What GTM experiments should an early-seed startup run first?

Proof tests come before channel tests

The first GTM experiments at early seed falsify one assumption about a named buyer, one promise, or the promised outcome. They do not scale acquisition. A channel test waits until those three have a read.

Sam Altman at Y Combinator wrote in 2016 that growth before a product people love fills a leaky bucket: people arrive and leave, and ads or publicity carry what word of mouth will not. gtmpreneur will not let you skip stages: problem evidence, founder-led sales until a handful of paying customers share one profile, then one motion you can repeat. Outbound Capital runs market, then product, then go-to-market, then pricing.

Use that order when you can name a buyer and deliver the outcome by hand. If people pay and then leave, test activation, not a new channel. If you cannot deliver the outcome manually, stop at interviews and do not sell a pilot you cannot staff. Validate the ICP before you raise volume. A first go-to-market strategy is what you write after a motion survives.

Write the stop before you write the tactic

A GTM experiment is a hypothesis, one change, one count, a window, and a decision written before the result.

Fill the card before you send. Example: an operations lead, a role just opened, and the weekly report done before the meeting. One change, on one list. Count buyer conversations, not opens, for two weeks. If it fails, the promise or the slice changes and no channel is added. Traction in only one slice means the next test keeps that slice.

Stackmatix names the failures: a change too small to read, a noisy metric, a forgotten window, or three changes at once. Most early tests they describe run one to four weeks. Before product-market fit, Stackmatix keeps tests on the problem and the activation moment, not on scaling acquisition.

Dan Hockenmaier, with Gustaf Alströmer, says to talk founders out of the backlog. Most of the gain sits in one channel and a handful of tactics. You need one cohort where retention levels out. Heavy acquisition, especially paid, before that pattern is too early.

Six experiments, in the order the risk allows

Run the next row only after the row above has passed.

Order	Experiment	Proves	Does not prove	Stop if
1	Two slices, listening only	A recent workaround and someone who can act	Payment, or a channel that scales	Nobody names the last time it happened
2	Three outcomes, one list rule	Which outcome earns a qualified next step	That another channel would win	Polite replies, no meeting, no scope
3	A costly next step	Payment, a scoped pilot, or named internal time	Retention	Everyone wants a free look
4	Concierge delivery	The outcome can happen by hand	That software does it alone	Extra effort still misses the outcome
5	Activation watch	New users hit the promised moment	That more traffic helps	People start and never arrive
6	Same motion, fresh batch	The path repeats on strangers	A second channel or a hire	Every win needed a new story

Outbound Capital starts wide, then narrows. gtmpreneur and Pier constrain first: two slices, not five. SalesCaptain uses 10 to 15 interviews in about two weeks, and three prompts: what breaks, who signs, and what would make them switch inside a quarter. Also ask the last time it happened, the workaround, and what they do if they do nothing. Skip “would you use this” and “would you pay today.” Superpublic says problem interviews are weak for proving payment. A paid pilot, with a clear scope and a stated support expectation, is their test of urgency and of whether a buyer will commit.

Read signals in order: attention, intent, activation, a value exchange, durability. Activation is the promised outcome, not a signup. Pier’s alternative is the spreadsheet, the intern, or doing nothing. A fake door fits only when those people already visit and you cannot talk to them.

1. Week 1. Two slices, listening only. Exit with one profile, one use case, one trigger, and the person who can commit. Pier’s trigger is the recent event that makes them look or act. On this page a stranger must be able to check it, such as a role just opened, new funding, or a leadership change. A vibe is not a list rule.
2. Week 2. Three outcomes for the same slice and the same next step: time back, a failure avoided, the current workaround removed. Not three tones, and not three channels. Keep what earns buyer conversations, or change the slice.
3. Week 3. Pier’s loop: a list in the low hundreds, small batches, one hook rewrite, then a scoped pilot or payment.
4. Week 4. If they miss the outcome, change onboarding, not the channel.

Founder-led GTM before a full team is this month.

Outbound, partners, content, and paid are different questions

Compare the four motions by the question each one can answer now.

Motion	Seed question	Honest count	Kill	Leave closed
Outbound	Will a named buyer meet on one promise?	Buyer meetings from a list a stranger could apply	One rewrite and still no meeting, or the email has no lawful basis	The promise changes every send
Partnerships	Will this one partner introduce buyers?	Introductions that become conversations	No named introduction in the window	No partner, only a wish
Content	Can one asset enter a live thread?	Conversations the piece caused	It never enters a sales conversation	You publish to look active
Paid	Will this audience take the step warm visits take?	Starts or booked calls, by source	Clicks come, the next step does not	The promise is still unheard

Outbound is first when you can name a title and a trigger. Pier chooses from where the buyer spends time, how long the sale is, and what you can already do. A short cycle leans outbound. See content that supports outbound. A community keeps the same message.

Under article 11.7 of the Dutch Telecommunications Act, commercial email is barred unless you can show prior consent. No consent is needed if a business recipient published the address for unsolicited commercial messages and you use it only for that purpose, or if they are already your customer, you offered an opt-out when you collected the address, the note is about your own similar offer, and every send has an opt-out. A scraped personal work address is neither. Outside the EEA, follow that country’s rule. Otherwise call or interview, and keep the proof.

Rank by the risk you can falsify this fortnight

The next experiment is the one that can kill the current risk: problem, commitment, activation, or repeatability. If the last people missed the outcome, do not buy more of them. Message and channel in the same week is two tests, so run the message first. Use one to two weeks for a conversation or a milestone, and up to about four weeks to watch activation. Do not kill a recurring problem on day fourteen. No buyer conversation in the batch means you change the promise or the slice, and you add no channel. Score impact, confidence, and ease only after a motion exists. Three or four channels need pipeline you can already attribute.

One live motion, one watch, and a closed backlog

A founder alone runs one distribution test and watches the people already in. Park the rest with a reopen date. Stackmatix keeps two to four live tests for a small team, split across acquisition, activation, and retention.

Do not change the list rule mid-batch. Do not let a sequencer send while the sentence still moves. GTM AI is clear: tooling can speed the calls and will not fix a vague promise. Who runs GTM experiments without a team is staffing, not extra motions.

Double down only when a second batch repeats

Add volume only when a fresh batch repeats the same promise, takes a costly next step, and reaches the outcome.

1. The chain holds through the outcome. Stop at the first break.
2. Inputs and a rough output are written so someone other than the founder can run them.
3. New accounts reach the outcome without a rescue you cannot describe.

Pier adds a second channel only after consistent output over a long stretch, not after a good week. gtmpreneur hires the first seller or marketer after the founders have sold repeatedly. When to hire the first growth marketer starts there. Stop if only friends convert, one logo is the story, or the win is opens and clicks.

Leave these tests on the shelf

Park a test whose no would not change the slice, the promise, or the roadmap. That includes a fake door with no matching traffic, paid as the first hearing of the promise, three channels from a checklist, a referral program before an outcome, three public price tiers before a payment, an SDR or agency hired to discover the motion, a split test on an empty page, and a savings claim you cannot show.

The first month I would run with you

In the first month I would run the two-slice listening sprint, keep one buyer, run one message contrast, and ask for a scoped next step before anything sends on its own.

I am Piet Baudoin, one person, Poldermarketing, and I work remote in Dutch and English. You can hire me freelance for a bounded project or for a few days a week. I am an AI-native growth marketer who builds and executes, strong in AI, content, automation, and workflows. Google Ads and Meta Ads are relatively new for me: I can set them up and review them, and I do not run them at scale. That is an all-in GTM solution for freshly funded startups anywhere, fully remote, from the first message to the first customers: marketing, AI, and automation, without a separate specialist for each part.

If the promise is still hard to say, routing will not fix it. The free growth scan shows how the site reads to a stranger. How I work is the shape of an engagement. Positioning and messaging is where I start when the words are the leak.