Verified against ChatGPT · 2026-08-13
Define your ideal customer profile from the deals that actually renewed, not your whole customer list
Builds an ICP definition weighted toward accounts that renewed and expanded rather than every closed-won deal ever, so the resulting profile targets accounts likely to stay and grow rather than just accounts easy to close once.
The prompt
Ready to copy — highlighted parts are example details you can swap.
Help me define our ideal customer profile. I don't want this based on every deal we've ever closed — I want it weighted specifically toward the accounts that renewed and grew, because a customer profile built from every close, including the ones that churned fast, will point our sales team back at the same easy-but-bad-fit accounts. ACCOUNTS THAT RENEWED AND EXPANDED Mostly 50-200 person B2B services companies, bought because a specific manager was drowning in a manual reporting task, adopted across the whole team within 60 days. ACCOUNTS THAT CHURNED OR NEVER EXPANDED Similar company sizes on paper, but bought during a broad 'digital transformation' initiative with no specific owner, usage stayed limited to one person, churned at renewal. OUR PRODUCT AND WHAT IT'S BUILT TO SOLVE Workflow automation tool aimed at eliminating manual reporting and status-update work for operations teams. WHAT WE CAN ACTUALLY FILTER FOR DURING PROSPECTING Company size, industry, job titles on LinkedIn, and whether a specific pain point is mentioned in outbound replies or discovery. Compare the strong and weak account groups directly, and identify what genuinely differs between them across firmographics (size, industry, structure), the specific problem they had when they bought, and how they were using the product a few months in. Do not just describe the strong accounts in isolation — the profile only becomes useful once it's clear what distinguishes them FROM the weak accounts, since two groups that look identical on paper but had different outcomes means the real differentiator is something else you need to dig into, not the obvious firmographic traits. Be explicit about which differentiating traits can actually be checked during prospecting (before a deal even starts) versus which ones are only visible after the sale — an ICP is only actionable if a rep can screen for it before investing time in a deal, so clearly separate "usable at prospecting time" traits from "only knowable in hindsight" traits and build the profile primarily from the former. Where the strong and weak groups don't clearly differ on some dimension, say so rather than manufacturing a distinction that isn't really there. Do not assume information about accounts that wasn't given to you. Output: a comparison table (Trait, Strong Accounts, Weak Accounts, Usable at Prospecting Time?), followed by a concise ICP statement built only from the traits that actually differentiated and are checkable early.
Customize
Optional — swap in your own details for the highlighted parts above.
Why this works
Comparing the strong and weak account groups against each other, rather than describing the strong accounts alone, is the single mechanism that keeps this from producing a generic "who are our best customers" description — an ICP built only by describing winners tends to just restate obvious firmographic facts (company size, industry) that may be shared equally by the accounts that churned, which makes the resulting profile look complete while filtering for nothing useful. Explicitly separating traits that are checkable at prospecting time from traits only visible in hindsight addresses a very common and costly gap in AI-generated ICP work: a model asked to build a customer profile will readily surface a genuinely differentiating trait like "adopted across the whole team within 60 days," which is a real and true pattern but completely useless to a rep trying to qualify a lead before a deal has even started, since that information doesn't exist yet at that point — without this instruction, the resulting document reads as insightful but can't actually be operationalized in prospecting. Requiring the model to say explicitly when two groups don't differ on some dimension, rather than manufacturing a distinction, guards against the tendency to produce an evenly-weighted list of differentiators regardless of whether the underlying data actually supports each one — a real ICP usually has a small number of traits that matter a lot and several that don't matter at all, and forcing the model to admit the latter is what keeps the final profile sharp instead of diluted with plausible-sounding but unsupported criteria. This structure specifically counters the common go-to-market mistake of an ICP built from an entire closed-won list, which quietly re-encodes and reinforces whatever the sales team has been closing regardless of whether those accounts actually stuck around and grew.
What you get back
Trait: presence of a specific named pain owner at time of purchase. Strong accounts: present in nearly every case — a specific manager drowning in manual reporting. Weak accounts: largely absent — purchase driven by a broad initiative with no individual pain owner. Usable at prospecting time: yes, checkable directly in discovery conversation. ICP statement: 50-200 person B2B services companies where a specific operations or reporting manager can name the exact manual task costing them time today, not a company undergoing a broad, ownerless digital transformation initiative.
Verified against
ChatGPT GPT-5.1 · 2026-08-13
Changelog
- 2026-08-13 — Initial publish, verified against ChatGPT GPT-5.1.
Need this built into your business?
If a prompt isn't enough — Google Ads management, built and maintained for you — that's Scult's day job.
EXPLORE GOOGLE ADS MANAGEMENT
