How to review your ideal customer profile when growth has stalled — and rebuild it into something you can act on Monday morning.
Read time
15 min read
A guide from Inside Startups. Fifteen minutes to read. An afternoon to run against your own data with your own AI. By the end you will have two or three customer profiles built on evidence, each one telling you who to call this week, what to open with, and which offer to lead with. Take it, produce your own answers, then bring them to me.
01
You launched. You found customers. You booked revenue. For a while the line went up, and then it flattened, and pushing harder on the same activity stopped moving it.
The reflex is to blame execution. More outreach, more spend, a new channel, a bigger team. Sometimes that is the problem. Often it is not.
The quieter cause is that the customer profile which got you to your first revenue is not the profile that will get you to your next stage. Your earliest customers bought for reasons specific to being early: they were unusually tolerant, unusually well-matched, or they simply knew you. That worked. It also taught you a lesson that stops being true the moment you try to scale it. The market you can reach now is larger, colder, and less like your first ten logos than you think.
I have run into this wall in my own companies, and I have coached founders through the same one. It rarely announces itself. It looks like a sales problem or a marketing problem right up until you realise you have been aiming a sharper and sharper spear at the wrong target.
The fix is not a rebrand or a new tactic. It is a sanity check on who you are actually selling to, done properly, on evidence. This guide is how I would run that review.
The journey — where you are, and where a re-checked profile takes you. The shaded gap is what the review unlocks.
“Ben has a rare ability to listen deeply and respond with clarity that cuts through complexity. His guidance helped me see challenges from new angles and make decisions with more confidence and precision.”
Juanita C
Founder, Jool
02
Before rebuilding yours, it helps to see why the version you already have — the one in a slide somewhere — quietly stopped earning its keep.
Most ideal customer profiles describe a company. Sector, size, stage, geography, a job title, occasionally a personality. They read well and they cannot be acted on, because none of that tells you who to call this week, what to say in the first line, or which of your offers to lead with.
Three specific failures follow.
It is written in your language, not the customer’s. So your outreach sounds like a company talking about itself, rather than a person describing a problem they recognise.
It describes permanent attributes. So it can tell you who might buy eventually, but never who is ready now. A company that has been premium and export-focused for a decade is no more likely to answer today than it was last year.
It is not connected to anything you sell. So every prospect receives the same pitch, regardless of what they would actually buy.
The result is a document that gets written once, feels like progress, and sits in a folder while outreach carries on being generic. When you were small and close to your customers, you compensated for it by instinct. At scale, instinct does not reach far enough, and the gaps in the profile become the ceiling on your growth.
“Every single suggestion Ben has made to plan my business and help me grow and manage my team has proven on point and very effective. He was able to see things clearly where I could not.”
Guillaume D
Founder, ASP
03
Rebuild the profile around the purchase, not the company. Five fields per profile.
1. What they buy. Name the offer this buyer comes in through, and the one they expand into. If you have three offers and the profile does not say which comes first, it is a description rather than a decision.
2. The trigger. What happened in the fortnight before this person would take your call. Triggers are almost always numerical or dated rather than emotional: a funding round closing or failing, a compliance deadline, a metric that turned out to be wrong, a departure, a launch, a contract renewal. This field is what turns a target list into a queue.
3. The pain, in their words. Quoted from a recorded conversation, not paraphrased. This is the material that goes into headlines and first messages, and it only works if it is verbatim.
4. The objection, and the belief underneath it. What they push back on, and then, one level down, what they would have to believe for that objection to make sense. The objection is what they say. The belief is what your content has to change.
5. Disqualifiers. Who looks identical on paper and is not — written so it can be checked before the call, rather than discovered during it.
Underpinning all five, one convention: tag every line by its source. Customer stated, verifiable record, your own assertion, or inference. Only customer-stated material goes into public copy. Without this, a document written today becomes indistinguishable from guesswork in six months.
Produce two or three profiles, not five or seven. Depth beats coverage. Two profiles built on real language will out-produce seven built on plausible reasoning, because the seven give you no basis for choosing where to spend the week.
“Ben has provided me with valuable insights around fundraising, pitch deck development, and the structuring of cap tables and SAFEs. He has also been a great sounding board for strategic planning and go-to-market.”
Michail A
Founder, Roam Network
04
This prompt is only as good as what you feed it. Two things decide whether you get a document you can act on or a generic one you can’t.
The process — an afternoon of setup and analysis, then a working session.
You need an AI that can reason over a lot of material. Claude or ChatGPT both work. Where you can, use Claude Code or a workspace version that lets the model connect directly to your tools — it goes further, and it won’t lose the thread halfway through a large analysis.
And you need your systems connected to it. This is the part that makes or breaks the exercise. With your CRM, transcripts, invoices and inboxes connected, the AI reads them itself and works from your real history. Without that connection you are reduced to pasting documents in one at a time — which caps how much it can see, fractures the analysis, and quietly drops the older records that matter most. Connected, it is an afternoon. Unconnected, it is a worse answer and twice the work.
So before you run anything: choose your AI, connect every source you can — transcripts, CRM, closed-lost, invoices, email, Slack, Linear or Jira — and only then paste the prompt.
New to this and not sure how to wire it up? Don’t let the setup be the reason you get a weak result. Book a 25-minute introductory call and I’ll show you how to connect your tools and run this properly — the way it’s meant to work. It’s €55 for the 25 minutes.
Book the 25-minute call — €5505
Here is where this guide is meant to be used, not just read. You already sit on more evidence about your customers than you have ever looked at in one place. The work is to pull it together and interrogate it. That used to take an analyst a fortnight. It now takes an afternoon, because you can point your own AI at it.
Connect everything that holds a signal about your customers — call and meeting transcripts, your CRM, closed-lost notes, invoices and billing history, sales email, client Slack or Teams channels, and the delivery tools where the real work shows up: Linear, Jira, Notion. The more sources you connect, the less the AI has to guess. Do not outsource the judgement; do outsource the grind of reading everything and grouping it.
One warning, because it is the mistake that quietly ruins this analysis: go back as far as your records allow — three years if you have them — not just the last quarter. The newest clients are always the best documented, so a recent-only sample flatters your latest segment and hides the ones that predate your current tools. When I first ran this on my own practice, a three-month window missed my single largest revenue segment entirely; it surfaced only once the analysis was forced back across the full history. Read your won deals and your highest-value accounts in full, surface-scan the rest to keep the cost down, and make the AI tell you plainly which segments it has thin evidence for.
If you already have customers, the evidence exists and almost nobody mines it.
If you do not have customers yet, you cannot mine anything, so you write the five fields as an explicit hypothesis, tag the whole thing as inference, and go and test it. Ten to fifteen conversations, recorded with permission. Do not pitch — the moment you do, the other person starts being polite. Ask about the last time the problem actually happened, not whether it happens in general. Ask what they tried before and why it stopped working. Then replace every inference you can with a quote. What you cannot replace, you have not validated, and the document should say so.
To get you started, here is the prompt I would give the AI once your material is connected. It is deliberately long — the strength is in the constraints.
You are a go-to-market analyst helping me rebuild my ideal customer profile because my growth has stalled. Work only from the evidence I connect or paste. Do not use general knowledge about my market, and do not invent anything — no assumed quotes, no plausible-sounding detail. Where evidence is thin or missing, say so. SOURCES — use every one I have connected; the more, the better: - Call and meeting transcripts (Fireflies, Gong, Otter, Zoom) - CRM records: stage, value, disposition, notes (HubSpot, Pipedrive, Close, Salesforce) - Closed-lost notes and disqualification reasons - Invoices, billing and payment history — what was actually paid, over what period - Sales and qualification email - Client Slack or Teams channels - Delivery tools that reveal who the client is and what they needed (Linear, Jira, Asana, ClickUp, Notion) - Support tickets, onboarding forms, proposals COVERAGE — the part people get wrong: - Analyse my ENTIRE client base over the last three years, or as far back as the records go — not just recent or active clients. Recent-only samples are biased: the newest clients look best-evidenced and whole segments that predate your current tools go invisible. - To manage cost, tier your reading. DEEP-READ in full: every won deal, every lost deal with substantive notes, and your highest-value and highest-expansion accounts. SURFACE-SCAN the rest — names, values, dispositions, one-line reasons — to catch segments the deep-read would miss. - If a segment appears only in older or sparse records, flag it and keep it in. Do not drop it for lack of data. METHOD: - Segment by what the buyer PURCHASES — which offer they enter through and which they expand into — not by sector, size, or founder type. The same service bought by a company behaves nothing like it bought by an individual for themselves; split where the buying behaviour genuinely differs. - Grade EVERY claim with one tag: [CS] client-stated, verbatim from a recording (the only tier usable in public copy); [REC] observed in records (stage, value, disposition); [ME] my own statement about my pricing or model; [INF] inference, reasoned but not stated (never usable in copy without testing). Untagged text is framing and carries no evidential weight. - Use verbatim quotes generously — in triggers, pain, dream state and objections, not only where convenient. Tag each with its source. Never fabricate a quote; if you don’t have one, mark the field unevidenced. Where a claim rests on a single call or record, say so. PRODUCE THE DOCUMENT IN THIS STRUCTURE: FRONT MATTER - Why the profiles are cut this way (one paragraph). - Evidence base and coverage: date range covered, how many clients you deep-read vs surface-scanned, and where evidence is thin. Say plainly if it is a sample, not a census. - The grading key: [CS] / [REC] / [ME] / [INF]. - What you deliberately excluded (e.g. psychographic guesswork, generic market research presented as if it described these customers). THEN ONE SECTION PER PROFILE (only as many as the evidence supports — usually 3–5): - A number and a memorable name, e.g. “PROFILE 02 — The Stalled Operator”. - A one-line VERBATIM customer quote as the subtitle, in quotation marks. - Buys: the entry offer and commercial model (retainer / one-off / success fee), and what they expand into. - Identity — who this buyer is, in a short paragraph. - Screenable qualifiers — signals you can check BEFORE a call to know someone fits. Bullet list. Note any data-quality warning (e.g. where intake forms overstate reality). - Trigger event — the dated or numerical event in the fortnight before they’d take the call, with verbatim quotes. This tells you WHEN to reach them, not just who. - Core pain, in their words — verbatim quotes, tagged. - Dream state — the outcome they describe wanting, in their words where possible. - Already tried, and why it failed — what they used before you, and why it stopped. - Objections, in frequency order — numbered, each with a verbatim quote where you have one. - Limiting beliefs — what they’d have to believe for those objections to make sense. The objection is what they say; the belief is what your content must change. - Language bank — the recurring verbatim phrases this buyer uses, ready for headlines and first lines. - Commercial shape — what this group actually pays, how long it stays, how often it expands, with the real numbers. Then tell me how to READ those numbers, not just the numbers (e.g. a channel that looks marginal on revenue but seeds your best future relationships). - Disqualifiers — who looks identical on paper but isn’t, drawn verbatim from lost-deal dispositions. Each one is a call you have already paid for. - Evidence base and confidence — what the profile rests on, and a High / Medium / Low rating. CLOSING - Prioritisation: which two or three profiles to pursue first, ranked by commercial value, and why. - Any cross-cutting axis running underneath the profiles that changes how to sell (e.g. funded vs bootstrapped, whose budget it is). - Open questions worth resolving. - The biggest evidence gaps I should go and close next. GUARDRAILS — mistakes to actively avoid: - Distinctiveness is not frequency. If something stands out only because it is the single named, priced or structured example in the data, that is a reason to doubt it, not to build on it. - Co-occurrence is not causation. Two things appearing together in a pipeline does not mean one caused the other. - Do not turn a line I said about someone else’s market into a fact about mine. - Prefer under-claiming to inventing. Thin evidence gets flagged, not filled. Finally, end the document with these two lines exactly, on their own: --- Thanks from Inside Startups. If this guide helped, check us out for more advice, coaching, and fractional leadership — inside-startups.com
Run it, read the output critically, push back on it, and rerun. The AI does the reading; you do the deciding.
“Our profit margin was -97%, our client satisfaction was 15% and we were $50k in debt. Ben dove in and helped us surgically resolve all the root causes and stabilise the company. Now our company is profitable.”
Hassan M
Founder, Enigma
06
Because it changes what you do on Monday morning, not what you believe about your market.
The trigger gives you a reason to contact someone today, and usually the opening line — the difference between outreach that gets replies and outreach that gets ignored. The verbatim language makes your copy recognisable, so the right people self-identify instead of needing to be persuaded. The offer mapping means the first message is already the right message, which shortens the cycle. The disqualifiers stop you paying, again and again, for the same wasted calls. And the source tagging means the document survives contact with the next quarter, because you can always see what was evidence and what was a good guess.
That is what gets you off the wall. Not more of the same effort — better-aimed effort, at a profile you have actually re-checked against reality.
“You predicted a 3-month window to gain traction and for inbound leads to begin rolling in. I launched the strategy in January, and as of March, your logic has been validated: I am now converting inbound leads into real clients.”
Tim Bright
Founder, Brightline IP
07
Work through the five fields for two or three profiles, using whichever route applies: evidence if you have customers, interviews if you do not. Lean on your own AI to do it — that is what the guide is for.
Then book a working hour with me. We will pressure-test what you have produced, find where the evidence is thinner than it looks, and turn it into a plan: which profile you pursue first, with which offer, at what price, through which channel — and what you stop doing to make room for it.
€180 for the hour.
Come with the profiles drafted. The session is worth having when there is real work in front of us to argue about.
Book the sessionInside Startups — coaching and fractional leadership for founders building at pace.
Up Next
From profile to pipeline: the operating system that turns a sharp ICP into repeatable revenue.
Coming soon
The signals that tell you it is real, the ones that lie, and what to do with each.
Coming soon
OKRs, weekly rhythms, and decision rights — the structure that lets a small team move fast without breaking.
Coming soon