Most startup positioning problems look like writing problems. The homepage is vague, so someone rewrites the homepage, and three weeks later it’s vague again — because the words were never the problem. The company hadn’t decided what it was, for whom, against what. Positioning is that decision. The copy is just the decision, published.
The method below is the evidence-first approach April Dunford made canonical in B2B, folded into how we run it inside the audit. It has one governing rule: positioning is discovered from your best customers, not invented in a workshop.
Start from the customers who already chose you
Take your five to ten best customers — best meaning they bought quickly, use the product heavily, and would be upset to lose it — and mine the evidence they’ve already given you. Why did they buy, in their words? What were they doing about the problem before? What almost stopped them? Win-loss notes, onboarding calls, and support threads hold most of this; two or three short calls fill the gaps. You’re listening for the pattern in who they are (their situation and pains — not their industry and headcount) and what you actually replaced.
That second one is where startups flinch. The honest alternative is rarely the competitor in your pitch deck — it’s a spreadsheet, an agency, an intern, or doing nothing and hoping. Position against the real alternative and the differentiation writes itself. Position against a competitor your buyers haven’t heard of, and you’re explaining two products instead of one.
Write it as four answers, not a template
Skip the madlib. Positioning templates produce sentences no human says out loud, and the fill-in-the-blanks format lets you complete the exercise without making the decision. Instead, write four answers in the plainest language you can:
Who is this for?
Described by situation and pain, not firmographics. “Founders doing the marketing themselves before the first hire” beats “B2B SaaS companies, 10–50 employees.” An ICP written in firmographics is one of the surest signs the customer-evidence step got skipped.
What would they use instead?
The honest list, including “nothing.” This is the question that makes positioning real — every claim of differentiation is relative to this answer.
What can you do that the alternative can’t?
Capabilities, then the value they unlock for this specific buyer. If a competitor could paste your homepage over theirs and nothing would look wrong, this answer isn’t done.
Why should they believe you?
Proof sized to the claim: a quantified case study, a live demo, a free first step that shows rather than tells. Unproven claims read as noise to a buyer who’s heard everything.
Those four answers are your positioning. The homepage H1, the sales deck, the LinkedIn bio — all of it should be derivable from them, and none of it should require a thesaurus. If the answers are right, plain words carry them.
Test it: one leading signal, one lagging number
The 10-second test. Show the homepage to a stranger in your target market for 10 seconds, then ask: what is it, who’s it for, why is it different? Nielsen Norman Group’s page-abandonment research found the first 10 seconds decide whether a visitor stays at all — so this is the test your positioning takes a thousand times a month whether you run it or not. Run it on your three closest competitors’ homepages too; the gaps you find are positions nobody is holding.
Percent of pipeline inside ICP. The lagging number that keeps everyone honest. Positioning aimed at one buyer that fills the pipeline with another has been overruled by the market — a common and quiet failure at seed, where heavy word-of-mouth reliance means the pipeline reflects the founder’s network, not the positioning. Listen live, too: when a prospect says “so you’re basically like X,” that miscategorization is free positioning research.
Both tests are built into the marketing audit — it walks your competitors’ homepages, runs the 10-second test on yours, and asks for the pipeline-in-ICP number directly. And once the positioning decision is made, it cascades into everything downstream: which channels fit, what the content says, who the outbound targets. Kindling’s free audit is the ten-minute version of this whole guide — it reads your site the way the stranger would, and tells you plainly what it saw.