On a Tuesday in March, Sophie at ONAH told me she wanted to move quickly but had no idea what came next. Her HR product looked promising but there was no finished website nor clear guidelines. By Wednesday afternoon, she had a positioning statement, two sticky one-liners (“the inbox no longer sets the agenda”, “stop answering emails, start doing HR work”), and homepage copy good enough to ship. She was surprised it was AI-written.
A few good signals were enough.
If you’re a first-time technical founder, three to six months past pre-seed, with a few customers from your network, this post is for you. The bar for writing a sharp startup positioning statement is lower than you’ve been told. The cost of waiting is higher.
”I don’t have enough customers” is the wrong frame
Nowhere in Obviously Awesome does April Dunford say you need thirty customers. The book tells you to list your best customers. For most early-stage founders that list is short.
The real problem is what people do when their best customers disagree. They panic, decide the sample is too small, go off to talk to twenty more. Two months. Positioning pushed to “after the next round.”
Michele Hansen makes the load-bearing point in Deploy Empathy: “If you find yourself with five interviews under your belt and you’re hearing wildly different things from each person, that’s a sign your problem definition scope could be narrowed down.”
Read that twice. Divergence at a small sample-size isn’t always a sample size problem, but a scope problem. Your positioning is too broad and three buyers are pulling it three directions. The fix isn’t more interviews but a clearer definition of who you should talk to.
The waterfall: ICP, positioning, USPs, messaging
Here’s the frame I keep coming back to, and it’s the most important idea in this post.
Data from your ICP waterfalls into your positioning. Positioning into USPs. USPs into messaging. Messaging into homepage copy, cold emails, sales scripts, LinkedIn posts, ads, and so on.
The waterfall is a system I’ve been building for four years out of product marketing notes, templates, and frameworks. For Sophie I bundled it into a tool that takes raw onboarding input and runs it down the chain in one pass: ICP → positioning → USPs → messaging → copy. The first end-to-end run was on her project. The next run will be faster. The point isn’t the tool, though, it’s the chain. Any founder can run the same chain manually with three customer interviews and a day of focused work.
Worth being explicit about one thing: customer interviews and the waterfall are different activities. Interviews gather raw input. The waterfall transforms that input into usable statements and copy. People conflate them and get stuck on the input phase, hoping more interviews will eventually produce a position. They won’t. The transformation step is its own work.
Most founders try to write the bottom first. They sit down for a homepage and discover they don’t know who it’s for. They draft cold emails and every line could be sent by any competitor. They blame the copy. The copy is fine. The waterfall is dry.
The waterfall works because the bottom forces coherence at the top. You can’t write a hero without a value statement. You can’t write value without a position. You can’t write a position without an ICP. So when you sit down to ship, you discover what’s missing upstream, and the work fixes itself in order.
This is why “we’ll do positioning later” is a trap. Without it, every downstream artefact gets stuck or generic. With it, even a half-right hypothesis gives you something to test. Testing is the only thing that produces better data than three customers.
What to actually extract from three customers
Each of your three customers contains more signal than a hundred survey responses. Ask the right questions and shut up to hear the answers.
Adapt the Switch Interview from Bob Moesta’s Jobs to be Done framework. Four things to hunt for:
- The struggling moment. When did they first realize their old way wasn’t working?
- The desired outcome. Beyond features, what did they want their world to look like?
- The anxieties. What almost stopped them from buying?
- The competing habits. Spreadsheets, three tools, an intern, nothing?
Add Dunford’s two. Magic wand: if you could change anything about how you currently do this, what would it be? Dinner party: how would you describe what we do to a friend over dinner? The first surfaces the unmet need. The second surfaces your real category, in their words.
Tip: write their phrasing down verbatim. Customers give you better copy than you’ll ever write.
Not all three customers are equal
Most founders average their three customers and call the average a position. Bad idea. Two are probably mildly interested. One is in pain. That last one is the data point.
Matt Lerner at SYSTM calls this finding the suffering segment. Calm, PayPal, Deel, Vanta, and Gong didn’t change their products: they found customers whose pain was acute, urgent, and frequent enough to overcome inertia.
So look at your three and ask: which came complaining about something specific? Which had a deadline? Which had already tried two other tools? That’s your wedge.
Frederic Kerrest of Okta puts it bluntly in Zero to IPO: “Smart startups begin by targeting just one or two sectors, focusing on the top three or five companies. These are called lighthouse accounts. If you can win them, you can often own the whole vertical.”
Three isn’t too few. Three is the right number for a wedge. Lighthouses, not a sample.
”But what if I pick the wrong three?”
The most common objection I hear, almost word for word: “We don’t actually know which of the three is the real signal. What if the next ten look completely different?”
Fair argument. Also not actionable. At some point you stop throwing spaghetti at the wall, focus on the three you have, find what overlaps, and commit.
Two paths follow. Path one: focus until it works, prove the wedge, expand to adjacent industries with data. Path two: focus until it doesn’t work, then pivot. The second isn’t a loss. You uncovered an industry that doesn’t work for your product, which is information you wouldn’t have without the focus.
The version that loses is “let’s keep our options open and reach broadly.” That’s not optionality. That’s the absence of a thesis.
Filling Dunford’s five components at n=3
Here’s the structural payoff. Dunford’s five-plus-one components of effective positioning aren’t statistical questions. They’re qualitative pattern questions. Three good customers are enough to answer all of them, today.
1. Competitive alternatives. What would your customers use if you didn’t exist? Spreadsheets, three Frankensteined SaaS tools, an agency, an intern, or nothing. The honest answer is rarely your obvious-looking competitor.
2. Unique attributes. What can you do that the alternatives can’t? Not what you wish you did. What you actually do today that nothing else on their list does. Three customers give you two or three real ones, not the twelve in your pitch deck.
3. Value (and proof). What do those attributes enable? Tie every attribute to an outcome they mentioned. If you can’t, it’s a feature, not a benefit. Proof comes from their own words: “We used to spend six hours a week on this. Now it’s twenty minutes.”
4. Target market characteristics. Describe the customer who values your attributes most. Not all B2B tech companies. Not all HR teams. The specific shape of company, role, stage, and trigger that turns interest into urgency.
5. Market category. What frame of reference do you want to be evaluated in? Dunford spends a chapter on this and founders still avoid it. ZYGO spent weeks debating internal terminology. “Operating system” was in the mix and got rejected: ZYGO is more than an OS, more than software. “Robotics software platform” was the only term that covered the full scope. Three words. Ended the debate. Unblocked the website, the PR, the partnerships.
Plus one: relevant trends. What macro shift makes your category urgent? AI, regulation, remote work, cost pressure, talent shortage. Pick one. Seasoning, not the meal.
You can fill all six in a focused half-day if you’ve done the interviews. Sophie and I did it in one Wednesday.
The deliverable: a sharp hypothesis, not a final answer
Your output is a positioning statement, written down, in one paragraph. The goal isn’t to be right forever. It’s to be specific enough to test.
A useful template, from Dunford:
For [target customer] who [problem], we are a [market category] that [unique value]. Unlike [competitive alternative], we [key differentiator].
Userlist’s repositioning case study is one of the cleanest worked examples I’ve seen. They moved from “email automation for SaaS” (a feature) to “customer messaging for SaaS” (a category) with the differentiator “more efficient than building it yourself, less complex than Intercom or Customer.io.” That last sentence is the whole game. Not the best, the right size for a specific kind of company. Dunford calls that move Big Fish, Small Pond.
ZYGO’s “robotics software platform” works the same way. Three words that tell a specific buyer what bucket you fit.
Your first version will feel uncomfortably narrow. Good. As Arielle Jackson writes in First Round Review, “If you have a compelling product that people actually need, you should be able to write a decent positioning statement. If you’re struggling, that’s a sign.”
Expect two of your three to nod and one to push back. The one pushing back is either outside your ICP or pointing at the part you haven’t sharpened. Don’t flatten to make all three comfortable. Flat positioning is no positioning.
Testing it without burning the place down
Sharp positioning is a hypothesis. Test it where the cost of being wrong is low and feedback is fast.
Three places to deploy first:
- Homepage hero. Headline, subhead, one-line value statement. Watch scroll depth, time on page, demo-request rate.
- Cold outreach. Rewrite the top-of-funnel email. Category descriptor in the subject line. Unique attribute in line one.
- Discovery calls. Open with the positioning statement. Watch faces.
Signs it’s working: calls get shorter and more qualified. Cold reply rate goes up. Objections get more specific (“we already use X for that,” not “what does this do?”). Wrong-fit prospects self-reject.
Peep Laja built Wynter to a reported $2M ARR in two years on the premise that product-market fit without message-market fit is dead. Position now, refine messaging continuously. The position is the stake. Messaging is what you sand against real reactions.
The cost of waiting
Let’s get specific about what “we’ll do positioning later” actually costs.
Cold email benchmarks for 2025 sit at 3.8 to 4.0 percent reply rates, around 306 emails per B2B lead. With sharp positioning, those numbers improve. With generic positioning, you compete on volume and someone with a bigger SDR team wins. So you send more, get fewer replies, blame the channel, try a new one. Channel sprawl is what positioning failure looks like in your calendar. (I wrote about that in the do-everything trap. Sophie features there too, on the channel side of the same story.)
Stack that against pre-seed burn of $20-50k a month and 18-24 months of runway. Three months of generic positioning is 60 to 150 grand. Three months you also can’t test claims or generate data.
What changes at n=10
You will rewrite this. That’s not a flaw. That’s the design.
Todd Jackson at First Round Capital puts the threshold cleanly: “Find three to five customers with a problem worth solving. Gong felt PMF when 11 of 12 design partners chose to buy.” Three is where you commit. Ten is where you confirm. Thirty is where you specialize.
At n=10, you’ll see which patterns held and which were artifacts of your first three. The position gets edited, not replaced, and the edits are cheap because your downstream artefacts show you exactly what to update.
Once you’ve shipped the first version, audit it: pull the ICP doc, the homepage, the cold email, the sales script, check they tell the same story. (I wrote a step-by-step audit method you can run in an afternoon.) Then translate the position into the messaging framework that feeds the rest of GTM. And if the ICP step still feels mushy, do that first. The waterfall: the top fixes the bottom.
The point of writing it at three isn’t that you’ll be right. It’s that you’ll be specific. Specific is testable. Generic is just expensive.
Pick the wedge, write the paragraph, ship the homepage. The next thirty customers come from the position you commit to today, not the cleaner version you’ll write next quarter.