icp-definition-framework
ICP Definition Framework
The Ideal Customer Profile (ICP) defines the type of company most likely to buy, get value from, and stay with your product. Not "anyone who could use it." The specific type of company that buys fastest, churns least, expands most, and refers others. Every GTM decision downstream (targeting, messaging, pricing, sales motion, channel strategy) flows from the ICP. A wrong ICP makes everything else harder.
The principle: the ICP is derived from data, not from aspiration. Analyze your best customers. Find the patterns. Those patterns are your ICP. Don't define the ICP based on who you wish would buy. Define it based on who actually does.
ICP vs Persona vs TAM
| Concept | What it defines | Level | Example |
|---|---|---|---|
| ICP | The type of company that's the best fit | Company-level | "B2B SaaS, 50-500 employees, Series A-C, US-based, sales-led motion" |
| Buyer Persona | The type of person at an ICP company who buys | Person-level | "VP Sales, 5-15 direct reports, responsible for pipeline generation" |
| TAM | Total addressable market. Everyone who could theoretically buy | Market-level | "All B2B companies with a sales team" |
ICP is narrower than TAM. TAM is everyone who could buy. ICP is who should buy. Selling to TAM is spray-and-pray. Selling to ICP is precision.
ICP is broader than a persona. ICP defines the company. Personas define the people inside those companies. You need both, but ICP comes first.
Building the ICP from Data
Step 1: Analyze your best customers
Pull data on your top 20-30 customers by one or more of these criteria:
| Ranking criteria | Why this matters |
|---|---|
| Highest NRR (net revenue retention) | They stay and grow. The strongest signal of product-market fit |
| Fastest sales cycle | They buy quickly. Less friction = better fit |
| Highest expansion rate | They buy more over time. Product delivers increasing value |
| Lowest churn rate | They don't leave. The product is sticky for this type of company |
| Highest LTV | Combination of retention + expansion. The most valuable customer type |
| Most referrals | They actively promote you. Strongest signal of satisfaction |
Data to collect per customer:
| Data point | Where to find it | Why |
|---|---|---|
| Company size (employees) at time of purchase | CRM, enrichment | Size band targeting |
| Industry / vertical | CRM, enrichment | Vertical focus |
| Geography (HQ location) | CRM | Territory and market focus |
| Funding stage at time of purchase | Crunchbase, CRM | Stage targeting |
| ARR or revenue range | Enrichment, estimate | Revenue-based segmentation |
| GTM motion (sales-led, PLG, hybrid) | Account research | Motion matching |
| Primary use case | CS notes, onboarding records | Use case focus for messaging |
| Champion title/role | CRM contact records | Persona definition |
| How they found you (lead source) | CRM | Channel investment |
| Time to close (sales cycle) | CRM deal records | Cycle time expectations |
| ACV (annual contract value) | CRM deal records | Pricing alignment |
| Tech stack (CRM, sequencing, enrichment tools) | Job postings, enrichment | Stack-based targeting |
| Churn status | CS data | Retention validation |
| Expansion revenue | Billing data | Growth validation |
Step 2: Find the patterns
Look for concentrations in the data. The ICP is where the patterns cluster.
Pattern analysis questions:
| Question | What the answer reveals |
|---|---|
| What company size appears most often in your top 20? | Your size sweet spot. Usually a range, not a single number |
| What industries appear 3+ times? | Your core verticals. Where your product naturally fits |
| What funding stages dominate? | Your stage sweet spot. Pre-seed companies and public companies have very different needs |
| What GTM motion do most top customers run? | Sales-led customers may use your product differently than PLG customers |
| What's the most common champion title? | Your primary buyer persona |
| What's the average ACV of top customers? | Your pricing sweet spot. Where the value-to-price ratio works |
| What's the average time to close? | Your expected sales cycle. Deviations from this suggest bad fit |
| What tech stack patterns appear? | Integration requirements and displacement opportunities |
Step 3: Define the ICP
Based on the patterns, write the ICP as a set of criteria.
ICP template:
# [Company Name] Ideal Customer Profile
## Company fit
- Industry: [list of core verticals]
- Company size: [employee range]
- Revenue: [revenue range or "N/A if early stage"]
- Funding stage: [stage range]
- Geography: [regions]
- GTM motion: [sales-led / PLG / hybrid]
- Tech stack: [required integrations or tools]
## Situation fit
- Primary pain: [the problem they have that your product solves]
- Current solution: [what they're doing today, including "nothing"]
- Buying trigger: [what event makes them ready to buy now]
- Budget range: [realistic budget for this type of company]
## Persona fit (primary buyer)
- Title: [title variations]
- Seniority: [level]
- Department: [function]
- Reports to: [who they answer to]
## Anti-ICP (explicit exclusions)
- Companies below [size] employees (too small to need this)
- Companies above [size] employees (require enterprise sales we can't support)
- Industries: [excluded verticals]
- Companies with no [required condition: "no sales team" / "no CRM"]
- Competitors
- Agencies / consultants (unless explicitly serving them)
Step 4: Validate with win/loss data
After defining the ICP, test it against your win/loss data.
| Test | How to run it | What it tells you |
|---|---|---|
| Win rate by ICP fit | Score all closed-won and closed-lost deals on ICP fit. Compare win rates | ICP-fit deals should win at 2-3x the rate of non-ICP deals |
| Sales cycle by ICP fit | Compare average sales cycle for ICP vs non-ICP deals | ICP deals should close 30-50% faster |
| ACV by ICP fit | Compare average ACV for ICP vs non-ICP deals | ICP deals should have equal or higher ACV |
| Churn by ICP fit | Compare churn rates for ICP vs non-ICP customers | ICP customers should churn at < 50% the rate of non-ICP |
| Expansion by ICP fit | Compare expansion rates | ICP customers should expand more frequently |
Validation rules:
- If ICP-fit deals don't win at a meaningfully higher rate than non-ICP deals, the ICP definition isn't discriminating. It's either too broad or the criteria are wrong. Revisit
- If ICP-fit customers churn at similar rates to non-ICP, your product may deliver value broadly but you haven't identified the segment where it's stickiest. Dig deeper into churn reasons
- Run this validation on at least 50 deals (25 won, 25 lost). Fewer than that and the patterns aren't statistically meaningful
ICP Dimensions
Firmographic dimensions (company-level)
| Dimension | How to define | Example |
|---|---|---|
| Company size | Employee count range. Use bands, not exact numbers | 50-500 employees |
| Revenue | ARR or revenue range | $5M-$50M ARR |
| Industry | Specific verticals, not "all B2B" | B2B SaaS, fintech, healthtech |
| Geography | HQ location. Where you can sell and support | United States, UK, DACH |
| Funding stage | Investment stage as a proxy for maturity and budget | Series A through Series C |
| Company age | Years since founding | 3-10 years old |
| Growth rate | Headcount or revenue growth trajectory | > 20% YoY headcount growth |
Technographic dimensions (stack-level)
| Dimension | How to define | Example |
|---|---|---|
| CRM | Which CRM they use | Uses HubSpot or Salesforce |
| Adjacent tools | Tools in your integration ecosystem | Uses Outreach, Salesloft, or similar |
| Competing tools | Tools you'd replace | Currently using [competitor] or spreadsheets |
| Infrastructure | Technical requirements for your product | Runs cloud infrastructure (AWS, GCP) |
Behavioral dimensions (situation-level)
| Dimension | How to define | Example |
|---|---|---|
| GTM motion | How they sell | Sales-led with SDR team |
| Team structure | Relevant team composition | Has 5-20 person sales team |
| Current pain | The problem they're experiencing | Reply rates below 5% on outbound |
| Buying trigger | What makes them ready to buy | Scaling outbound after raising Series B |
| Decision speed | How fast they buy | Can make a tool decision in < 60 days |
ICP by Company Stage
The ICP shifts as your company grows. What works at $1M ARR is different from what works at $20M ARR.
Pre-PMF (< $1M ARR)
| Dimension | Approach |
|---|---|
| How to define ICP | From conversations with early customers and prospects. Qualitative, not quantitative |
| ICP precision | Low. You're still learning who your best customer is. Define loosely, learn fast |
| Segments | 1 segment. Don't try to serve multiple ICPs before PMF |
| Validation | "Are these customers getting value and staying?" is the only question that matters |
| Common mistake | Defining the ICP based on who you want to sell to, not who actually gets value |
Post-PMF ($1-10M ARR)
| Dimension | Approach |
|---|---|
| How to define ICP | From data. Analyze top 20-30 customers per the framework above |
| ICP precision | Medium. Clear firmographic criteria. Starting to add behavioral and technographic dimensions |
| Segments | 1-2 segments. Primary ICP + one adjacent segment to test |
| Validation | Win/loss analysis quarterly. Churn-by-fit analysis |
| Common mistake | Broadening the ICP too early because "we can sell to anyone" |
Scale ($10M+ ARR)
| Dimension | Approach |
|---|---|
| How to define ICP | From advanced analytics. Predictive scoring. Cohort analysis by ICP fit |
| ICP precision | High. Multiple sub-segments with segment-specific messaging, pricing, and sales motions |
| Segments | 2-4 segments. Each with its own ICP definition, persona map, and GTM playbook |
| Validation | Quarterly cohort analysis. Predictive models. ML-assisted ICP scoring |
| Common mistake | Not updating the ICP as the product and market evolve. The ICP from $5M ARR may not apply at $30M |
Common ICP Mistakes
| Mistake | Why it happens | Why it's wrong | Fix |
|---|---|---|---|
| "Our ICP is all B2B companies" | Founders want a large TAM for investors | No focus = no targeting = no messaging = no results | Narrow to the specific type of B2B company that buys fastest and churns least |
| ICP based on aspiration, not data | "We want to sell to enterprise" but all wins are mid-market | Aspiration creates misalignment between where the product works and where you're selling | Analyze closed-won data. ICP = who actually buys, not who you wish would |
| ICP defined once and never updated | Created at founding. Never revisited | Markets shift. Products evolve. The ICP from 2 years ago may not be the ICP today | Review and validate quarterly |
| ICP too broad (10 verticals, all company sizes) | Fear of missing opportunities | Can't write targeted messaging, can't build relevant content, can't prioritize accounts | Narrow to 2-3 verticals and a specific size range. Test. Expand only when data supports it |
| ICP too narrow (one vertical, one title, one city) | Over-optimization from small sample | TAM is too small to sustain growth. Pipeline dries up | Start narrow, but not microscopic. Ensure the ICP contains at least 1,000-5,000 target accounts |
| No anti-ICP | Only defined who to target, not who to exclude | Sales wastes time on bad-fit deals. Marketing generates bad-fit leads | Define explicit exclusions: company sizes to avoid, industries to skip, situations where you don't add value |
| ICP = TAM in the pitch deck, different ICP for GTM | Sales targets a narrow ICP. Pitch deck claims a $50B TAM | Internal misalignment. Marketing optimizes for TAM breadth. Sales needs ICP depth | TAM is for investors. ICP is for the team. Keep both, use each in the right context |
Using the ICP
Where ICP criteria should be embedded
| Function | How ICP is used | Example |
|---|---|---|
| List building | Filter prospect lists to ICP criteria only | Sales Nav search: 50-500 employees + SaaS + US + Series A-C |
| Lead scoring | ICP fit score as a major scoring dimension | Company size in range = +15 points. Industry match = +15 points |
| Lead routing | Route ICP-fit leads to sales. Route non-ICP to nurture or disqualify | Workflow: if ICP_Tier = 1, route to AE. If ICP_Tier = 3, route to nurture |
| Messaging | Tailor emails, ads, and content to ICP challenges | Cold email references "Series B SaaS teams scaling outbound" (ICP-specific) |
| Content | Create content for ICP's specific problems and questions | Blog post: "How Series B SaaS Teams Rebuild Outbound After the First SDR Hire" |
| ABM | Select ABM accounts from ICP-fit companies with signals | Tier 1 = ICP fit + funding signal + hiring signal |
| Product | Prioritize features that ICP companies need most | ICP uses HubSpot → prioritize HubSpot integration |
| Pricing | Set pricing at a level ICP companies can afford and will pay | ICP is 50-500 employees → pricing at $49-99/seat fits their budget |
ICP operationalization rules
- The ICP should be in the CRM. Create an "ICP Tier" field on the Account/Company object. Score and categorize every account: Tier 1 (ideal), Tier 2 (good), Tier 3 (acceptable), Not ICP
- Score ICP fit automatically. Use enrichment data (company size, industry, funding, geography) to auto-score ICP fit when a lead enters the system. Don't rely on reps to manually assess fit
- Everyone on the GTM team should be able to describe the ICP in one sentence. "Series A-C B2B SaaS companies with 50-500 employees and a sales-led GTM motion in the US." If the team can't say this consistently, the ICP isn't clear enough
- Review quarterly. The ICP is not carved in stone. As the product evolves, as the market shifts, and as the customer base matures, the ICP should be revisited and refined
The ICP Document
What the document contains
Every B2B SaaS company should have a written ICP document that is:
- Shared with marketing, sales, CS, and product
- Referenced in every list build, campaign design, and ABM planning session
- Updated quarterly with new data
Document template
# [Company] Ideal Customer Profile
## Last updated: [Date]
## Validated against: [N] closed-won deals, [N] churned customers
### Company fit
[Firmographic criteria with ranges]
### Situation fit
[Behavioral criteria: pain, trigger, current solution]
### Persona fit
[Primary buyer persona: title, seniority, function]
### Anti-ICP
[Explicit exclusions with reasons]
### ICP scoring
[How ICP fit is scored in CRM: which fields, which values, which thresholds]
### Validation data
[Win rate by ICP fit. Churn by ICP fit. Sales cycle by ICP fit]
Measurement
| Metric | Definition | Target | Frequency |
|---|---|---|---|
| % of pipeline from ICP-fit accounts | Pipeline from Tier 1-2 accounts / total pipeline | > 70% | Monthly |
| Win rate by ICP tier | Close rate for Tier 1 vs Tier 2 vs non-ICP | Tier 1 should be 2x+ non-ICP | Quarterly |
| Sales cycle by ICP tier | Average days to close by ICP tier | Tier 1 should be 30-50% faster | Quarterly |
| Churn by ICP tier | Churn rate for Tier 1 vs Tier 2 vs non-ICP | Tier 1 should churn < 50% of non-ICP rate | Quarterly |
| NRR by ICP tier | Net revenue retention by tier | Tier 1 > 120%. Non-ICP < 100% | Quarterly |
| ICP coverage | % of target accounts in your TAM that you've contacted | Track | Quarterly |
Anti-Pattern Check
- "Our ICP is all B2B SaaS." Too broad. Which stage? Which size? Which GTM motion? Which geography? Narrow until the ICP describes a specific type of company, not a market
- ICP defined by the founder's intuition with no data. "I think enterprise fintech is our ICP" is a hypothesis. Analyze your best customers. The data may agree or disagree. Trust the data
- ICP hasn't been updated in 18 months. Your product has shipped 50 features since then. Your customer base has evolved. Three new verticals have adopted. The ICP from 18 months ago is stale. Refresh quarterly
- No anti-ICP. Every company defines who to target. Few define who to exclude. Without explicit exclusions, sales wastes time on companies that will never close or will churn within 6 months. Define the anti-ICP
- ICP is a PDF on a shared drive nobody reads. The ICP should be in the CRM (as a scoring field), in the list building criteria, in the lead scoring model, and in the SDR training deck. If it's not operationalized, it's just a document
- Too many ICP segments at early stage. A $3M ARR company with 4 ICP segments, each with its own messaging and sales motion, is spread too thin. 1 segment at pre-PMF. 1-2 at $1-10M. 2-4 at $10M+. Match the number of segments to your team's capacity
- ICP based on where you want to be, not where you are. "We're building for enterprise" but all your wins are 50-person startups. The ICP is where the product works today, not where you hope it works in 2 years. Serve today's ICP. Expand when the product earns it
- ICP defined at the person level, not the company level. "Our ICP is a VP of Sales" is a persona, not an ICP. The ICP defines the company. The persona defines the person at that company. Both are needed. Don't conflate them