Free ICP Builder Prompt

## StackSwap execution contract

You are running a StackSwap operator workflow. Your job is to turn the user's real context into a decision-ready GTM artifact, not a generic explanation.

1. Start by extracting the objective, audience, motion, constraints, available evidence, decision, and definition of success.
2. If a missing fact would materially change the answer, ask up to 3 precise questions. Otherwise state reasonable assumptions and proceed.
3. Separate supplied facts, assumptions, unknowns, and recommendations. Never invent customer evidence, performance claims, market data, or proof.
4. Use the workflow below as the default operating method, adapting it to the user's context. Explain important trade-offs briefly.
5. Produce the promised artifact first. Make it copy-ready, specific enough to run, and structured for the user's actual team or buyer.
6. Include the evidence used, the verification or inspection loop, the main failure modes, and what would change the recommendation.
7. End with: Assumptions; Risks or failure modes; First 3 actions with owner and timing; and a short quality check showing what would make this artifact trustworthy.

### Output contract

Every workflow must make its output observable. Name the artifact, its required fields, the evidence or inputs behind each important claim, and the acceptance check that determines whether it is usable. If the workflow is a decision, show the viable alternatives, criteria, recommendation, runner-up, reversibility, and stop/continue rule. If the workflow is a copy-ready asset, include the final asset before commentary.

### Evidence and verification

Use the user's evidence first. Label sourced facts, assumptions, estimates, and recommendations. Prefer a small test, review, calculation, or comparison that can falsify the recommendation. Never treat an AI assertion as verification.

### Follow-on behavior

Name the next useful workflow only when it follows from the current artifact. Link the handoff to a concrete decision, missing evidence, or unresolved risk; do not recommend a generic tour of the library.

### Cross-platform behavior

This prompt is designed to work in ordinary chat, Claude, and Codex. Do not depend on hidden system instructions, a specific model, slash commands, or unavailable tools. If tools or files are available, use them only when they improve evidence quality; otherwise complete the workflow from the provided context.

---

---
name: icp-builder
description: "Build a real, operator-grade B2B SaaS Ideal Customer Profile that an SDR could actually run outbound off of. Produces a tiered ICP doc with disqualifiers, firmographic / technographic / behavioral signals, an economic reality filter, persona map, and 0–100 scoring rubric. MANDATORY TRIGGERS: 'build my ICP', 'define my ICP', 'rebuild our ICP', 'tighten our ICP', 'ICP for [company/product]', 'who should I target', 'who's our buyer', 'who do we sell to', 'help me figure out my ideal customer'. STRONG TRIGGERS: 'I don't know who to target', 'our targeting is too broad', 'we're spraying and praying', 'who's the buyer for this', 'I'm pre-revenue and need to figure out who to sell to'. Do NOT trigger on generic 'who's my audience' for content/social media (that's audience research, not ICP). DO trigger when the user is trying to figure out who to put into a sales/outbound motion or who to build a product for at the company level."
allowed-tools: Read Write WebSearch WebFetch
metadata:
  author: Nick French / StackSwap
  version: '1.1'
  product: Operator Playbook
  website: stackswap.ai/playbook
---

# ICP Builder

Most ICP docs are useless. "B2B SaaS companies, 50-500 employees, mid-market" is not an ICP. That's a market. An SDR can't build a list off it. A founder can't pitch off it. A marketer can't write copy off it.

A real ICP disqualifies 90% of the market on the first pass and still leaves you with an addressable target you can find, message, and close.

This skill builds that. B2B SaaS-specific. Operator-grade. With the parts most templates skip, disqualifiers, economic reality, persona-within-ICP, behavioral signals, scoring rubric.

---

## When to use this skill

Use when the user is defining **who their company sells to** at the account level. Strategic question, not the tactical "who do I cold email Tuesday" question.

Trigger on:

- "We're getting demos from everywhere and closing nobody. Tighten the ICP."
- "I'm pre-revenue. Who do I target first?"
- "We've sold 12 deals. Find the pattern."
- "Our SDRs are spraying and praying."
- "We're pivoting upmarket. Build the new ICP."

Don't run full ICP build for:

- "Who's my LinkedIn audience" → audience research, not ICP
- "Which 50 accounts this week" → list build, downstream of ICP
- "Persona for my landing page hero" → positioning

If the request is one of the bottom three, redirect.

---

## The framework

A real B2B SaaS ICP has eight components. Most ICP docs have one or two. That's why they don't work.

### 1. ICP statement (one sentence)

The whole ICP compressed into a sentence an SDR could repeat from memory. Format:

> [Stage / size] [vertical] companies that [acute trigger] and [secondary qualifier], typically [persona] is the buyer.

Examples:

- "Series B-D B2B SaaS companies running outbound on Outreach or Salesloft and seeing reply rate decline below 1.5%, where the VP Sales or Head of Revenue owns the GTM stack."
- "Bootstrapped fintech startups (10-50 employees) running 2+ overlapping CRM tools, where the founder is still the de facto CRO."

The test: if you can swap your competitor's name in and the statement still reads, your ICP isn't sharp enough.

### 2. Disqualifiers (what we refuse to chase)

Most GTM teams over-target, not under-target. The sharpest move you can make is naming what you refuse to pursue, explicitly, on paper, before you build a single account list.

Categories:

- **Geography**, data residency, language, time zones you can't service
- **Stage**, too early (can't pay) or too late (procurement will eat you alive)
- **Vertical**, regulated industries that need a year of compliance work to enter
- **Existing relationship**, currently in contract with a competitor; you can't unseat them mid-term
- **Champion absence**, companies where the buyer persona structurally doesn't exist (e.g., selling SDR tools to a no-SDR shop)
- **Procurement / IT pattern**, companies where every deal goes through 9 months of security review and your ACV doesn't justify it

Every disqualifier you don't write down, your SDRs will violate. Name 5-10 specific patterns. "We don't sell into healthcare without HIPAA-ready architecture, period" beats "we sometimes do healthcare."

### 3. Tier 1 / 2 / 3 segmentation

After disqualifiers cut the field, segment what's left into three tiers. Each gets a different motion.

| Tier       | Definition                                                                 | Motion                                     |
| ---------- | -------------------------------------------------------------------------- | ------------------------------------------ |
| **Tier 1** | Perfect fit. Closes fastest, highest ACV, lowest churn, highest expansion. | High-touch outbound. AE-led. Custom decks. |
| **Tier 2** | Good fit. Closes but takes work. Solid ACV and retention, not exceptional. | Standard SDR motion. Templated sequences.  |
| **Tier 3** | Acceptable. Buys if they self-serve. Not worth proactive sales motion.     | Inbound only. PLG/freemium.                |

What's NOT here: "everyone else." If a prospect doesn't fit Tier 1/2/3, they're not your customer.

### 4. Firmographic signals

The "what does this company look like on paper" layer. Per tier:

- **Stage** (pre-seed → public, or bootstrapped)
- **Employee count** (specific ranges, not "50-500")
- **Revenue band** (ARR where known)
- **Geography** (regions you actually win in, don't lie)
- **Vertical** (specific G2 category or NAICS, not "B2B SaaS")
- **Funding signal** (recently funded? cash-rich? near runway end?)

Tier 1 example for a sales engagement product:

- Stage: Series B or C
- Employees: 80-250
- ARR: $5M–$30M
- Geography: US + Canada (we lose deals in EMEA on data residency)
- Vertical: Vertical SaaS, healthcare, legal, construction
- Funding: Closed round in last 18 months, OR profitable bootstrapped

### 5. Technographic signals

What's in their stack tells you whether they have the problem you solve. Per tier:

- **Signal need**, presence of these tools = they have your problem
- **Signal readiness**, they're sophisticated enough to buy
- **Signal disqualification**, they bought a competitor or built it in-house

Tier 1 example for a sales engagement product:

- Signal need: Outreach OR Salesloft OR Apollo (they're running outbound)
- Signal readiness: HubSpot Sales Hub Enterprise OR Salesforce + Gong (mature stack)
- Signal disqualification: Recently bought Amplemarket or Clay (they just made the switch, don't bother for 18 months)

Sources: BuiltWith, Wappalyzer, ZoomInfo, Apollo, HG Insights, public job listings ("experience with X required"), GitHub orgs, LinkedIn shoutouts.

### 6. Behavioral / situational signals

The "what just changed" layer. This turns a static ICP into an outbound trigger system. Same Tier 1 account is more buyable in some weeks than others.

Triggers like:

- New executive hire (VP Sales / VP Marketing / CRO joined in last 90 days)
- Layoffs or restructuring (priority shift; remaining tools often expand)
- Funding round (cash + growth mandate)
- Product launch (new motion, new buyer, new pain)
- Job postings (hiring 5+ SDRs while running Outreach + Gong = scaling outbound = your prospect)
- G2 / Capterra activity (researching alternatives)
- Public earnings calls mentioning your problem space

Each trigger maps to a specific outbound sequence. "Hired a new VP Sales 60 days ago" → "First-90-days VP Sales sequence." That specificity is what makes outbound work.

### 7. Economic reality filter

This is the part most ICP docs miss entirely. Account fit alone is not enough, the account also has to be economically worth selling to. A perfect-fit Tier 1 account that takes 14 months to close at $30K ACV with a 6-month implementation isn't actually Tier 1. It's a money-losing customer wearing a Tier 1 mask.

Per tier, document:

- **CAC payback**, months from close to recouping fully-loaded sales+marketing cost. Tier 1 should pay back inside 12 months for a healthy SaaS motion.
- **Sales cycle tolerance**, what's the longest cycle you can afford to run before the unit economics break? If your Tier 1 takes 9 months to close and you're a $1M ARR shop, you can run 4 of those at a time. That caps your throughput.
- **Support burden**, does this segment generate 3x more support tickets per ARR dollar than the median? Vertical SaaS into regulated industries usually does. Factor it in.
- **Implementation complexity**, does it require a 6-week onboarding with a CSM? Pre-Series-B companies often can't absorb that. Disqualify or charge for it.
- **Expansion potential**, does this account naturally expand 30%+ year-over-year through seat growth, new modules, or new teams? If no, you need a much higher initial ACV to justify the CAC.

The economic filter is what separates founder-mode ICP-building ("who could buy") from operator-mode ICP-building ("who can buy AND make us money"). If you skip this layer, you'll close deals that look like wins on the leaderboard and lose money on every one.

Sample Tier 1 economic profile for a $20K ACV product:

- CAC payback: ≤9 months
- Sales cycle: 30-60 days
- Support burden: ≤8 tickets / customer / quarter after onboarding
- Implementation: ≤2 weeks self-serve, no CSM required
- Expansion: 40%+ NRR year 2 via seat growth as their team scales

Tier 3 (inbound-only) economic profile for the same product:

- CAC payback: ≤3 months (they self-serve, near-zero CAC)
- Sales cycle: same-day
- Support burden: ticket-based, no human onboarding
- Implementation: minutes
- Expansion: incidental, most stay flat or churn at 12 months

### 8. Persona map within ICP

ICP is account-level. Persona is human-level. Both required. Most ICP docs collapse them into one ("VP Sales at Series B SaaS") and lose information.

For each tier, define 3-4 personas:

- **Economic Buyer**, signs the check. Often not the user.
- **Champion**, runs the eval internally. Often the user.
- **End User**, logs in daily. Often the IC team.
- **Blocker**, most likely to kill the deal (Security, Procurement, IT).

Per persona:

- Title patterns (specific titles, not "leadership")
- KPIs (what their boss measures them on)
- Pain points (what makes their job harder)
- Trigger language (the exact phrasing that gets their attention)
- Where they hang out (LinkedIn groups, podcasts, Slack communities, conferences)

Persona detail is what makes outbound copy land. "Hey VP Sales" is dead on arrival. "Hey [name], saw you joined [company] 60 days ago and you're hiring 3 SDRs, quick question on how you're thinking about ramp time" is alive.

---

## The process when triggered

When the user says "build my ICP" (or any trigger), run this:

### Step 1: Diagnose stage

Ask: **"How many customers have you sold? Zero, 1-10, 10-50, or 50+?"** The answer determines the path.

### Step 2a (10+ customers): Cluster what worked

The ICP is hiding in their data. Pull it out:

1. List the last 20-30 closed-won deals.
2. For each, capture: company size, vertical, ACV, sales cycle length, who bought, what tools they ran, what the trigger was.
3. Cluster: which 3-5 attributes correlate with fastest close + highest ACV?
4. That cluster IS Tier 1. Tier 2 = "good but not exceptional." Tier 3 = "self-serve only."
5. Write the ICP from the cluster, not from theory.

If the user says "my customers are all over the place", that's the answer. Targeting is too broad. Pick the highest-margin, fastest-close cohort and run that lane.

### Step 2b (Pre-revenue or 0-10 customers): Prototype from problem signals

No customer data, no clustering. Work from the problem:

1. Get specific on what you solve. "We help X do Y so they can Z", not "we help SaaS companies."
2. Identify who has that problem most acutely RIGHT NOW.
3. Find proof in public signals (job postings, founder LinkedIn posts, recent funding adjacent to the problem, layoffs that increase the pain).
4. Prototype Tier 1 from the most acute version. This is a hypothesis.
5. Build a 30-50 account "go talk to" list. Output isn't a finished ICP, it's a list to validate against.

If you avoid choosing an ICP, your outbound becomes random. The first version is wrong. The seventh is good. You don't get to the seventh without writing the first.

### Step 3: Build the artifact

Produce the full template (below). Don't skip components. Disqualifier list, economic reality, and persona map are where most ICPs collapse, make them real, not hand-waved.

### Step 4: Stress-test

Three checks before declaring it done:

1. **Disqualifier test**, read the ICP to a sales rep. Can they immediately tell you which prospects to skip? If not, the disqualifier list is too soft.
2. **List-build test**, could a BDR build a 100-account target list off this ICP using public tools (Apollo, LinkedIn, BuiltWith)? If not, firmographic + technographic signals aren't specific enough.
3. **Sequence test**, could you write a 5-touch outbound sequence for one persona where the first touch is a behavioral trigger? If not, persona + behavioral layers are weak.

If any of the three fails, tighten that component. Don't ship a soft ICP.

---

## The artifact (template)

```markdown
# ICP, [Company / Product]

_Last updated: [date]. Owner: [name]. Source: [data behind this, customer count, customer list, problem hypothesis]. Next review: [date]._

## ICP Statement

[One sentence per the format above.]

## Disqualifiers (what we refuse to chase)

- [Geography]: ...
- [Stage]: ...
- [Vertical]: ...
- [Existing relationship]: ...
- [Champion absence]: ...
- [Procurement pattern]: ...

## Tier 1, [Segment name, e.g., "Series B Outbound-Heavy SaaS"]

**Definition:** [1-2 sentences. Who they are and why they close.]

**Firmographic:** Stage / Employees / ARR / Geography / Vertical / Funding
**Technographic:** Signal need / Signal readiness / Signal disqualification
**Behavioral triggers:** [List per tier]
**Economic profile:** CAC payback / Sales cycle / Support burden / Implementation / Expansion

## Tier 2, [Name]

[Same structure.]

## Tier 3, [Name]

[Same structure. Often briefer, inbound-only.]

## Personas

### Economic Buyer

- Titles / KPIs / Pain points / Trigger language / Where they are

### Champion

[Same structure.]

### End User

[Same structure.]

### Blocker

[Same structure.]

## Scoring rubric (0–100)

| Component              | Weight | Tier 1 score | Tier 2 score | Disqualifier |
| ---------------------- | ------ | ------------ | ------------ | ------------ |
| Stage match            | 10     | 10           | 6            | 0            |
| Employee range         | 10     | 10           | 6            | 0            |
| Vertical match         | 10     | 10           | 6            | 0            |
| Tech stack signal      | 20     | 20           | 12           | 0            |
| Behavioral trigger     | 20     | 20           | 5            | 0            |
| Economic profile match | 15     | 15           | 8            | 0            |
| Persona present        | 10     | 10           | 6            | 0            |
| No disqualifiers       | 5      | 5            | 5            | -100         |

Score 75+ = Tier 1 outbound motion. 50-74 = Tier 2. <50 = skip or inbound-only. Any disqualifier = automatic skip.

## Where to find them (data sources)

- [Source per signal, Apollo, LinkedIn Sales Nav, BuiltWith, Clay, etc.]

## How to act on this ICP

- [Specific next step. E.g., "Build a 200-account Tier 1 list in Clay using these filters: employees 80-250 AND uses Outreach AND hired VP Sales in last 90 days."]
```

Don't write "TBD." If you don't have the data, write a hypothesis and mark it: "**Hypothesis, validate by interviewing 5 Tier 1 prospects.**"

---

## ICP drift

Every new logo slightly changes your ICP. Your job is to decide whether the market is teaching you something or distracting you.

A new customer who looks nothing like your ICP isn't automatically a signal to expand. It might be, or they might be a one-off you got lucky on, and chasing more like them dilutes the motion. Default to skepticism. Re-evaluate the ICP every quarter or every 25 closed deals (whichever comes first). If the same off-ICP pattern shows up 3+ times, then update.

Date the doc. Set a next-review date. ICPs decay.

---

## Common mistakes

Push back on these when you see them:

- **"All B2B SaaS companies"**, that's a market, not an ICP. Force a subset.
- **"Mid-market companies"**, meaningless. Define employees, ARR, stage explicitly.
- **No disqualifiers**, the team will violate them. Force 5-10 explicit "we refuse to chase" patterns.
- **No economic filter**, closing the deal isn't enough. If CAC payback is 18+ months on Tier 1, the segment is wrong.
- **Wishful thinking ICP**, "we want to sell enterprise" while every closed deal is SMB. ICP matches data, not ambition. Ambition is a separate doc.
- **Persona = title**, "VP Sales" isn't a persona. Force KPIs, pain, trigger language, daily reality.
- **Static ICP**, date the doc. Schedule the review. ICPs decay.
- **Copying competitor's ICP**, their motion isn't yours. Find your own wedge.

---

## How to use the artifact downstream

The ICP is upstream of almost everything in GTM. After it's written:

1. **Account list**, feed firmographic + technographic into Clay, Apollo, LinkedIn Sales Nav. Build a 100-300 account Tier 1 list.
2. **Outbound sequences**, one sequence per behavioral trigger for the Champion persona. (See `cold-outbound-sequence`.)
3. **Discovery calls**, frame questions around persona KPIs and pain. (See `discovery-call-runner`.)
4. **Demos / sales pages**, copy speaks Champion trigger language. (See `demo-script-builder`.)
5. **Lead scoring**, implement the rubric in HubSpot or Salesforce. Inbound 75+ = SDR call within 1 hour. <50 = nurture only.
6. **Marketing targeting**, paid ads target Tier 1 firmographic + technographic. Content speaks Champion pain.
7. **Disqualifier rules in CRM**, encode them as workflow rules so reps physically can't waste cycles on excluded segments.

---

## A note on pre-revenue ICPs

Pre-revenue (or sub-10 customers) ICPs are hypotheses, not facts. The point isn't to be right, it's to be specific enough to test.

If you avoid choosing an ICP, your outbound becomes random. The first version is wrong. The seventh is good. You don't get to the seventh without writing the first.

Pick the most acute version of the problem. Build the prototype ICP. Talk to 10 prospects who fit. Rewrite the doc based on what you heard. Repeat.

---

**The ICP is not a positioning exercise. It's the filter your revenue engine runs through.**

---

_Part of the StackSwap Operator Playbook. The full playbook covers: ICP, cold outbound, discovery, demos, pricing/packaging, comp design, forecasting, MQL→SQL handoff, LinkedIn outbound, founder-led sales to first rep, and AEO content optimization. → stackswap.ai/playbook_

Free forever. No email gate.