The Eight Sources Every B2B GTM Foundation Needs

A B2B GTM foundation needs eight kinds of evidence before anyone drafts strategy from it: website and public positioning, buyer call recordings, customer interviews, LinkedIn profiles of real people in the persona seat, a competitive and market scan, sales and pipeline data, product and delivery reality, and the operator's own recorded judgment. Work them in order.

If your AI drafts a positioning statement in ninety seconds, the ninety seconds are not the problem. The problem is what it drafted from.

Most GTM foundations in this market are built on a website crawl, a competitor list, and the operator's memory of a kickoff call. That base produces work that reads plausible and dies on contact with a real buyer. You find out in month two, when the outbound gets no replies and the client asks why.

So here is the utility of this piece: the eight kinds of evidence that should be in a GTM foundation before anyone — human or AI — drafts strategy from it, what "good" looks like for each, and how to close the gaps fast. Work the list in order. Each source unlocks the ones below it.

Source 1 — Website and public positioning (the claim you're inheriting)

What it is: the client's own site, pricing page, product pages, and any positioning deck they've already shipped.

What good looks like: you can state, in one sentence, who they currently claim to be for. If you can't, that's your first finding, not a missing input.

Why it's first: everything after this is a test of whether that claim survives evidence.

Source 2 — Buyer conversations (the only source that can quantify a pain)

Recordings and transcripts of sales calls, discovery calls, and win/loss debriefs. Not summaries — the actual words.

What good looks like: at least [add minimum count] recordings across won, lost, and churned deals, close enough to now that the market hasn't moved under them.

This is where pain becomes real. As one of our own interviews put it: "If you cannot quantify the pain, the pain is not really a pain yet." A transcript is the only place a dollar figure or an hour count shows up in a buyer's own mouth.

Source 3 — Customer interviews (the story behind the persona)

Adjacent to Source 2 but different in intent. Sales calls tell you what buyers say when they're being sold to. Interviews tell you what happened.

What good looks like: the interview asks for narrative, not lists. Not "what are your challenges" — instead: "tell me about the last time this went really badly, and the next time you expect it to." Behind every good persona is a customer who can narrate both.

Source 4 — LinkedIn profiles of 2–3 real people who ARE the persona

Source 4 is the newest required source in a GTM foundation, and the one operators resist most — because pulling two or three LinkedIn profiles feels like a shortcut when it's the opposite.

Pick two or three named humans who genuinely sit in the target seat. Not composites. Real profiles. Two to three is the working minimum: one profile is an anecdote, and past three you stop learning new vocabulary and start collecting duplicates.

What good looks like:

• Their actual job title and how it differs from the title you assumed
• Their career path — what they did before, which tells you what they'll believe
• What they post and repost, which is your vocabulary source
• Company size and stage, which sets the buying-committee shape

A persona built from three real profiles stops being a demographic sketch and starts being someone you could recognize on a call. It also settles arguments that composites can't: as Stijn puts it in the internal training, positioning foundation — who is it for, what's it for, value prop — comes before anything else, and brand voice and visuals only work once that foundation is done. Three real profiles are the fastest way to finish "who is it for."

Source 5 — Competitive and market scan (what buyers get told when you're absent)

Not a feature grid. The scan you need answers: what do buyer LLMs and search engines actually recommend when someone asks for a solution in this category?

What good looks like: named competitors with visibility counts, the queries where you're cited, and — more valuable — the queries where you're absent entirely. Absence is the whitespace. It's where the next quarter of content lives.

Source 6 — Sales and pipeline data (the reality check on the story)

CRM stages, cycle length, deal sizes, close rates by segment.

What good looks like: enough volume to tell you which segment actually converts, versus which one the founder likes talking about. These two are rarely the same company.

Source 7 — Product and delivery reality (what you can honestly promise)

Roadmap, onboarding flow, support tickets, implementation notes.

What good looks like: you can distinguish what ships today from what ships next quarter. This is the guardrail against a positioning that the product can't service — the fastest way to lose an engagement in its second month.

Source 8 — The operator's own judgment, recorded as a source

Your read on the political dynamics, the founder's real appetite, the constraint nobody wrote down.

What good looks like: it's written down and attributable, not carried in your head across four concurrent clients. Judgment held only in memory is the single most expensive thing in a fractional practice — it's the reload cost you pay several times a day.

What the foundation checklist does on its own

Enumerating eight sources is easy. Getting eight sources into a base while you're mid-engagement on four clients is the actual work — which is why in T2D3 OS the foundation checklist is deliberately pushy. The scale of that work is the point: a fractional running four clients is typically logged into four different CRMs, four analytics stacks, four Slack workspaces and four Drives, and fractional practices lose an estimated 50–60% of their working time to stitching disconnected tools together instead of shipping billable deliverables.

It surfaces the gap rather than waiting to be asked. And it arrives having done homework: the public sources it can gather itself, it gathers — the site scan, the competitive visibility read, the market vocabulary — so what's left for you is the part only you can get. For the sources that need a human, it drafts the ask: the interview request, the recording pull, the list of three profiles to confirm.

That order — evidence first, output second — is what the Kalungi Masterclass session landed on too: as Yusuf framed it, regardless of whether you call it SEO, AEO, GEO or LMO, the goal is brand visibility, and strong positioning, a clear ICP, a technically sound site and original credible content remain paramount.

You can snooze any of it. You cannot make it quietly disappear, because a foundation with three of eight sources will still generate confident output — and confident output on a thin base is the failure mode this whole practice exists to prevent.

Run it this week

1. Score your current foundation out of eight. Be honest about partials.
2. Fix Source 4 first — three LinkedIn profiles is an afternoon.
3. Pull the three most recent lost-deal recordings for Source 2.
4. Write down Source 8 before you forget it.

Everything downstream — scoring, querying, auditing — reads from this base. Get the base right and the rest is judgment. Get it wrong and the speed just gets you to the wrong answer faster.

Last updated [publish date]