Last November, after taking a strategy workshop with YO from Fresh Milk Workshop, a Taiwanese dairy startup, I wrote a very long reflection. I had just realized that I’d spent all my time heads-down building features without ever thinking clearly about why we were building them. So after the workshop, for the first time, I wrote our vision down in full: Rebuilding how humans absorb, connect, and evolve knowledge. Once it was written, I felt much more grounded. I even believed that as long as I looked up and saw our North Star ten years out clearly, I’d be able to judge any new opportunity or suggestion on the spot.
Then, from the end of last year into this year, we went through months of fumbling around and running into walls, and step by step we narrowed the product from a knowledge platform down to what we build now, a memory system called Cairn. Looking back, the vision I wrote last year really did hold us up. It kept the team from losing its way while the product changed direction. But once we actually started building Cairn, I found myself stuck in a different kind of confusion.
As a founder you often hear two kinds of advice. Some people remind you to think hard about what your company will look like in ten years. Others tell you that what matters most right now is focus: serve a small group of people well first. Each makes perfect sense on its own, but when I carried that big vision back into my daily work, it was surprisingly hard to make decisions with it. We want the understanding people build up while working with AI to carry forward and be reused. That direction fits the vision perfectly, but it stretches into so many situations. Every time I thought of another group who might benefit, that direction seemed worth pursuing too. The product wasn’t even finished, and in my head the scope of who we served had already expanded several times over 😅
Of course I know time and resources are limited. But knowing that wasn’t enough to help me make trade-offs, because the directions we set aside might genuinely be valuable, and I had a hard time proving they weren’t worth the investment. Then recently I took Cindy’s course on early-user needs discovery and PMF, read the books she recommended, Diffusion of Innovations and Crossing the Chasm, and then took a Lean Canvas course from Octopus Bro (his nickname) at Xiaoshuwu, a Taiwanese chain of rentable meeting and study rooms. Suddenly the problem I’d been stuck on for months all clicked together. If last year’s strategy workshop taught me to look up at the vision in the distance, these recent courses and books finally showed me how to take the next step under my feet. I hadn’t felt a breakthrough like that in a long time, and honestly it’s pretty exciting.
Change the reference point, and even your competitors change
The first thing in this round of learning that woke me up was when Cindy talked in class about the “reference point” we choose when we look at things.
I used to feel like almost anyone could be my competitor. From the memory-system angle I’d see note-taking and knowledge-management tools, and if I followed our vision of “absorbing, connecting and evolving knowledge” further up, even consultants and content services seemed related. The wider I looked, the more opportunities I saw, and the harder it got to decide who we should actually compare ourselves to right now.
Cindy talked about Jobs to Be Done in class, and it snapped me back to reality. Whether someone picks your product has little to do with product categories or user labels. What matters is what trouble they’re stuck in right now and what they’re trying to move forward. That made me realize that while I can imagine a very wide range of competitors, I can’t keep expanding it based on my own imagination. Even if two tools both help with learning in the abstract, that doesn’t mean a user facing a problem would actually weigh them against each other. Real competition still comes down to what that person is trying to get done right now, and what options they already have at hand.
I also went back to the problem I’d written during a class exercise: “How do we get users to treat us as their own memory bank?” Only after writing it did I notice that the sentence had already stuffed our product’s form into the question. Once you ask the question that way, you naturally keep thinking about how to get people to store more and how to lower the barrier to using it. Why they absolutely need to keep that content in the moment, I hadn’t asked nearly as clearly.
But pulling the question up a level has its own risk. If rephrasing things lets me find a bigger market for what I already wanted to build, how do I know whether I’m getting closer to a real need, or just getting better at explaining my own product? So reference points took on another meaning for me. They shape which problems I see and which constraints I ignore. Being able to look from a different angle matters, but after switching I still have to check whether that angle maps to the choices users actually make.
Why a ten-year vision didn’t tell me what to do now
Once I’d figured out reference points, I finally understood why I felt so grounded after writing the vision last year, yet this year I was still being torn back and forth between “think further ahead” and “focus on what’s in front of you.”
When I first set the vision last year, part of me expected that if the long-term direction was clear enough, the priorities right in front of me would emerge on their own. Actually living through these past months, I found that a big enough vision has room for many directions by design. When every option looks consistent with the vision, the vision alone can’t rank them. It reminded me of what Richard Rumelt says in Good Strategy Bad Strategy: goals and strategy are two different things. A vision tells us where we want to end up in the long run, but strategy means first recognizing which wall is blocking us most right now, then concentrating our limited effort there.
Advice I used to think was contradictory was really answering questions at different levels. Talking about the ten-year vision is about confirming what’s worth committing to for the long haul, something that won’t waver even as the product changes direction. Talking about who to serve first is about which people we can actually reach, and whether we have a chance to understand one specific pain point deeply enough. There are many layers of judgment between those two, and you really can’t cover them all with one line about “what we want to become.”
My view of milestones changed along with this. I used to see milestones mostly as a list of things we planned to finish. Now I’ve started asking what we should know at the end of each stretch that we didn’t know before, so that we have good enough reasons to plan the next one. The entry point I pick now doesn’t have to look exactly like where we end up in ten years, but I have to be able to say what evidence this first step will build up for the decisions that come after. Thinking this way, I can also let go more calmly of directions that are genuinely valuable but that we don’t yet have enough reason to put resources into.
Why one group’s approval doesn’t carry over to the next
Reading Diffusion of Innovations and Crossing the Chasm with these thoughts about stages in mind, I noticed that the way people adopt new things has more layers than I’d thought.
I’m the kind of person who sees a new AI tool and wants to grab it and play with it right away, and when I’m talking with people about the product, I easily feel a kinship with those who like trying new things. But reading Rogers made me realize that “early adopter” isn’t a fixed personality label. It’s about how early someone starts using something new within a particular circle. My familiarity with this group actually made me underestimate how much ordinary people need to confirm before they decide. Switching to a new tool is a risk in itself, and before deciding whether to use something, most people really need to see others in a situation like their own already using it, with no problems, before they dare to follow.
Reading Crossing the Chasm made that gap feel even sharper. The book puts it bluntly: the people willing to play along with you early on might buy in just from hearing you talk about the vision and the picture of the future. But the majority who come later and want to use it for real work are very pragmatic. They need to see that the use case is already mature, with successful examples around them they can refer to, before they’ll even consider adopting it.
I tried putting this back into a work situation we all know. Say someone spends a weekend redoing an old report with a new tool. Just seeing it do things they couldn’t do before is amazing, and even if the formatting breaks, they’re happy to spend time fixing it. But someone who has to hand a final report to a client first thing tomorrow morning will definitely check whether the numbers can be verified, and whether they can recover right away if something goes wrong. Even the same person might make completely different choices playing around on the weekend versus delivering on Monday.
That suddenly made it clear to me why the approval a product gets early on often doesn’t extend outward. Flaws some people are willing to tolerate may be exactly the risks another group can’t afford. When your first group of friends says your thing is great to use, their forgiving use cases and their enthusiasm for tinkering are wrapped up in that verdict too. If I happily accept “great to use” and strip away all the assumptions behind it, I can easily overestimate how far that experience applies. This is where I paused longest while reading Crossing the Chasm. When the market needs to expand, I have to work out again why the next group makes its decisions. I can’t expect them to simply be people who heard about it later and will eventually accept the same pitch.
It turns out there are so many differences inside a segment
All the concepts I’d picked up from the books only really sparked in my head once I took Octopus Bro’s Lean Canvas course.
What stuck with me most from the whole class was that no matter what we discussed, he kept pulling us back to a concrete “one person”: ideally someone you can name, know where to find, and whose actual experiences you can write down. One example from class: when Xiaoshuwu serves freelancers, the thing users compare it to might be a business center, but when the person walking in is a student studying for exams, Louisa Coffee, a popular Taiwanese café chain, might enter the comparison. It’s the same act of looking for a space, but change the situation and market boundaries that seemed obvious shift with it.
That example instantly connected the reference points from Cindy’s class with the market segmentation I’d read about in the books a few days earlier. Segmenting a market isn’t sitting at a computer drawing demographic boxes. It’s knowing which people make choices using similar criteria, and what life or work situations those criteria grow out of.
And even once you’ve chosen a group, their current relationship with the product can be very different. People who haven’t heard of you first need a chance to get to know you. People who have used it but didn’t come back need us to go back and understand what actually happened that time. If you lump both situations together as “we need more users,” the actions that follow easily lose their aim. This is yet another way of slicing things than the early-versus-late adoption in the books. Someone very willing to try new things might still try once and leave, and someone who was cautious might keep using it for good once a pain point is truly solved.
When Octopus Bro talked about core users at the current stage, he specifically reminded us to think about “who can bring in the people who are truly a fit.” That made me rethink what it means for someone to be willing to share. Beyond the act of sharing itself, I need to know whose words that circle treats as experience worth listening to, because the first people to rush in and try something new aren’t necessarily the most persuasive people in the group.
Putting all of this together and going back to the beachhead market in Crossing the Chasm, I finally understood completely why it’s not just about shrinking your target numbers. You have to concentrate fire on fully solving one important scenario, build a complete solution that lets people actually finish their work, and build up real cases the next person can see and trust. With Cairn we’re still confirming our early core use case, which is different from the book’s situation of crossing into the mainstream market. But this way of thinking has already got me asking how much reference value the results of the group we pick now will actually have for the next group. When I realized that theory I thought I’d understood from the books a few days earlier could finally be used to take apart the choices in front of me, that excitement of a breakthrough was very real 🙌
Focus also changes how I understand feedback
After putting the threads from these courses and books together, I came to a new understanding of “focus” that I’d never thought about before.
I used to think of focus mainly as allocating limited time and development resources. Now I’ve started to feel that choosing who to serve first also directly affects whether I can build real understanding from user feedback. Each person’s input makes sense on its own, but if what they’re trying to get done differs a lot, combining all of it just gives you a to-do list that keeps getting longer. This person wants a simpler interface, that person wants more control. After accommodating both, there are definitely more features, but I’m not necessarily any clearer on which person’s which problem actually got better.
To use an analogy from data analysis, which I know better, it’s like dumping data from different experimental conditions into one pool and rushing to compute an average result. The differences that really matter often get washed out in the merging.
So now I also see focus as a way to let observations build on each other. If several user interviews in a row revolve around very similar work situations, the next person’s experience has a better chance of adding to or contradicting what the previous person said. And when we make a change to the product, it’s clearer which metric to go back and check to see whether it worked.
Of course, this is just where my reasoning has taken me so far, and I’m still figuring it out myself. If I segment too early or too rigidly, I might end up only hearing what I want to hear, and every time I hit a counterexample, comfort myself that the person just isn’t a fit for our product 🫠 After choosing a group, I still have to stay alert to whether that so-called shared problem is just something I made up in my head, because some people look alike on the surface but care about completely different things when they actually pay or make a choice.
Last year’s workshop taught me to look up and see the vision clearly, and it helped us hold on to our North Star while we were feeling our way through the product’s change of direction. These past few weeks of SLP courses and books gave me a different kind of groundedness, the feeling of standing on solid floor. I haven’t found all the answers yet, but I’m starting to be able to tell, for each step under my feet, which parts rest on real evidence and which only sound reasonable. The further I go, the more I feel there’s still so much to learn. I hope to keep improving like this, one step at a time, and I’d love to hear from anyone who has walked this road before 🙏