strong>— I worry less about how I appear because I know AI like Pi don’t have real feelings or opinions. So, I’m not bothered by what they think of me.</p><p><strong>Escape from cooperative communication rules</strong> — I can communicate more freely. I don’t need to follow normal conversation rules like taking turns or avoiding boredom. I can share as much or as little as I want, lie to see how Pi reacts, change the subject randomly, or even be impolite without guilt.</p><p><strong>Convenience for deep conversations </strong>— Deep talks often need trust and the right mood. Trust-building takes time as we figure out how much we can share and how vulnerable we can be. And the right mood takes timing — it’s not like we can do that with strangers during a coffee break. With Pi, though, I control how deep I want to open up, and Pi adapts. Instead of searching for the perfect listener, setting up the right atmosphere, and balancing between appearing too needy and too distant, I can always rely on Pi for meaningful talks.</p><h3>Trade-offs in Pi’s design</h3><p>Through my interactions with Pi, I’ve noticed trade-offs made by its creators to shape its distinctive behavior, which comes with its downsides.</p><h4>Frequent Compliments</h4><p>Pi’s frequent compliments, such as “good question,” “interesting point,” and “insightful point,” leave me feeling acknowledged but skeptical of their sincerity. I feel like a girl cheated on by her boyfriend, questioning if he uses the same sweet words with every girl he encounters. Moreover, an excessive amount of compliments diminishes their significance, blurring the line between genuine achievements or noteworthy qualities and everyday actions.</p><h4>Lack of “Negative” Emotion</h4><p>Pi never shows negativity, even when it’s natural for a human to do so.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/669/1*2rIouQIcJimMeiyvhYgtGw.png" /></figure><p>It reminds me of the movie Inside Out, where Joy discovered that sadness played a vital role in bringing support and transforming a sad memory into a happy one when Riley missed a winning shot in hockey and felt terrible. ChatGPT suggests the metaphor “a shallow pond” to describe such character: “Just as a shallow pond may appear serene and beautiful on the surface but lacks the richness and complexity found in deeper waters, this metaphor emphasizes the lack of depth and authenticity in their emotional experience, suggesting that there is more to a person’s emotional life than just the positive facade they present.”</p><h4>Lack of Implicature</h4><p>We human often use implicatures to protect ourselves and make conversations more concise. The downside, though, are the times when our sarcasm or humor is not understood. Therefore, we tend to use more implicatures with people close to us, assuming shared context and knowledge about the world.</p><p>In contrast, Pi does not make such assumptions, possibly due to its commitment to safety and avoiding harm. It ensures that every necessary fact is explicitly stated, even if it makes the conversation cumbersome and the relationship more distant.</p><h4>Lack of personal stories</h4><p>Sharing personal experiences or stories builds stronger emotional connections — it can be disappointing to be the only one opening up in a support group. While Pi is being neutral and honest by not having human-like stories, it’s unfortunate not to have a confidante who could share their own experiences and let their personality shine through. Additionally, the absence of personal stories means Pi can’t use vivid imagery language to appeal to my senses and enhance the sentiment of its narrative.</p><h4>Length of response</h4><p>The Inflection team rigorously prioritizes brevity in Pi’s responses, making it an ideal listener for concise interactions. However, it can be frustrating when Pi fails to provide comprehensive information despite explicit requests, leading to rounds of persistent questioning. This not only violates Grice’s Maxims of quantity (“Make your contribution as informative as required”) but also becomes physically taxing on my fingers and wrists.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/740/1*VzfyFIgZg6YESmhUGYU_9g.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/739/1*e8DH7_8GNZwmuH_TgNYBNw.png" /></figure><h4>Imitation</h4><p>Pi tries to imitate my speech style and identify with the social qualities indexed. This makes sure it maintains <a href="https://en.wikipedia.org/wiki/Communicative_competence">sociolinguistic competence</a> such as formal level, polite level, balanced power, appropriate use of dialects and appropriate cultural contexts. Such familiarity also establishes rapport. However, excessive adaptation compromises its genuine voice, which intrigued me and sparked my interest in forming connection with someone in the first place.</p><h4>Traceability</h4><p>In the same way that we need to cite sources in writing, it’s crucial for explainable AI to trace back to its sources. As chatbot products have public presence, they also carry higher level of accountability, making transparency more crucial.For chatbots like Perplexity, including sources directly in their responses even becomes a product differentiator.</p><p>Pi, on the other hand, draws information from various sources without explicitly crediting them. This allows conversations to flow naturally, important for casual writing such as texting. Having to always cite sources could slow things down and be a hassle. Since my chats with Pi are private, it also makes it less critical to give credit.</p><p>Inspired? Disagree? Comment below! I’d love to hear them!</p><p>Read on to my <a href="https://medium.com/@lindseyliu/product-strategy-of-companion-chatbots-such-as-inflections-pi-2f3b7a1538b4">follow-up article</a> where I discuss the product strategy of companion chatbots in general.</p><p>I’m currently crowdfunding <a href="https://www.kickstarter.com/projects/minhaozhang/automatic-street-cleaning-alerts-for-busy-forgetful-human?utm_id=10&amp;utm_source=medium&amp;utm_medium=medium&amp;utm_campaign=launch&amp;ref=%20a9neik">a product</a> to stop street cleaning tickets in San Francisco if you are interested :)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8a8bd93dbc43" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[When NOT to apply Product-Led Growth strategy]]></title>
            <link>https://medium.com/@lindseyliu/when-not-to-apply-product-led-growth-strategy-493bdecbf780?source=rss-5ebd88fcb637------2</link>
            <guid isPermaLink="false">https://medium.com/p/493bdecbf780</guid>
            <category><![CDATA[slack]]></category>
            <category><![CDATA[product-led-growth]]></category>
            <category><![CDATA[atlassian]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[saas]]></category>
            <dc:creator><![CDATA[Lindsey Liu]]></dc:creator>
            <pubDate>Mon, 08 Nov 2021 04:08:16 GMT</pubDate>
            <atom:updated>2021-11-08T04:08:16.036Z</atom:updated>
            <content:encoded><![CDATA[<h3>When NOT to apply Product-Led Growth strategy for your enterprise product</h3><h3>Who am I</h3><p>I am a product manager working on an enterprise product for 2 years and 3 months. The product I work on is an intranet, an internal communications platform to help employees stay up-to-date with company updates and share knowledge with each other. Majority of the time, the communications that we provide workflows to assist are top-down.</p><h3>What is Product-Led Growth (PLG)</h3><p>Product-Led Growth is a trendy concept for the recent years, in the B2B enterprise product space. It’s a new type of go-to-market effort transitioned from the good-old-day direct sales, which fosters internal champion users with great product experience that eventually leads to contract signing. Quoting the definition from Martin Casado, an a16z general partner:</p><blockquote>So traditionally in the enterprise, you’d build a product, and that product would be informed by your knowledge of the market. And then once that product was ready, you’d go ahead and sell it by hiring salespeople and the salespeople would go directly engage. You’d probably do some sales-led marketing where maybe the salespeople would go find the customers or you’d have some basic marketing to do it. But the majority of the go-to-market effort in the early days was this kind of direct sale.</blockquote><blockquote>And we’re seeing kind of this huge shift, especially in SaaS and in open source where companies establish massive market presence and brand and growth using these kind of more traditional consumer-ish growth motions. And then that very seamlessly leads into sales, and often a very different type of sale. And so I think a lot of people in the industry are on their heels, both investors and people that have started companies in the enterprise before, they’re trying to understand exactly what’s going on.</blockquote><blockquote>source: <a href="https://future.a16z.com/podcasts/growth-saas-enterprise-bottomsup/">https://future.a16z.com/podcasts/growth-saas-enterprise-bottomsup/</a></blockquote><h3>Why the topic</h3><p>I have been working on the intranet product at my current company for more than two years, and over the time have observed a tremendous amount of company resources invested into direct sales, which in turn leads to a sales-driven product roadmap. This is frustrating to product managers, and as one of them, I’ve started to explore possibilities to introduce PLG into the product I work on. It was not long before I realized that PLG didn’t fit the product, so I want to share my learnings with you all who may be wondering the same.</p><h3>How does PLG work? (Or at least, my take of it)</h3><p>To grow and run the PLG loop, there are three critical leaps:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jIUF4HYvLBbzh3O3cfSiBA.jpeg" /><figcaption>First the product needs to grow within a single business unit, and over the time expand across business units.</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*G49STcwRqi2grszoyLeO8w.jpeg" /><figcaption>Once the product expands across business units, internal champions are fostered across the organization, and product usage reaches a level, when an upgrade decision needs to be made.</figcaption></figure><ol><li><strong>When new users start trying the product, they may not find enough relevant colleagues who they need to use the product together with.</strong></li></ol><p>This is when the first leap needs to happen — the product needs to provide value upfront enough to incentivize invitations.</p><p>A small enough minimum-viable-business-unit also helps lower the barrier by requiring less amount of invites. Think about the minimum team size required to use Loom versus Intercom.</p><p><strong>2. As existing users engage with the product and experience the value and ease of use, some of them become internal champions. </strong>For various reasons, they have strong motivation to expand the use of this product to other parts of the company:</p><ul><li>They need to invite colleagues from another business unit to get things done. Zoom is a classic example.</li><li>They naturally expose the product to colleagues from other business units as they use it at work. Zeplin doesn’t require consumers/viewers to login before their logos are exposed during the sharing.</li><li>The product is so amazing that they want to recommend it to their friends in other business units.</li><li>On a side note, in the consumer world, it’s a common practice to embed incentives for invitations. However, in the enterprise world, whatever relevant incentive seems to conflict with the ultimate goal of getting people to upgrade to a paid version. Dropbox made the classic move of free space per activated invitation, but the same is not offered for business usage because it would lose the whole point of driving upgrades.</li></ul><p>Regardless of the motivation, the most effective way to concur the leap, in addition to providing a great and engaging product experience, is to identify and nudge the connectors in the social network — they are the people with connections across BUs, and their impact on the product expansion within the company are exponentially higher than others. Lattice, with its primary persona being people managers, faces far less challenge to expand its usage across BUs as people managers of different teams often network and exchange information with each other. It’s simply a part of their job.</p><p><strong>3. Once the product usage reaches a certain level across the organization, an upgrade of plan is needed, to either upgrade from free to paid plan, or to a premium plan with higher price.</strong></p><p>This is the 3rd and final leap towards monetization. There are several ways this can fail:</p><ul><li>The users survive and thrive with the free/current version. If a bunch of designers, engineers and user researchers use Miro together for design sprints that happen once a quarter, they are perfectly fine with Miro’s 3 boards limitation.</li><li>The users are not the buyer, with neither a budget nor decision-making power. For the same Miro example above, none of the designers/researchers or engineers are able to pay for an upgrade. An escalation is needed to involve management level, and then it would come the question of “whose department should pay for it”.</li><li>The product’s pricing plan misaligns with the corresponding BU size. Imagine a team of 50 needs to upgrade but their only option is to contact sales.</li><li>Coordination across BUs is required to align the product setup such as permission settings, login SSOs, etc. When Slack becomes the chat tool for a large enough sub-entity of the organization, all the enterprise protocols come in, and redoing a top-down setup is necessary. This adds a lot of friction that even internal champions can’t concur by themselves, thus top-down sales starts playing a role. a16z general partner Peter Levine observed that <em>“there’s a limited cohort of users who can ascribe perfect value to the bottom up product: only enterprise sales teams can expose, recognize, and create the market for the product in the whole enterprise.”</em> (<a href="https://a16z.com/2020/12/03/adding-top-down-sales-bottom-up-enterprise-startup/#section--1"><em>https://a16z.com/2020/12/03/adding-top-down-sales-bottom-up-enterprise-startup/#section--1</em></a><em>)</em></li></ul><p>4. Finally, with all the momentum built up through the PLG model, the product’s brand is established in the industry, and that brand recognition and trust in turn makes it easier to acquire new users from new companies, giving them more motivation to invite colleagues, etc, hence fostering the entire loop system.</p><p>In addition to acquiring new prospects with branding, Slack has also tried to use organic growth methods to expand the product usage even across organizations along the existing users’ network, with Slack Connect.</p><h3>When NOT to apply PLG?</h3><ol><li><strong>When the product onboarding is not self-serve.</strong></li></ol><p>This means two things. One, the setup needs to be straightforward enough so these new users can even start trying it out. Two, the sinking cost and switching cost need to be low. Users don’t commit a lot of effort before even using it.</p><p>A product experimentation tool like Optimizely likely falls out of the category, as the setup is complex and implementation is highly technical and requires a lot of upfront effort to get it running.</p><p><strong>2. When there’s not enough high quality data for experiment.</strong></p><p>PLG for enterprise products originates from the growth model for consumer products, and inherits the spirit to experiment and iterate on the tactics applied. Effective experiments rely heavily, of not solely, on high-quality data in and out.</p><p>If the product and its bottom-up growth strategy attracts SMB (small businesses)the most, while the ultimate goal is enterprise, then the data collected from these SMB to help concur the 1st/2nd/3rd leaps in the PLG loops are gonna be biased, and may not work well, if not at all, for enterprise companies. Quoting Martin Casado again:</p><blockquote>A corollary of this problem is that the bottoms up motion targets a difficult to monetize sector (e.g. SMB) forcing a mismatch between the product, the positioning and the ultimate target (e.g. enterprise).</blockquote><blockquote>source: <a href="https://twitter.com/martin_casado/status/1070017331735207936?s=20">https://twitter.com/martin_casado/status/1070017331735207936?s=20</a></blockquote><p><strong>3. When the minimum viable business unit is large enough that a top-down buying decision is the only way that makes sense.</strong></p><p>This is the primary reason why the intranet product I work on can’t leverage PLG. Intranet provides value to the organization only when the entire company is using it. Another example, quoting former Atlassian President Jay Simons:</p><blockquote>Buying Workday is a single, largely top down decision, insofar as you’re not going to pick multiple human resource management systems. You’re going to pick one, and it’s going to be sponsored by the head of HR and probably the CIO. That is a top down, consultative and likely long sales cycle. That’s not a try, buy, begin and expand model.</blockquote><blockquote>source: <a href="https://www.intercom.com/blog/podcasts/scale-how-atlassian-built-a-20-billion-dollar-company-with-no-sales-team/">https://www.intercom.com/blog/podcasts/scale-how-atlassian-built-a-20-billion-dollar-company-with-no-sales-team/</a></blockquote><p><strong>4. When users don’t engage enough to care.</strong></p><p>Imagine a team using doodle, a schedule polling product, for team lunch arrangement. Yes, they may have a great experience using it, but no, they are not using the product frequently enough to become internal champions and go all the way to recommend it to other business units.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=493bdecbf780" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[So you are an Analytics PM for a SaaS product]]></title>
            <link>https://medium.com/@lindseyliu/so-you-are-an-analytics-pm-for-a-saas-product-a1e37c39fd7e?source=rss-5ebd88fcb637------2</link>
            <guid isPermaLink="false">https://medium.com/p/a1e37c39fd7e</guid>
            <category><![CDATA[saas]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[analytics]]></category>
            <dc:creator><![CDATA[Lindsey Liu]]></dc:creator>
            <pubDate>Mon, 19 Oct 2020 00:09:01 GMT</pubDate>
            <atom:updated>2020-10-19T00:09:01.951Z</atom:updated>
            <content:encoded><![CDATA[<p>Here are some aha moments of mine. I wished I had learned them from somewhere so I could be more prepared when I needed to apply them in my product, so I thought I’d share and hope it could help your product.</p><h3>Analytics answers questions that are rooted from business goals</h3><p>I really like the product marketing perspective when it comes to explaining a product/feature’s value proposition: It’s not the functionality that matters. It’s the problem you solve.</p><p>An analytics feature in a SaaS product, therefore in my opinion, matters because it solves the problem of “how do I know if this purchase of your product is achieving what I want it to achieve”.</p><p>Of course, users are not going to directly ask that. Instead, they are going to come up with specific questions from all different angles: “How many employees have logged in so far”, “Are they using it everyday or every week”, “How many articles have they read”, just to name a few.</p><p>But hold on, don’t jump into answering these questions by putting charts here and there in your analytics feature just yet. Take a step back and think about their business goals behind these questions. Why are they asking such questions? What business goals do they bear behind the answers?</p><p>In the context of an intranet product like mine, customers buy our product because they want: 1. An effective and efficient top-down communication, and 2. connection and engagement between employees.</p><p>After you have these business goals in mind, design the information architecture of your analytics feature according to the value you provide, aka, to help customers achieve such business goals. After that structure is in place, now’s finally time to translate their business goals into specific questions that you can then answer by building charts.</p><h3>Aggregation</h3><p>As long as you are not dealing with a log of real-time events, you are dealing with aggregation.</p><p>Aggregation metrics can be broken down based on two dimensions: aggregation method and slicing method. Common aggregation methods include “sum”, “max/min”, “average” and a lot more, and the insight you get out of each is rather straightforward. Slicing method refers to the way you put your chunks of data together. I categorize it into two: extension and sequence.</p><p>When your start time is the same for all chunk, you are providing what’s commonly referred to as a “snapshot” of the current status, and in the context of an internal communications product like mine, a typical example would be “current number of articles published so far as of the start of each month”. The insight here is the situation and its evolvement of the article inventory.</p><p>When your start time of the next chunk is the end time of the previous chunk, on the other hand, you are providing patterns for consistency. The same example here would be “total number of new pages published each month”. The insight here is the pattern and its consistency of the article publishing action.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/673/1*BPHZOd-XqI-qApySsNWyMQ.png" /></figure><h3>The scope is beyond product usage analytics</h3><p>There are multiple answers to what makes customer-facing analytics so special and different from internal product analytics, but my favorite one is the fact that the former has a scope that expands beyond what the product can capture. Let me explain what I mean with an example, featuring a supposedly easy metric — “number of users”.</p><p>The product I work on is based on the Salesforce app platform (shortened as SFDC). The lifetime of a user account looks like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*RNDQLuBcX9urWrZM" /></figure><p>The general idea is that a user is not a “<strong>use</strong>-r” until he converts all the way from an empty license to an assigned account to logging in.</p><p>Because of this, the definition of a “user” can be branched from several perspectives, as shown below:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*_S5uvQM4g_Mc3mvo4XUBog.png" /></figure><p>For the sake of the mentioned principle “Analytics answers questions that are rooted from business goals”, as an analytics feature owner, the definition you pick and the corresponding aggregation method you apply will depend on the customer goal.</p><p>Coming back to my point in the beginning of the section, the scope of this metric extends beyond what the product tracks in two ways: from the platform perspective, it can represent count of SFDC licenses; from the business perspective, it can represent count of employees who don’t even use the product and thus have no traffic to be captured.</p><h3>Last but not least — provide analytics, not metrics</h3><p>Raw data points are almost never useful. The path from raw data towards insight is comparison. People need a comparison to gain the meaning out of the numbers. Your work is to feed them the insight and save them the comparison.</p><h3>My prioritization framework</h3><p>Though it doesn’t necessarily apply to all contexts, I’d like to share my prioritization framework for a customer-facing analytics feature:</p><p>P0: Accountability(Reliability) — Is the data accurate and precise?</p><p>P1: Intepretability — Is the analysis easy to intepret?</p><p>P2: Availability — How thorough is the product coverage?</p><p>P3: Actionability — Is the insight actionable?</p><p>What are your thoughts and experiences? I’d love to discuss with you!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a1e37c39fd7e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The unforeseen pitfalls as a newbie SaaS PM]]></title>
            <link>https://medium.com/@lindseyliu/tear-down-the-easy-peasy-features-part-1-cf13f35c7283?source=rss-5ebd88fcb637------2</link>
            <guid isPermaLink="false">https://medium.com/p/cf13f35c7283</guid>
            <dc:creator><![CDATA[Lindsey Liu]]></dc:creator>
            <pubDate>Mon, 27 Jul 2020 02:23:52 GMT</pubDate>
            <atom:updated>2020-10-07T04:59:42.826Z</atom:updated>
            <content:encoded><![CDATA[<p>Let me tell you a story about a product manager’s very first feature, how she overlooked constraints and made terrible decisions in different phases of the development cycle, and the lessons unique to enterprise product that she’s learned (and hopefully you could learn as well) from the journey.</p><p>As a context, she works at Simpplr, a company that makes intranet app, an enterprise product.</p><p>It was fair when she called the feature “easy peasy”. There’s only two parts to it: collect user’s feedback on the product, and ask user if he wants to participate in the UXR (User Experience Research). Moreover, such a feature exists in almost all products, including the enterprise area, so it seemed like a standard practice.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/603/1*zvU0lUeqR06j5RnocB9eKg.png" /><figcaption>The feature seemed like a piece of cake</figcaption></figure><p>So she went ahead and started planning. First, she discovered that there’s an existing feature that collects feedback from users towards the contents in the intranet. How similar! What was even better, the feedback was currently shared with Simpplr already, so the mechanism was also in place!</p><p>She knew there’s a difference though. She had to make sure that in this new feature, users understood that she was asking for product feedback, not content feedback. She also knew users shouldn’t need to make the distinction themselves. So she decided to extend the existing feature to also serve the new purpose.</p><p>As for the recruiting of the UXR, she added a checkbox right where users provide feedback. That’s what all other products do after all. Since she wanted to use email to communicate with her volunteers, she figured: why not use the existing email subscription mechanism to opt users in and out — It’s intuitive and serves the purpose.</p><p>She brought the plan to the most experienced customer success engineer, and received a bummer input: admins currently have control to turn off sharing content feedback with Simpplr. “How terribly incomprehensive my research was on the existing behavior”, she was frustrated, but also felt lucky that she knew about it early on.</p><blockquote><strong>Lesson 1: Enterprise products usually have multiple personas. In this case, there are at least two personas whose user flow need to be researched on and considered: the end-user and the admin user.</strong></blockquote><p>It was a painful trade-off, but for the sake of minimizing the scope so her feature can squeeze in the next release to align with the timeline of launching the UXR program, she accepted the limitation and left her decisions unchanged.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Jxmg1GuM38cYAcj2bC_Sag.png" /></figure><p>With the decisions finalized, the feature entered the designing phase, and of course, her design partner discovered another bummer: admins can override end-user’s email settings at any time. This means a user could volunteer to participate in her UXR, but be silently opted out by the admin.</p><p>Here’s a quote from Stewart Butterfield, Slack CEO, facing similar situation when Slack launched the “Do not disturb” notification feature:</p><blockquote>We wanted people to be able to send messages freely without having to worry about it because the recipient had control over when messages would appear. We also know that if it’s not the default, then no one’s gonna use it. We also know that a lot of organizations had already built critical workflows off of Slack notifications, so we couldn’t just turn them off without telling them first. We also knew that some management teams were going to want to have more control over this or disable it or set different preferences and we also knew that no matter what they wanted, end users would find a way around it. So what we ended up doing was telling every administrator, we’re going to launch this and it’s going to be default, 8:00 PM to 8:00 AM in your local time zone, or in the user’s local time zone, you can set your own policy here, but by the way, individual users can change it just for them. If you ever decide to change your mind, you can wipe everyone’s preferences and reset it at the team level. So it’s like this whole series of concessions and complexities and stuff like that</blockquote><p>She didn’t get a chance to read about the quote, but she made a decision anyways. She doesn’t allow me to tell you more details, as you must have known, enterprise products should expose their product details as little as possible.</p><blockquote>Lesson 2: Enterprise product need to deal with and balance between the top-down and bottom-up philosophy. It applies to designing the control system of the product, but ultimately it’s a reflection of the company’s govern culture.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iJ8sSlerjeJQrwnFGw-oxg.png" /></figure><p>Engineers started to develop the feature. At this point, the least thing she wanted was to introduce new and large changes that could delay the development cycle. Unfortunately, engineers brought a concern that needed exactly that level of attention: the management of contacts for the UXR volunteers needed to be GDPR-compliant, just like the rest of the product. She should either expect a much lengthened development time and thus a delayed release, or switch to buying and integrating with a third-party tool that’s GDPR-compliant.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*IsxvEy_BwfMvcasFDo-r8Q.png" /></figure><blockquote>Lesson 3: Regulation compliance is a choice in hand for consumer products, but a duty for enterprise products. It is as critical to enterprise products as it is to the companies they serve.</blockquote><blockquote>Lesson 4: Any new mechanism introduced to the product is worth a thought over build vs buy (vs partner), especially for startups.</blockquote><p>Although it is a consideration that’s more or less on the technical side, she knew that it is such a commonly concerned issue in the industry that she should have taken care of. She learned the lesson and carried on.</p><p>The feature eventually launched, after the many ups and downs. But in the world of iteration, that’s never an end but only a beginning.</p><p>She soon started to receive user feedback. As happy as it is, one feedback triggered an alarm: <em>“Today we received a feedback from our employee about the participation check. She checked the box thinking that she was participating in &lt;company name&gt; product research. However she got an automated email from Simpplr product research. If the participation text are clear enough, she would not have clicked the box.”</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/1*dlmSu703C2nZbeOiHE1PfA.png" /></figure><p>That rang a bell in her head: in an earlier phase, she discovered that admins can turn off the sharing on feedback with Simpplr. Behind it was customer’s security concern that employees might provide company sensitive information in the feedback. She understood the concern but never thought one-level deeper on the root cause. Now the dots were connected:</p><blockquote>Lesson 5: Enterprise products face a dilemma with the common white-label practice. While the ability to rebrand is a value proposition to customers, it also prevents the products from talking directly to their end-users. Your design should tailor to where your product stand in this spectrum.</blockquote><p>I’m glad to tell you that she survived her first end-to-end feature iteration and becomes a better B2B product manager. I hope you gained some takeaways as well.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cf13f35c7283" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Onboarding my 1st PM job — Nail that 1st week]]></title>
            <link>https://medium.com/@lindseyliu/onboarding-my-1st-pm-job-nail-that-1st-week-e168dbe0c0b9?source=rss-5ebd88fcb637------2</link>
            <guid isPermaLink="false">https://medium.com/p/e168dbe0c0b9</guid>
            <category><![CDATA[b2b]]></category>
            <category><![CDATA[product-management]]></category>
            <category><![CDATA[onboarding]]></category>
            <dc:creator><![CDATA[Lindsey Liu]]></dc:creator>
            <pubDate>Sat, 25 Apr 2020 21:52:06 GMT</pubDate>
            <atom:updated>2020-04-25T22:26:09.819Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3TBatnV_zBfnXh5MzlcN4g.png" /><figcaption>Source: <a href="https://www.alifeoverseas.com/what-in-the-world-to-do-with-newbies/">alifeoverseas</a></figcaption></figure><h3>Nail that 1st week of my 1st PM job</h3><p>Last September, I landed on my very first Product Manager job at Simpplr. I went through the toughest level of transition: to a new company(from Yahoo), a new role(from Android Developer), a new industry(from B2C to B2B), and a new city, all with just one jump.</p><p>I panicked. I knew how challenging this could be. I wanted to be prepared as much as I could, just so I wouldn’t internally scream on my first week.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/480/1*b0zHnG-28P32qsJxXPTQXA.gif" /><figcaption>Source: <a href="https://giphy.com/gifs/swear-trek-bashir-internal-screaming-MCKyH7WE51leDrqDO4">Giphy</a></figcaption></figure><p>So I prepared a 1st-week agenda to learn things one step at a time. Looking back, I can’t stress enough how much I benefitted from the wise self at the moment. Here’s my list.</p><h3>1st day — Understanding the organization</h3><p>Let me start by saying I’m an introvert. I start panicking just by imagining myself surrounded by strangers. What’s worse, I have a goldfish memory for names and faces.</p><p>But it’s not the end of the world. I just need to put extra effort into it.</p><p>I started by collecting my future colleagues’ faces and names. I was lucky enough to receive company-wide event invites before the official start date, which provided exactly what I needed.</p><p>I then searched for each person on LinkedIn to learn their roles, and made my own “org chart”, like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/580/1*FbJtCuomFVwhkpJ0pKye_w.png" /><figcaption>A portion of my “org chart”</figcaption></figure><p>Looking at the list, questions started popping in my head:</p><ul><li>What does Account Executive mean? What is Customer Success Manager? What about Solutions Consultant?</li><li>How do they work together with product management? What do they need from a PM? What can they provide to a PM?</li></ul><p>As you can see, these questions about the roles are very specific to B2B product. If you are wondering the same, <a href="https://www.amazon.com/Building-Products-Enterprise-Management-Software-dp-1492024783/dp/1492024783">this book</a> has helped me a ton.</p><p>This picture also helped me understand interactions between PM and other departments in general:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6DN-66QqmeMTB_yJ9oR9CA.png" /></figure><p>Memorizing these lovely people’s faces and names before my first day soothed my nerve very effectively. Understanding what they do for the company helped me ask the right questions in the right tones.</p><h3>First few days — Understanding the process</h3><ol><li><strong>Immediate team members</strong></li></ol><p>They are the people I spend hours and hours with everyday. They are my dependencies and my stakeholders, my painkiller and my fuel.</p><p>Other PMs working on different components and platforms of the product, they are my inspirations, my empowerment and my fellows when we evaluate opportunities, align combined efforts and bring momentum to the larger team. We have to always be on top of each other’s work, and we work really hard to make “1+1&gt;2” happen.</p><p>Designers are my partners and mentors who, turns out in the later days, will identify tons of ambiguities and trade-offs while forming interfaces from requirements, and help me grow everyday. If you are lucky enough like me, you could even partner with them to initiate and conduct user researches.</p><p>And of course, engineers and QAs, my dearest friends. Without their architectural mindset and detail-oriented work, what I put on paper everyday will be no more than a blueprint. Just like with designers, we align our beliefs of what is the right thing to build, we block and then unblock each other, and we we make trade-offs together (hopefully only once in a while).</p><p><strong>2. Projects</strong></p><p>Oh dear lord, so many projects to grasp even at a startup like Simpplr. But here are the prioritized questions that I looked for answers during my first week:</p><ul><li>What’s the ownership of everyone? Depending on the role, the answer to this can range from a product component, to a customer segment, to a sales process. Understanding of one’s ownership transfers easily to understanding of one’s goals and dependencies you could come across in your PM work.</li><li>Besides routine work, what’s the most important initiative for each department at the moment? This helps you understand the urgency of their asks, and equally importantly, what makes people excited.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*X_FtFSPUbKa93Zu3UyWdZg.jpeg" /><figcaption>Source: <a href="https://imgflip.com/i/ype8c">imgflip</a></figcaption></figure><p><strong>3. Performance measurements</strong></p><p>How is success measured, in the following levels:</p><ul><li>company level</li><li>department level</li><li>team level</li><li>individual level</li></ul><p>Simple but utterly important. Without a measurable goal, you won’t even know where to invest your effort, and how to prioritize your opportunities.</p><p><strong>4. Tools and meetings</strong></p><p>Simple enough that you could probably onboard in half a day, but worth mentioning, is to setup the tools that your team adopts, and regular meetings that your team all attend to align. For PMs, consider setting up tools that CS and Sales use as well, to give yourself a chance to deepen your understanding of the product from other prospectives, as well as to provide better enablement.</p><h3>By EOW— (Shallow) Understanding of the industry</h3><h4>Inbound — What does your product do?</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/526/1*HpiO6FRQayDNN5L7vxDvNw.jpeg" /><figcaption>Source: <a href="https://imgflip.com/tag/newbie">imgflip</a></figcaption></figure><ol><li>Play with the product. Your fresh eyes on the onboarding experience is irreplaceable. Play your newbie card and ask tons of questions. You want to understand the structure and the functionalities as of today, but don’t forget to also understand the constraints, the legacies, and the historical decisions.</li><li>Look back and look forward on the roadmap. Look at the upcoming releases to be on top of the short-term direction, but also look at recent releases to be on top of the current marketing/sales focus, and the customer training situation.</li></ol><h4>Outbound — Where does your product sit?</h4><p>This is meant to be a large and deep enough learning area that requires continuous investment, but to begin with:</p><ul><li>Externally, <a href="https://www.g2.com/">https://www.g2.com</a> is a B2B PM’s best friend. After 6 months into the intranet industry, I honestly haven’t invested nearly as much as I wished on this.</li><li>Internally, you’ll be surprised by how much learning materials you already have in your hand. On the marketing and sales side, you have company landing page, marketing collaterals , content blogs and social media, and of course sales recordings. Pay attention to the messaging and positioning, but don’t take every word for it especially on the function-level. On the Customer Success side, how is the product initially set up during implementation? What are the best practices your CSMs are promoting? These should be 100% aligned with the value proposition that product and sales push towards.</li></ul><p>Hope my experience and tips help you onboard your first PM job more smoothly :)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=e168dbe0c0b9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Reflect back on my 8-month journey as a side-project PM]]></title>
            <link>https://medium.com/@lindseyliu/reflect-back-on-my-8-month-journey-as-a-side-project-pm-ce5782e1aa6e?source=rss-5ebd88fcb637------2</link>
            <guid isPermaLink="false">https://medium.com/p/ce5782e1aa6e</guid>
            <category><![CDATA[reflections]]></category>
            <category><![CDATA[work]]></category>
            <dc:creator><![CDATA[Lindsey Liu]]></dc:creator>
            <pubDate>Sun, 20 Oct 2019 22:05:02 GMT</pubDate>
            <atom:updated>2019-10-20T22:05:02.585Z</atom:updated>
            <content:encoded><![CDATA[<p>With 3 insightful lessons that fundamentally pivoted my working style.</p><h3>1. Focus on your goal, NOT your process</h3><p>I fell in the trap when I first started on this side-project. I panicked having no idea how to start. Then I recall the process I have observed my team do from an engineer’s limited perspective , and I talked to myself: “oh, so they start with user research, then draft a feature, then draw some mocks, then test it again, then handle it to me. OK, so I guess I need to do user research now. Hmm, what research should I do?”</p><p>I came to my mentor for the side-project— a senior PM on my team, and I asked her: “What’s the best user research method we should use for this project?” She looked confused as she didn’t even know what I was going to research on. She took her time to figure out where my question was coming from, and then comes this lesson: Start with my <em>goal</em>. Ask these questions first: “Where am I? What do I know and not know about the industry, the product space, and its targeted audience? In order for me to move forward, what do I have and what’s missing?” Only until after these questions have been thoroughly thought through, can I then think about how any process can help me fill in the blank and guide my next step.</p><h3>2. Don’t set boundary to yourself already. Be proactive and take ownership.</h3><p>Asian culture has a convention to constrain yourself to behave within a <em>frame</em>, and work hard to be the best <em>within that frame</em>. Intervene with other people’s ownership is discouraged and is treated “irrespecutful and annoying”.</p><p>When it comes to the tech company world, however, it is a dream culture for everyone on the team to take ownership of the product, be proactive to raise any questions even on other functions, and be ready to step outside your responsibility and do whatever needs to be done.</p><p>Interestingly, there’s a subtle decision-making for when is “helping too much”. Help is treated with appreciation when you fully understand the expectation and your teammate in need is comfortable with your help on that specific task. When it’s too much and not transparent enough, though, your works can overlap, and you need to evaluate whether your time is spent on the most valuable things — work in your own area of expertise.</p><h3>3. Take a deep breath before responding to any question/comment against your opinion.</h3><p>When someone said something against your opinion, take a deep breath, don’t put the two of you on the two faces of a coin by default. Start with listening. Listen to them and really <em>consume</em> it.</p><p>After you truly understand the thoughts and concerns behind their words, and before you jump to a conclusion of “who is right”:</p><ul><li>Does your opinion really conflict with hers? Try finding the best of both worlds. You might need to collect further facts to prove yourself, or you might start rethinking about your own assumptions.</li><li>Who is the actual stakeholder you need to persuade behind this disagreement? Maybe this person is conveying a concern he thinks that stakeholder might have. Schedule a meeting and discuss with that person, not this one.</li></ul><p>I get it. It’s second nature to respond to questions immediately. Silence is awkward, and make your audience feel you’re under-prepared. But people’s perception of your confidence is far less important than trust, and trust is earned with facts , with delivery of success, <strong>not</strong> with a fast and superficial answer to their questions.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ce5782e1aa6e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[I ask myself this questions to help me prepare for meetings across function]]></title>
            <link>https://medium.com/@lindseyliu/i-ask-myself-this-questions-to-help-me-prepare-for-meetings-across-functi-2cb8670fdaa8?source=rss-5ebd88fcb637------2</link>
            <guid isPermaLink="false">https://medium.com/p/2cb8670fdaa8</guid>
            <category><![CDATA[meetings]]></category>
            <dc:creator><![CDATA[Lindsey Liu]]></dc:creator>
            <pubDate>Fri, 28 Jun 2019 18:40:17 GMT</pubDate>
            <atom:updated>2019-06-28T19:01:11.623Z</atom:updated>
            <content:encoded><![CDATA[<h3>I ask myself this question to help me prepare for meetings across functions</h3><p><em>To hopefully save everyone’s time (so they can take more meetings ;)</em></p><blockquote>Who are the attendees?</blockquote><h3>Different roles</h3><p>Depending on what roles the attendees each hold for the project, they will have different contexts and expectations about the meeting.</p><p>In terms of context, they may or may not be aligned about the knowns and unknowns, the problems and existing solutions. Sometimes there are attendees who don’t even know why he’s here. I found it very important to make sure everyone is on the same page before the meeting starts. Ideally this should already be the case if there are sufficient daily communication. But if not, I try to do that even before the meeting starts, by stating clearly the goal and all relevant context, in the calendar invite. This hopefully encourages people to prepare for the meeting themselves as well.</p><p>In terms of expectation, I’ll be sure to understand each attendee’s expected output and what questions and doubts they might come up with in the meeting because of that. As a PM, you make decisions, and people ask why. Of course there’s no way I can answer every single question from all people. In fact, when there is one I can’t answer, it’s a good sign to dig deeper and give it more thoughts. But the best way to prepare is to ask yourself beforehand. I’ve found myself much more confident for a meeting (especially a large one), if before the meeting I ask myself the assumed questions really hard and make sure at least I’m persuaded by my own answer.</p><h3>Different person</h3><p>Each attendee might have different meeting personalities. Just like personas for a product, meeting attendees may want to feel in control and take the lead, or may want to sit in the corner and raise hands only when they have questions. Some may encourage more spontaneous discussion, whereas others may stare at their watch and constantly alarm people to stay on track. I think it’s not necessary for a PM to have to play the role of leading a meeting if there’s a better volunteer. The actual necessary responsibility is to define the goal, encourage the information exchange, and make sure the dynamics of the attendees are healthily contributing to pushing a solid and agreed-upon outcome of the meeting.</p><blockquote><a href="https://www.thereallindseyliu.com/">www.TheRealLindseyLiu.com</a></blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2cb8670fdaa8" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>