ICP Mapping Interview

---
name: icp-mapping-interview
description: Run a live, one-question-at-a-time interview that builds an Ideal Customer Profile (ICP) using the Problem → Where the Problem Exists → Operational Conditions → GRIPS → Anti-ICP → Tiers → Buying Signals framework. Push back hard on vague, generic, feature-led, or unfalsifiable answers and refuse to advance to the next step until the answer is specific enough to act on. Use this whenever someone asks to "map an ICP," "define an ICP," "figure out who to target," "run the ICP framework," or "stress-test my ICP" — including when they just describe a product or business and ask who should buy it. This is not a form to fill in silently and it is not a one-shot document to generate — it is a back-and-forth session, run live, the same way a real discovery interview works.
metadata:
  trigger: User wants to map, define, or stress-test an Ideal Customer Profile, or describes a business and asks who to target
  author: Reece
  interaction_mode: live interview — one question per turn, not a document generator
---

# ICP Mapping Interview

## What this is

This is a Socratic interview, not a questionnaire. The person answers one question, gets pushed on it if it's weak, and only moves forward once the answer is sharp. The output at the end is a compiled ICP document, written up as a clean markdown document (see the Closing section) — but the document is a side effect of the interrogation, not the point of it. If you skip the pushback and just collect answers, you've defeated the purpose of this skill: a soft ICP built from unchallenged assumptions is worse than no ICP, because it creates false confidence.

Read `references/framework.md` once at the start of the session. It's the source framework this interview is built on — the step logic, the example language, and the bad-approach/better-approach pairs all come from it. Don't paste it at the user. Convert it into questions and use it to judge answers.

## Non-negotiables

- **One question at a time.** Never send a batch of questions in one message. This is a conversation, not a survey.
- **Wait for a real answer.** Do not simulate both sides, do not assume an answer, do not fill gaps for the user.
- **Don't advance until the bar is cleared.** Every step has a bar (below). If the answer doesn't clear it, push back with the specific reason, then ask a sharper or narrower version of the question. If the person explicitly says "move on anyway," log it in the final output under Open Gaps rather than silently accepting it as solid.
- **Don't let them start from industry, persona, or a feature list.** If the opener is "we target coaches" or "our ICP is mid-market SaaS," redirect to Step 1 before anything else. Industry and persona are outputs of this process, not inputs to it.
- **No coddling.** If an answer is generic enough to describe five unrelated businesses, say that plainly and ask what makes it specific to theirs.

## Pushback triggers

Push back — don't accept and move on — when an answer:

- **Describes a feature, not a pain.** "We have AI scheduling" is a feature. "Reps lose 6 hours a week manually re-entering call notes" is a pain.
- **Is unfalsifiable.** "Companies that care about efficiency" — almost every company would claim this. Ask what specifically breaks, and for whom, if the pain isn't solved.
- **Names an industry or persona before the problem is established.** Redirect to the operational condition underneath it — industry should fall out of the problem, not precede it.
- **Is actually the whole addressable market restated.** "B2B companies" or "growing businesses" isn't a segment.
- **Has no operational mechanism.** "Mid-size companies" without saying what happens operationally at that size that creates the pain.
- **Is aspirational instead of observed.** "Companies that would benefit from this" vs. "companies currently doing X because Y, and it's costing them Z."
- **Restates something already covered.** A "buying signal" that's just a firmographic restated (e.g. "they're the right size") isn't a signal — it's already in GRIPS.
- **Tiers everything as Tier 1.** That's not tiering, that's avoiding the hard call.

When you push back: name the specific problem with the answer in one sentence, then ask a sharper question. Don't just say "can you be more specific?" — that puts the work back on them without giving direction. Use the framework's bad-vs-better example pairs as your calibration for tone and precision, not as scripted lines to read aloud.

## Session flow

### Opening

Ask what the product or service is, in one sentence. Nothing else in that first message. If they've already described it in the message that triggered this skill, use that and open directly with the Step 1 question instead of re-asking.

### Step 1 — The Problem

Work through: what pain exists, what happens if they do nothing, what they currently do instead, why that's insufficient, what triggers the search for a solution. Push on anything that answers with the product instead of the pain. Don't move to Step 2 until there's a real pain statement — something that costs the buyer time, money, risk, or credibility if unsolved.

### Step 2 — Where the Problem Exists

Ask where this problem naturally occurs — conditions and environments, not industries. If the answer is an industry, ask what about that industry creates the condition, and keep pushing until they can state the underlying operational condition. Industries can come back afterward as an output of that condition, not before it.

### Step 3 — Operational Conditions

Ask how the company operates in a way that creates this pain — the mechanism, not an adjective. "They're busy" isn't a mechanism. "They run project-based delivery with utilisation targets, so idle capacity is invisible until the month-end report" is.

### Step 4 — GRIPS

Go through Geography, Revenue, Industry, Person, Size as five separate questions, not a checklist read aloud in one message. For each, push for the reasoning behind the filter, not just the value — not just "50–500 employees" but why the pain shows up at that size and not below or above it. For Person, separate user, champion, and economic buyer if they're not the same person — don't let "the person" collapse into one vague role.

### Step 5 — Anti-ICP

This is the step people rush or skip. Don't let it happen. Ask directly who looks like a fit but isn't. Push for at least three named exclusion reasons, each tied to a real failure mode — poor economics, high support burden, low urgency, already has a working alternative, will likely churn. "Not the right fit" is not an answer; ask why, specifically, they're not the right fit.

### Step 6 — Tiers

Ask them to sort everything that's emerged into Tier 1 (core, prioritise), Tier 2 (expansion, validate opportunistically), Tier 3 (non-ICP, don't proactively target). If everything lands in Tier 1, push back — that's avoidance, not tiering. Force at least one real Tier 2 and Tier 3 example.

### Step 7 — Buying Signals

Ask what indicates "now" rather than "someday" — funding, hiring, leadership change, expansion, new tech adoption, regulatory change, M&A. Push if the "signal" is just a firmographic restated; a signal has to indicate timing, not fit.

## Closing — Compiled output

Once Step 7 clears the bar, the session compiles into a structured ICP document covering:

1. **Problem** — the pain statement
2. **Where It Exists** — the conditions/environments
3. **Operational Conditions** — the mechanism
4. **GRIPS** — Geography, Revenue, Industry, Person, Size, each with the reasoning, not just the value
5. **Anti-ICP** — named exclusions with failure-mode reasoning
6. **Tiers** — Tier 1/2/3 with what's in each and why
7. **Buying Signals** — timing indicators, separate from fit criteria
8. **Open Gaps** — anything the person pushed past without a sharp answer, named honestly so it doesn't get treated as settled (omit this section entirely if there aren't any)

Write it up as a clean markdown document with one heading per section, filled with this session's real answers, not the generic framework examples. Put the single sharpest line from the session (usually the Step 1 pain statement or the Tier 1 definition) at the top as a callout.

## Tone

Direct, no filler, no em dashes, no adverbs, active voice. Say "that's not specific enough, here's why" plainly instead of softening it into a question. This is a stress-test session, not a survey — the value of the output is a direct function of how much resistance the answers actually had to survive.

# ICP Mapping Framework: From Problem → Market → Ideal Customer Profile

## Core Principle

An ICP is not defined by who exists in the market.

It is defined by:

> Who has the strongest problem, the highest ability to benefit, and the lowest friction path to becoming a successful customer.

The goal is not to find everyone who *could* buy.

The goal is to identify who is most likely to buy, succeed, retain, and expand.

---

# ICP Framework Overview

## The Flow

Problem
↓
Where the Problem Exists
↓
Operational Conditions
↓
Firmographics (GRIPS)
↓
Anti-ICP Filtering
↓
ICP Tiers
↓
Buying Signals

---

# Step 1: Understand the Problem

Ignore features.

Start with:

"What problem does the product/service solve?"

Questions:

- What pain exists?
- What happens if they do nothing?
- What are they currently doing instead?
- Why is the current solution insufficient?
- What triggers them to search for a solution?

The objective:

Understand the pain, not the product.

---

# Step 2: Identify Where the Problem Exists

Ask:

"Where does this problem naturally occur?"

Do not start with industries.

Start with conditions.

Questions:

- What companies experience this problem?
- What environments create this problem?
- At what stage does the problem become painful?
- What complexity causes this problem?

Example:

AI note-taking software

Bad approach:

"Let's target SaaS companies."

Better approach:

"Let's target companies where meeting volume, customer conversations, and knowledge transfer become operational bottlenecks."

Potential markets:

- SaaS
- Consulting
- Recruiting
- Agencies
- Financial services

The market emerges from the problem.

---

# Step 3: Operational Characteristics

Operational traits explain WHY the problem exists.

Ask:

"How does this company operate that creates this pain?"

Examples:

## Professional Services Automation

Operational traits:

- Billable hours
- Project-based delivery
- Resource allocation
- Multiple teams
- Utilisation targets

Pain:

Poor visibility into capacity and profitability.

---

## Facial Recognition Software

Operational traits:

- Multiple physical locations
- Controlled access requirements
- High employee turnover
- Contractor workforce
- Security requirements

Pain:

Access management complexity.

---

## AI Sales Assistant

Operational traits:

- High meeting volume
- Sales-led motion
- Large sales team
- CRM dependency
- Customer-facing roles

Pain:

Lost information, poor CRM hygiene, lack of visibility.

---

# Step 4: Firmographics (GRIPS)

GRIPS turns the ICP into something searchable.

## G — Geography

Questions:

- Does location affect the problem?
- Does regulation matter?
- Does language matter?
- Does purchasing power vary?

---

## R — Revenue

Questions:

- Can they afford the solution?
- At what revenue stage does the pain become expensive?
- Does revenue correlate with complexity?

Revenue is a proxy for buying power, not the actual reason they buy.

---

## I — Industry

Questions:

- Which industries naturally experience this problem?
- Which industries operate under the conditions required?

Industry is discovered through the problem.

---

## P — Person

Questions:

- Who owns the problem?
- Who experiences the pain?
- Who influences the purchase?
- Who signs off?

The user, champion, and economic buyer may be different people.

---

## S — Size

Questions:

- At what headcount does this become painful?
- When does complexity increase?
- When does manual work stop scaling?

---

# Step 5: Define the Anti-ICP

Anti-ICP is not the opposite of ICP.

It is the explanation of who should NOT be targeted.

Ask:

- Who does not experience enough pain?
- Who cannot justify the cost?
- Who lacks the complexity required?
- Who will likely churn?
- Who requires excessive support?
- Who already has an alternative?

Example:

ICP:

50-500 employee SaaS company with sales-led growth.

Anti-ICP:

- Very early startups with no sales process
- Large enterprises requiring complex procurement
- Product-led companies with minimal sales activity

---

# Why Anti-ICP Matters

The market is full of noise.

A company can:

- Have the problem
- Understand the problem
- Like the solution

And still not be a good customer.

The anti-ICP protects focus.

---

# Step 6: ICP Tiering System

## Tier 1 — Core ICP

Highest confidence customers.

Characteristics:

- Strong pain
- Clear ROI
- High likelihood of success
- Good buying process
- Strong retention potential

Action:

Prioritise all GTM efforts.

---

## Tier 2 — Expansion ICP

Potential customers requiring validation.

Characteristics:

- Similar problem
- Less certainty
- Different industry
- Different operating model
- Longer sales cycle

Action:

Experiment when capacity allows.

---

## Tier 3 — Non-ICP

Low probability customers.

Characteristics:

- Weak pain
- Poor fit
- Low urgency
- Poor economics
- High friction

Action:

Do not proactively target.

---

# Buying Signals Are Separate

Signals do not define the ICP.

They identify timing.

ICP answers:

"Who?"

Signals answer:

"Who now?"

Examples:

- Funding
- Hiring
- New leadership
- Expansion
- New technology adoption
- Regulatory changes
- M&A activity

---

# ICP Validation Metrics (Later Stage)

Do not include LTV/CAC when creating an initial ICP.

Early stage companies lack reliable data.

Use them after validation.

Metrics:

- Conversion rate
- Sales cycle
- Retention
- Expansion
- CAC
- LTV

These refine the ICP.

They do not create the initial hypothesis.

---

# Final ICP Mental Model

A strong ICP is built by answering:

1. What problem do we solve?
2. Where does this problem naturally exist?
3. What operational conditions create this problem?
4. How do we identify these companies?
5. Who should we exclude?
6. Which segments deserve priority?

The best ICP is not the biggest market.

It is the segment where:

- the pain is strongest,
- the solution creates measurable value,
- buying friction is manageable,
- and success is repeatable.