Poldermarketing
What should a seed-stage startup say on its homepage to attract the right buyers?
Name the buyer, the change, and the workaround they can drop. On the first screen, put that headline, one line for how it works, and one button. Put proof a stranger can check in the next strip. Keep the round and the vision off that screen. Until buyers pull you toward a segment, treat the line as a thesis.
Say the change, the buyer, and the workaround
A seed homepage should tell one buyer what changes for them now, and which current workaround they can drop. The round and the long-term vision stay off that first screen.
April Dunford separates customer positioning from investor positioning. Customer positioning is the value the buyer gets today, against alternatives they already have. Investor positioning is the later vision of a large company. Serving both in one pass produces a muddy story. On a single-product company, the product and the company are the same story.
The first screen is for the champion, the person who builds the shortlist. If that person does not recognize the job, you never meet the other stakeholders. Define the ICP before the hero.
Climb from the alternative to the category
Build the sentence from what the buyer would do if you did not exist.
Dunford’s quickstart defines positioning as how your product leads at something a defined set of customers cares about a lot. Work in this order, because each step depends on the one before it: the alternative they would use, the capability those alternatives lack, the value of that capability, the customers who care a lot about that value, and the category that makes the value obvious. A fill-in-the-blank positioning statement is pointless and dangerous, because most products can sit in more than one category. A coined name, her “fluflommer,” has to be taught before anyone can want it, so it does not lead. When buyers know the category, narrow it, as in “embeddable database for mobile devices.”
Mucker adds the fork the headline has to follow. If buyers already know the category, put that name in the headline and then say how you are better inside it. If they do not know the name, do not lead with it. Lead with the job and the workaround.
On competition, beat the status quo first: a spreadsheet, a document, a manual process, an intern, or the limited tools inside a platform they already use. The quickstart gives one typical share of deals lost to no decision, and the competition note gives a range that contains that share and cites a third party. The notes do not conflict. Do not borrow either figure. A rival they have never shortlisted is a phantom.
Stackmatix puts the order in one sentence: we help this buyer reach this outcome without the old painful way, by this one mechanism. Take it from 10 conversations. Listen for “the hard part is.” If a founder, a builder, and a design partner cannot repeat the same sentence, it has not landed.
Keep the net wide until buyers pull
Before customers arrive, the public line is a thesis. Tighten who it excludes only after buyers show you the pull.
Dunford writes that across 16 in-house launches, positioning was never exact before launch. If you say the product is only for insurance companies, only insurance companies try it, and a miss teaches you only that insurance was wrong. Keep the public line a little loose, then tighten where demand appears. Publish an exclusion only after you have seen the mismatch, or when those buyers would break the product.
Once it is public, hold it on two clocks. Stackmatix says repeat it verbatim for 30 days, and judge it only after at least a full quarter of outreach. Do not rewrite because one customer arrived. Rewrite when you keep losing to one named alternative, or buyers file you in the wrong category. When you change it, change the site, the deck, and the posts at once. Change the outbound sentence in that same pass. That sentence is what positioning into campaigns carries, and what you validate before outreach.
Three lines on the first screen
On the first screen, use three lines: a headline that names the buyer and the change, one supporting line for the mechanism, and one button. Stackmatix on the seed site puts proof, the problem, and how it works below that hero, and ends with a final prompt. Put checkable proof in the next strip. An exclusion you have earned goes below this screen, not on it. Repeat the same button at the bottom.
Nielsen Norman Group says a concise statement of who you are and what you do stops people guessing whether they are in the right place. Two buttons of equal weight make them stop and choose between the buttons, which costs time and feels like strain. Give one button the weight.
Separate the clocks. That article’s lab flash is 50 milliseconds for visual appeal and up to five seconds for what looks prominent. Nielsen, citing Lindgaard, says that flash is not how a live page is used. On a real visit, test the message at about 10 seconds, the time people give a page before they leave.
| How you sell this month | Button | A weak default |
|---|---|---|
| The buyer can try it | Start, or open the docs | Book a demo |
| A person has to show it | A short conversation | A trial that does not exist |
| They need to see the job happen | The product on a real example | Learn more |
| You cannot take customers yet | A waitlist that says what arrives | A checkout |
Hedrick is right that “Learn more” does not work as the button, because it adds friction, and lists start free, book a demo, or read the docs. The same guide also wants the seed site to reassure Series A investors and early enterprise buyers. Keep that job off the hero. If the founder is still the demo, the button is a conversation: founder-led GTM before a full team.
Below the fold, use the buyer’s steps: the situation, what they do with the product, and what is true afterward. Discover, build, measure is the vendor’s plan. Show a price only when the buyer can choose a plan alone.
Show the job in the buyer’s words
Show the product doing the change, in the strip under those three lines, in phrases from calls.
Stock imagery suggests there is nothing to show. Logos that are not real customers break trust faster than an empty strip, and Hedrick finds three named customers stronger than a longer wall of weak ones. Use the real interface or the output. Content that supports outbound repeats the homepage sentence. The first go-to-market strategy picks the channel.
Use only proof a stranger can check
At seed, proof is whatever a stranger can verify. Rank it, and leave the slot empty rather than borrow a number.
- A named buyer, the situation, and the change, with permission.
- The product doing that change on a real example.
- A design partner, labeled as a design partner. A pilot logo sold as a customer is a false claim.
- A founder credential that touches this job.
- Nothing. Say you are early.
Skip an hours-saved figure or a usage count you have not measured. Add a number only when the buyer would confirm it.
A vague line survives a name swap
Messaging is too vague when a stranger cannot say who it is for, what changes, or why this product beats the workaround. The first words should name the buyer. On the seed site, buyers self-identify in the first five words, and empower or streamline fails.
- Paste the closest alternative’s name over yours. If the sentence still works, you named a category.
- Three people who must say it in public come back with three versions.
- A call transcript shares none of the words.
- The first words do not name a buyer.
- The hero says platform, operating system, AI-powered, future of, modern teams, scalable, robust, or mission-critical. Hedrick flags those last three.
Put the hard word in parentheses
A complex product stays clear when the first line is the buyer’s outcome, and a technical term appears only if that reader needs the exact word.
Kate Moran asks two questions: how many readers know the term, and whether they need that exact word. If most readers know the term, use it with no explanation. Explaining a basic term to experts signals the wrong reader. If few know it and they do not need the exact word, use plain language. If they need the word and most readers are new, put the plain words first and the term in parentheses. If most already know it but some do not, put the term first and the plain words in parentheses. Parentheses mark the extra as skippable, because people scan.
Spell each acronym out at the start of a major section, because people do not start at the top.
Hold the page in four situations
Keep the homepage as a short thesis, and do not publish the exclusions, in four situations.
- You have not had the 10 conversations yet.
- The page is there to close the round. The deck carries the future.
- Two products exist and you have not chosen which one a new buyer meets. With one product, keep one story.
- The headline names a change the product cannot do yet. Label it as later, or leave it off.
Agree the sentence and the one button from the table above, then build the other pages so they repeat that sentence.
I would draft the first screen from the calls you already have
I would read the calls you already have, write one sentence in the champion’s words, and put that sentence on the first screen with the product doing the job and one next step.
I am Piet Baudoin. Poldermarketing is a one-person eenmanszaak. I work remote, in Dutch and English, freelance for a bounded project or a few days a week. I am AI-native, strong in AI, content, automation, and workflows, and I build and run the work. Google Ads and Meta Ads are newer for me: I set them up and review them, and I do not scale a large ads program.
For a freshly funded startup anywhere, I work as one person across the message, the marketing, and the automation, from the first sentence to the first customers, without a separate specialist for each part. Accounts and prompts stay in your name.
If the current site cannot say who it is for, the free growth scan shows how it reads now. Positioning and messaging is where that sentence gets made. How I work is the shape of an engagement.
Questions people ask
How do I know if my startup messaging is too vague?
Paste a rival's name over yours. If the headline still works, it names a category, not a position. The line is also vague when three colleagues repeat three versions, when words from real calls never appear, or when the hero is future-of-work language. Specificity checks vagueness. A sharp line can still be a guess, so do not wall off other buyers until they are the ones who show up.
How do I explain a complex product without technical jargon?
Lead with the change in the buyer's own verb. Add a technical term only when this reader needs that exact word, and put the plain version beside it in parentheses. Experts can distrust a page that explains a basic term, while newcomers stall on an acronym. Spell acronyms out in each major section, because people do not read from the top. Architecture and security belong one click down.
Should a seed homepage mention the funding round?
Keep the round off the first screen. A buyer decides on the change they can get now, against the workaround they already use. The round is a story about the future, and mixing that story with today's value produces a muddy hero. Put the future in the deck. A short about note can say the company is early without turning the homepage into a pitch for the round.
What if there are no customers yet?
Show the product doing one job, name a design partner as a design partner, or cite a founder credential that touches this work. Skip logo walls you cannot defend and any time-saved figure you did not measure. An empty proof slot is cleaner than a borrowed one. Add a number only when the buyer would confirm it. If you are early, say that in plain words.
How long should one homepage line stay up?
Hold one public line long enough that people outside the company hear it twice. Give it a quarter of outreach before you judge it, and rewrite when you keep losing to a named alternative or buyers file you in the wrong category. A weekly hero change never produces that evidence. When you change the line, change the site, the deck, and the posts at once, and change the outbound sentence in that same pass.