How should an early-seed startup build its first go-to-market strategy?

Treat the first GTM as a learning system

The first go-to-market finds a repeatable path to the right customer.

Sam Altman separates growth from step one. A product people do not love is a leaky bucket: they arrive and leave. Airbnb slogged for 1000 days, then grew by word of mouth. Raising to hide that gap leaves the problem and raises the expectations.

Michael Seibel quotes Marc Andreessen: customers buy as fast as you can make the product, or usage grows as fast as you can add servers. Money piles up. You hire sales and customer support as fast as you can. Stay lean until loyal customers swamp a small team. Hold the problem tightly and the solution loosely, so the market pulls the product out. Staffing before retention and growth weakens the next raise. Justin.tv had users and revenue and no growth, with a large team. After that raise failed, they cut the team, and the gamers already streaming became Twitch.

Refuse a buyer before you describe one

An ICP you can run is a filter a stranger could apply this week.

Seibel retells Sequoia’s hair-on-fire test: the pain is urgent enough that someone tries a rough version. Geoff Ralston wants 10 customers with a burning problem, not 1000 with a passing annoyance, and a pocket of 10 to 100 people who love it. Drop accounts that cost more, in revenue or learning, than they return.

Line	What you must be able to say
Who	A role, in a kind of company you can name
Trigger	The event that makes this urgent now
Pain	Who feels it on their real cadence
Yes	Who can approve the next step
Alternative	What they do today, including nothing
Refuse	Who you will not take, even if polite

If you cannot name ten accounts that pass, and that you could reach without buying a database, you do not have a segment yet. A name on a list is not enough. You need the same trigger from several of them, in their words, plus the workaround they already run, such as a spreadsheet, a manual step, or a deadline. If you had to teach them the problem, they are not the segment. Define the ICP first, then validate it before outreach.

Shekhar Kirani says the first customers set the product’s DNA, so do not take that first set from one country, and do not paste one country’s employee band onto the filter.

Product, sales, and marketing share one loop

Before fit, product, sales, and marketing are one loop.

Ralston’s order is a quantum of utility, then talk and iterate, almost all the time. Skip conferences unless they are the route. Solve one acute problem. Ralston’s record of Buchheit’s 90/10 solution is nine tenths of the outcome for a tenth of the effort, now.

Peter Reinhardt says qualify before the demo. Segment’s list is MEDDIC: metrics, economic buyer, decision maker, decision process, identified pain, champion. While searching, those are questions. Wishes without use are a death spiral. Use, then a next request, is the turn. Professors sounded keen. The reference user said he probably would not use it. Counting classroom screens was the check. The landing page taught less than later conversations.

Sales is how the founder learns the buyer. Marketing is those words, placed where that buyer already looks. Kirani: if the founder skips early sales, it seldom works. The first commercial hire, if any, is a flexible individual contributor beside you, not a sales director or VP. Founder-led GTM before a full team is that sequence.

Match the motion to how this buyer can start

Pick the one motion this buyer can start this quarter.

Dan Hockenmaier calls it a power law: most of the gain is one channel and a handful of tactics. Paid acquisition before a cohort’s retention levels off is premature. Gustaf Alstromer adds that a weak product will not grow, and that you should ask how people heard of you, because tracking misses word of mouth.

Motion	Use it when	Do not use it when	Continue only if
Founder-led sales	The deal needs a conversation or setup	A stranger reaches value alone	Same segment uses it and takes a next step
Product-led	A new user reaches value with no long review	You must be in the account	A cohort returns without a chase
Partner-led	Someone trusted introduces this segment now	The list is logos	Intros reach the same value event
Public artifact	They already ask where you can answer	Nobody is searching	Inbound that passes the filter becomes use

Segment started on the last row: a library on Hacker News, then a hosted product, with the founders still in the conversations.

Kirani scores search on the right visitors converting, not on winning costly keywords. Keep outbound vertical even when the product is broad. He treats daily use, or use several times a week, as the sign the product will not be abandoned. Skip that test when the buyer’s real cadence is a weekly close or a monthly report, and judge that cadence instead. On a marketplace, Hockenmaier says paid is a surgical way to fill the thin side in one place, while the core stays on the channel that already repeats. Do not pay for the side that is already dense. Which experiments to run first is this choice.

If the weekly action is commercial email to people in the EEA, send it only with prior consent, a real sender name, and an easy opt-out in every message. An existing customer is the exception when the message relates to an earlier purchase. A company address is a different case. Do not treat a founder’s personal address as a company address.

Keep the whole bet on a single page

Before fit, the plan is one page. If a choice changes who you contact, what you build, or where the week goes, it is on that page.

Block	Write this
Beachhead	The six lines, including who you refuse, plus the same buying path and a network where these buyers already talk to each other
Promise	Buyer, status quo, outcome, mechanism, alternative
Offer	One use case, a time box, one success event, no open custom
Motion	One row above, and one named weekly action
Proof	A workflow, a number they track, or a reference
Score	Passing conversations, value events, next steps, who returns
Stop	The date, and the behavior that means narrow, rewrite, or quit

The value event is the first behavior that predicts the outcome you promised. A login, a meeting, a waitlist, or a compliment is not that event unless you have seen it predict the outcome.

Suppose the beachhead needs a conversation before anyone can start. That is the founder-led row. The same product for buyers who reach the result alone is the product-led row.

If a spreadsheet could say the sentence, it is not positioning. That sentence is the homepage message.

Ralston records Buchheit on the unit: taking less than it costs to serve them is not a strategy at higher volume. Kirani allows a rough early price if gross margin holds, then a unit that grows with the customer, such as a seat, usage, or storage.

Run the motion by hand until the path repeats

A good early motion is repetitive. Same buyer, same trigger, same alternative.

1. Every conversation passes the refusal list.
2. Start in their workflow, and fill the MEDDIC questions before you show the product.
3. Ship the 90/10 slice. Airbnb’s founders took the early listing photos themselves.
4. Continue only if the same segment used it and took a next step.

A famous long slog is a bad excuse for missing use.

A tiny team spends the week inside that loop

The week has five blocks: filtered conversations, one change at the sharpest stall, hands-on delivery, one note in the buyer’s words, and a page update of what you stop.

Do not split that week across an agency, a prospecting pod, content, and paid media. Operationalizing founder-led GTM starts after the path repeats, not before.

Hire only when someone who missed the calls could follow the path. Skip the program when the value event is unnamed, when you will not refuse a friendly logo, or when the custom work teaches nothing about the next buyer.

Stop, narrow, or rewrite when use is missing

Judge the week on behavior. A good mood in the room is not a metric.

What you see	What it means	What you do
Compliments, no use	Mild pain, or outside the workflow	Change the problem or the slice
Use, no next step	Wrong offer, or the wrong person	Rewrite the step, or change who you ask
Same objection each week	The objection is the product	One 90/10 change, then retest
Meetings outside the filter	The list is wrong	Fix the refusal line before more volume
New channel from impatience	The test is not finished	Hold until the stop rule fires
A leader hired to find it	Nothing is ready to hand over	The founder stays on the calls

What I run with a founder on the first motion

I run one motion with the founder, from the first sentence through to the first customers.

I am one person, an eenmanszaak named Poldermarketing. I work fully remote, in Dutch and English, with freshly funded startups anywhere. Hire me freelance, or for a few days a week. One seat covers marketing, AI, and automation, without a separate specialist for each part.

I am AI-native, and strong in AI, content, automation, and building workflows. Google Ads and Meta Ads are newer for me: I set them up and review them, and I am the wrong hire for scaling a large paid program. I build and execute, not only advise. On this brief I write the one-page bet and join the early calls. The free growth scan shows how the site reads. How I work is the engagement. Positioning and messaging is where the sentence is made.