Define an ICP from firmographics, triggers, and macro trends

01 Pin the business question before you touch any data.

An ICP without a decision attached to it is a research exercise. Start by naming what this definition has to settle in the next quarter: where to point the next sales hire, which vertical to build for, whether to move upmarket, what to cut from the roadmap. If a section of the ICP does not feed one of those decisions, you do not need it yet. Diagnose before executing, refuse the playbook ask

Run an intake interview with product and the founder before you analyze anything. Surface the real customer problem, what is launch-ready, who will care first, and what success looks like. That conversation exposes the assumptions you would otherwise bake in silently. An intake interview guide aligns PM and PMM before any GTM work begins.

The most important thing to settle here is the market itself. Before you optimize who inside the market to target, confirm the market is buying anything from anyone. A hungry market carries an average ICP. A great ICP cannot save a market with no urgency.

"Starving Crowd (market selection) matters more than Offer Strength, which matters more than Persuasion Skills. Most entrepreneurs obsess over the bottom of this hierarchy when the highest-ROI intervention is almost always improving the offer itself or selecting a better market."

· Alex Hormozi, $100M Offers, 2026-03-03 · “Market choice (Starving Crowd)…”

Picking the market is the highest-impact move in this whole pipeline. Get it wrong and every downstream attribute describes the wrong people precisely. Market and offer beat funnel optimisation

02 Pick your animal before you tune any tactic.

Decide the revenue-per-account band you are hunting, because that single choice sets the entire GTM architecture. You cannot field-sell a $10-per-year product and you cannot self-serve a $100k contract. The ICP for a high-volume, low-ARPA motion looks nothing like the ICP for a field-sales motion, and the mistake is choosing attributes before choosing the band.

"Flies generate $10/year per customer and need 10 million of them; Mice generate $100/year and need 1 million; Rabbits generate $1,000/year and need 100,000; Deer generate $10,000/year and need 10,000; Elephants generate $100,000+/year and need 1,000."

· Christoph Janz, Five Ways to Build a $100M SaaS Business, 2026-03-03 · “You can't hunt elephants…”

Write the animal and the ARPA target at the top of the ICP doc. The firmographics, the channels, and the sales motion all hang off it. If the band shifts later, that is a deliberate re-architecture, not a quiet drift.

03 Analyze the customer base you already have. Find the high-value clusters.

Segment the current base by firmographics, subscription characteristics, lifetime value, and usage. You are looking for the accounts that closed fast, stayed, expanded, and cost little to serve. Those clusters are the empirical core of the ICP. The largest contract that took a year of customization and three executive sponsors is expensive fit, not natural fit. Weight it down.

Firmographics tell you who fits. Usage tells you who gets value, and the two diverge more than teams expect. An account can match every firmographic filter and still never adopt the feature that creates retention. Pull both and cross them.

The cluster to learn from is not the average account and not the power user. It is the account on the cusp: high intent, low success, the one that wanted the product but barely got there.

"When you fix marginal user problems, you fix everyone downstream. Focus on making that person's experience stupid easy."

· Adriel Frederick, Lenny's Podcast, 2026-04-28 · “Design for the marginal…”

04 Rank verticals by revenue, retention, and fit. Then go narrow.

Score industries and segments by account volume, revenue, ARPA, retention, and fit. The temptation is to keep the top three because cutting feels like leaving money on the table. Resist it. A boring, undermonetized vertical with real willingness to pay beats a crowded segment where better-capitalized players already won.

"I lost $10M on Flow, competing with venture-backed Asana. Don't deadlift 300 pounds on day one. There's a guy doing $30M a year on government form-filling software."

· Andrew Wilkinson on Lenny's Podcast, 2026-04-28 · “Pick a niche where…”

The narrow market that feels too small is usually the right one. The FOMO trap, widening the ICP because the niche looks unprofitable, dilutes the message until no buyer recognizes themselves in it. Wingman doubled homepage conversion not by changing words but by cutting three buyer types and writing to one.

The repositioning is on the public record: the homepage moved from "Conversation intelligence for fast-growing sales teams" to "Your Revenue Accelerator," and the company doubled ARR in the six months before its Clari acquisition. Repositioning doesn't work until you cut three buyers and pick one

05 Define firmographic criteria. At least three specific attributes.

State the company-level attributes that bound the segment: size, industry, geography, tech stack, business model. "SMBs" is not an ICP. "Series B SaaS companies, 30 to 80 person sales teams, running Salesforce, selling into mid-market" is. Each attribute should be filterable in the CRM, because that is where the ICP lives or dies.

Keep this layer at the account level. The individual you sell to comes later. Mixing them now is the most common early-stage mistake.

"An ICP is not a buyer persona — it is an account-level definition that must be operationalized as a living mandate in your CRM, sales meetings, and product roadmap, not a static slide deck."

· TK Kader, Ideal Customer Profile template, 2026-03-03 · “An ICP isn't a…”

06 Map triggers and macro trends. Static firmographics are half an ICP.

Firmographics describe who fits. Triggers describe when they buy, and the macro trend describes why now. Without the timing layer, the ICP is a static list and sales prospects into accounts that match on paper but have no reason to move this quarter.

Account usage of a competitor is a trigger most teams over-trust. A list of every company running Vendor X is not a target list. Real switching readiness needs several signals to line up at once.

"An account using a competitor's product isn't a target by itself. Most competitive intelligence tools give you a list of 60,000 companies using Vendor X and call it a day, but true switching readiness only appears when five signals line up: install maturity and renewal proximity, product spend and expansion patterns, where the product sits inside the organization, adjacent technology changes, and segment-level switching behavior."

· Disha Thakkar, LinkedIn, 2026-04-10 · “True switching readiness requires…”

07 Run JTBD switch interviews with 5–10 best customers.

Interview the best customers from step 3, recent switchers from the past 90 days where you can. The goal is not feature feedback. The goal is the language they used to describe the problem before they knew your product existed, and the specific moment they decided to look.

"JTBD interviews help you define a customer's language — what they actually mean by 'easy' — the root cause of why they switched, and how their story connects, all of which directly impact your product marketing."

· Bob Moesta, Intercom podcast, 2020 · “JTBD interviews surface the…”

Trace the timeline backward from the purchase: what happened right before they started looking, what they tried first, what almost stopped them. The switch trigger is operationally valuable because the ICP can encode it directly, and campaigns can fire on it.

08 Build personas from interview data. Persona comes after the ICP, never before.

Only now, with the account-level ICP defined, build the people inside it. The champion plus one or two secondary personas, each with a real title, a real day-to-day, and a real pain traceable to an interview. Building "Marketing Mary" before defining her company is the inversion that sends reps cold-calling individuals with no budget and no urgency.

Keep the persona honest by refusing abstraction. Name three specific customers with three specific frustrations before you write a single archetype. The truth lives one level below the framework.

"The truth is one level down. Always."

· Shreyas Doshi, Get to the Core of the Thing, 2026-05-16 · “Three customer stories beat…”

The layer most personas skip is the situation: what the buyer is trying to do, what is breaking, what triggered the search. "VP of Sales" is an audience. "Reps are calling without notes and deals stall because nobody can tell where they stand" is a champion. Sell to the buyer's mindset, not to product features

09 Write JTBD statements per vertical. Anchor each to a named customer.

Convert the interview findings into statements in a fixed format, one per vertical: "When a [customer type] [situation], they hire [product] to [outcome], so that [bigger outcome]." Every statement must trace to a named customer who actually said something close to it. If you cannot name the customer, mark it a hypothesis and flag it for the next interview round.

These statements are the bridge between the account definition and the messaging. A vertical with no clean JTBD statement is a vertical you do not yet understand well enough to target.

10 Define exclusion criteria. Who you are not for, with reasons.

A definition that includes everyone protects nobody. Write down the segments you decline and why: too small to pay, wrong stack, a job your product does badly, a vertical with a structural reason it churns. Exclusion criteria are what give the ICP teeth, because they tell sales when to walk and tell product whose feature requests to track separately rather than build.

This is also where you account for the buyer who weighs doing nothing. Some accounts match the firmographics, hit a trigger, and still never decide. When you survey them, most made no purchase decision at all.

"When you survey buyers who didn't buy, 40–60% say they made no purchase decision at all."

· April Dunford on Lenny's Podcast, 2026-04-28 · “40–60% of B2B buyers…”

If a segment loses to inertia rather than to a competitor, the remedy is problem framing, not a sharper pitch. Misreading a status-quo segment as a competitive one points the whole GTM motion at the wrong fix. Status quo / no-decision is the real competitor No-decision-as-competitor vs. battle-card-driven competitive workflows

11 Build a scoring model and connect it to TAM, SAM, and SOM.

Turn the criteria into a weighted score a CRM can compute on every account: firmographic match, trigger presence, usage fit. The score is what makes the ICP usable in a pipeline review instead of a quarterly nod. Then frame the market sizing honestly. TAM is everyone. SAM is the ICP segment. SOM is the initial customer profile you can realistically win in the next 12 months. Most teams confuse TAM with ICP and end up sizing a market they will never serve.

12 Operationalize the ICP as a living mandate.

The ICP is not done when the doc is written. It is done when CRM filters enforce it, pipeline reviews open with ICP-fit distribution before deal updates, and the roadmap prioritizes ICP-segment requests over outliers. A definition that gets a nod and then gets ignored is the default failure.

One artifact carries this well: a single source-of-truth document the whole team educates the market from, the CTO and the CS rep saying the same thing.

"I created a 'Realm of Relevancy' document that captured every critical piece of information relevant to our product's world, and used it to align positioning, messaging, content, and sales conversations. This doc enabled our growth from five figures ARR to $1.5M ARR in 9 months and helped secure Series A funding."

· Obaid Durrani, LinkedIn, 2026-04-10 · “A single 'Realm of…”

Set a quarterly review cadence with a named owner. Triggers and macro trends move; an ICP written once and never updated decays into the set-and-forget trap.