The GTM engine: how to find your first paying customers in 12 stages

Most founders launch, post about it, and wait. Two weeks later there are a few free users and zero paying ones.

The product usually isn't the problem. The problem is that nothing is finding the people who are ready to buy right now, and nothing is learning from who actually paid.

This is the engine I use instead. It has 12 stages in 4 phases, and it loops. Every stage hands one thing to the next, so you always know what to do next and what "done" looks like.

UNDERSTAND        LISTEN                         ACT                        LEARN
link → brief → ICP → trigger → detector → signal → lead → playbook → campaign → sequence → funnel → loop
                     ↑_____________________________________________________________________________|
                                       what paid tunes what you listen for

Phase 1: understand who buys

This phase hands off one thing: a filter of who fits and who doesn't.

1. Link. Everything starts from your product's page. If the homepage is vague, everything after it will be vague too. Fix the page before you go looking for people.

2. Brief. Read your own site like a buyer would. Write down the job it does, what it replaces, the price, how buying works and what proof you have. Mark how sure you are about each line: something you can show (a payment, a quote) beats something you believe.

3. ICP (ideal customer profile). Turn the brief into a filter. Write "fits if" lines (a solo founder, selling software, launched in the last 90 days) and "not if" lines (agencies, enterprises, people who want it free forever). Every line should be checkable on a stranger from the outside.

Phase 2: listen for who's ready

This phase hands off a ranked list of people who are ready now.

4. Trigger. A trigger is the moment someone becomes ready to buy. "Any tool for this?" "I'm done with my current app." "Just launched and nobody signed up." Triggers come in two kinds: cold ones from strangers in public, and warm ones from people already touching your product (signed up but stalled, hit a limit, payment failed).

5. Detector. A detector is a saved search that watches for one trigger, in one place, every day. Where to look (a subreddit, X search, a review site, a launch directory), what to match (the phrases people actually use) and how fresh a match has to be.

6. Signal. A signal is one real person caught in a trigger. Keep the exact post, when it was posted and when you caught it. Throw out the ones that are stale, off-topic or from competitors, and write down why. The "no"s teach you as much as the "yes"es.

7. Lead. A lead is a signal that passes your filter. Merge the same person seen in different places into one profile, then rank them: fresher triggers and more than one signal go first.

Phase 3: act, one play at a time

This phase hands off messages actually sent, each one traceable.

8. Playbook. A playbook is how you help someone in a specific moment: answer the question they asked, offer the thing they're missing, show the fix. Pick plays that fit your stage. On a waitlist, the offer is early access, never a trial.

9. Campaign. A campaign is one playbook, run on a short list of leads, until a date, with a stop rule written down before you start ("stop after 30 people if fewer than 3 reply").

10. Sequence. The steps, in order: help first, one follow-up, then stop. You send every message to a stranger yourself. Put a unique tag on every link so you know which message did what.

Phase 4: learn what worked

This phase hands off one tested change, and then it loops.

11. Funnel. Count every step: reached, replied, visited, signed up, activated, paid. Only your payment provider marks someone as paid, never a form or a guess. Under 10 people, look at counts only; around 30 and up, rates start to mean something.

12. Loop. Every number gets a line, a threshold. When one crosses its line with enough data behind it:

• if it's a guardrail (people asking you to stop, broken links, a source that went quiet), act right away: pause it, fix it;
• if it's performance (a reply rate dropping), diagnose it, write one hypothesis, and test one change, with the win rule set before the test.

Ship what wins, revert what doesn't, write down what you learned, then move the line. What paid flows back into which triggers you watch and which playbooks you run.

The rule underneath all 12 stages

Every decision is data. Your edits to a draft, the leads you skip, who replied, who paid. Record them from the first day, with what decided them and when, and the engine gets better every week instead of every quarter. Recompute them later and they're useless; throw away the "no"s and there's nothing to learn from.

The one thing I'm still fixing in my own setup: an id on every message I send. Without it, nothing after "sent" can be traced back to the trigger, the play and the detector that found the person.

Start this week

1. Rewrite your filter: 3 "fits if" lines and 3 "not if" lines.
2. Pick one trigger, the clearest one your buyers say out loud.
3. Set up one detector for it: one place, the exact phrases, checked daily.
4. Answer five people this week. Help first, no pitch unless it fits.
5. Tag every link and count replies, visits and signups.
6. On Friday, look at the weakest step and change one thing there.
7. Repeat. Add a second trigger only when the first one has produced a result you can read.

Where VibeDraft fits

I'm building this engine into VibeDraft: it reads your site, finds the people asking for what you built and drafts the messages. You send every one. It's in waitlist mode right now.

If you want it running for your product, join the waitlist below.