The GTM Map

The steps, and what each one costs to skip

08 steps · 01 gate · 04 feedback lines

Gate

Product / market fit

BlocksEverything. This is the only true gate on the map.

A buyer you can name, a problem they already spend money or hours on, and a pitch you have heard repeated back to you in their words. Not a hypothesis you like. Evidence you got from talking to people who could pay.

Skip it and

Every step to the right will run beautifully and produce nothing. You will spend your best names learning what ten customer conversations would have told you in a week, and burnt lists do not grow back.

ICP and personas

NeedsThe gate.

BlocksBuying signals, and through them everything else.

Who buys, who blocks, who uses. Company shape, team size, the stage they are at, and the different words each role uses for the same problem. One line per persona, not a document.

Skip it and

The list looks impressively large and converts at noise level. Worse, every number downstream becomes unreadable: you can never tell a weak message from the wrong audience, so you tune the wrong dial for months.

Buying signals

NeedsICP.

BlocksList building, message.

Fed byThe learning loop.

The observable event that says now rather than someday: a new person in the seat, a job posting for the team that owns the problem, a round raised, a market entered, a question asked in public, a review read. A signal is something that happened, with a date on it, that you can point at.

Signals are not one thing. They come in fidelities, and the trade is always the same: the tighter the signal, the smaller and better the pool.

Fidelity of the signalSize of the poolFit

They are shoppingsearched, read reviews, asked in publichighest

They just changednew role, new market, new budgethigh

They match the shapeindustry, size, stack, geographylow

Skip it and

You are writing to the whole market on your calendar instead of theirs. The reply rate drops to the floor, and you reach that floor by spending the exact names you most wanted to reach later.

List building

NeedsICP and signals. Both.

BesideMessage and hooks.

BlocksOutreach.

Turn the definition into named people with a working way to reach them. Enrich, verify, deduplicate against everyone you have already touched, and remove current customers, current pipeline, competitors and anyone who asked you to stop.

Skip it and

You write to people who left the company, to addresses that bounce, and twice to the same person from two channels. The platform decides you are a problem before a single human has judged your message.

Message and hooks

NeedsSignals.

BesideList building.

BlocksOutreach.

The angle per segment, the first line, the one piece of proof, and the ask. Personalisation is a spectrum: a merge tag at one end, one checkable fact about that specific person at the other. The fact end is the only end that works, and a person without a fact should be dropped from the list rather than written to badly.

Skip it and

The list was fine. Nobody will ever find out, because the first line read like a bot and the thread ended before your product was mentioned. This is the most expensive step to get wrong, because it fails silently.

Outreach

NeedsList and message. Both.

Loops withReplies.

Fed byThe learning loop.

Contact, at a pace the channel tolerates: accounts warmed up before volume, a daily ceiling, spacing between touches, follow-ups that add something rather than repeat, and more than one channel so one platform cannot switch you off. Outreach and replies are not a sequence, they are one running loop.

Skip the pacing and

The account gets restricted, and every name you had queued behind it goes with the account. You will be rebuilding reach instead of selling, at exactly the moment the messaging finally started working.

Replies and qualification

Loops withOutreach.

BlocksMeetings.

Fed byThe learning loop.

Read what came back, answer like a person, and sort quickly into: real now, real later, and no. Speed beats polish here by a wide margin, and a reply is the one thing on this map you cannot automate the judgment out of.

Skip it and

A reply left sitting for four days is a lost deal that you already paid for. This is the cheapest revenue anywhere on the map and the step most often left unstaffed, because it does not feel like work until it is gone.

Meeting

NeedsA qualified reply.

ExitEverything after this is your sales process, not this map.

The conversation the entire map exists to create. It is the exit point, not the finish line: what happens inside the meeting belongs to a different chart, and mixing the two is how people end up measuring outreach by closed revenue three quarters later.

Learning loop

RunsContinuously, alongside 01 to 07.

Feeds back intoSignals, list, message, outreach.

Tag every batch you send with its signal, its segment and its date. Once a week, read reply rate by slice and change one thing. Two rules keep this honest: do not decide on noise (a slice under roughly 30 sends or 3 replies is a story, not data), and change one variable at a time or you will not know which one moved.

Skip it and

You keep the anecdotes and lose the data. Year two looks exactly like year one, except the lists cost more and the good names are already used.

The GTM map

The steps, and what each one costs to skip

Product / market fit

ICP and personas

Buying signals

List building

Message and hooks

Outreach

Replies and qualification

Meeting

Learning loop

What the map is for

Find where you actually are

Split own from buy

Lay your stack over it