How to Define ICPs and Buyer Personas for B2B Targeting

How to Define ICP and Buyer Personas for B2B Targeting

To build a B2B targeting system that improves pipeline quality, follow these 5 procedures. You will need CRM access, 24 months of closed-won and closed-lost data, a firmographic enrichment source, and an intent data feed. The full sequence takes 4 to 6 weeks. The Starr Conspiracy recommends finishing ICP definition before persona work.

Step summary

1. Define your B2B ICP from closed-won revenue concentration.
2. Audit and tier your TAM with firmographic fit scoring.
3. Build buyer personas anchored to validated ICP accounts.
4. Map the buying committee across demand states.
5. Activate intent signals against the ICP and committee map.

Personas inherit firmographic boundaries from the ICP, which is why the sequence is non-negotiable. Run this 6 to 8 weeks before quarter planning. You will leave with exportable Tier 1 lists, scoring fields, role-based messaging, a committee map, and intent rules sales agrees to. If you want definitions, go to YouTube. If you want execution, keep reading. We do not sell AI experiments. We build targeting systems that actually work.

Prerequisites / What You Need Before Starting

Each prerequisite below is verifiable, meaning a colleague could confirm you actually have it. Group them as data, tools, and people.

Data:

• 24 months of closed-won and closed-lost opportunities exportable from your CRM, with deal size, sales cycle length, and product mix.
• Firmographic enrichment for every account in your CRM, covering employee count, revenue band, industry code, tech stack, and headquarters geography.
• An intent data feed with topic-level scores refreshed at least weekly.

Tools:

• CRM admin access with the ability to build static account lists.
• Marketing automation platform (MAP) access with the ability to route by score and sync to Salesforce campaigns.

People:

• Sales leadership willing to spend 2 hours validating the draft ICP. Without this, you are building an artifact, not a targeting system.
• 4 to 6 weeks of calendar time. If you are trying to do this in a week, you are doing persona theater, not ICP definition.

If you do not yet have a documented go-to-market strategy, pause and complete that first. ICP work without GTM context produces a list, not a system.

Step 1: Define Your B2B ICP From Closed-Won Revenue Concentration

Export every closed-won opportunity from the last 24 months. Calculate revenue concentration by firmographic attribute: industry, employee band, revenue band, geography, and tech stack. Identify the smallest set of attribute combinations that produces 70% to 80% of closed-won revenue. That set is your draft ICP. 70% to 80% is a Starr operating heuristic, not an industry standard. Calibrate up if your closed-won set is small, down if it is broad.

The aspirational ICP and the actual ICP are almost never the same. We have watched this fail the same way for 25 years. If 62% of closed-won revenue came from 200 to 2,000 employee professional services firms in North America running a specific CRM, that is your ICP, regardless of what the deck says. Document it as a one-page firmographic profile with named inclusion and exclusion criteria. Exclusion criteria matter more, because they stop sales capacity from leaking into wasted SDR sequences and inflated CPL.

If sales objects that the product is new and there is not 24 months of data, use closed-won plus stage-4-plus pipeline and label the ICP "Draft, recalibrate at 12 months."

• When to use: You have at least 12 months of closed-won data and an active sales team.
• When not to use: Pre-revenue or fewer than 20 closed-won deals. Build a hypothesis ICP from win-loss interviews and revisit at 20 deals.

Governance: the demand gen leader owns the ICP, marketing ops owns the scoring fields, and sales leadership co-signs inclusion and exclusion. Recalibrate every 6 months, or sooner on product expansion, pricing change, or new market entry.

Confirm sales leadership signs off on inclusion and exclusion criteria in writing before proceeding. Output: a one-page ICP firmographic profile, feeding Step 2.

Step 2: Audit and Tier Your TAM With Firmographic Fit Scoring

Run your draft ICP against your firmographic enrichment provider to produce a total addressable account list. Build a fit score for every account using the formula: fit score equals the weighted sum of matches across industry, employee band, revenue band, geography, and tech stack. Weights are set by which attributes drove the strongest revenue concentration in Step 1. Tier 1 and Tier 2 outputs from this step define where personas must be validated first in Step 3.

Segment the list into 3 tiers:

• Tier 1: strict-match ICP, all 5 criteria met, one-to-one motion.
• Tier 2: 4 of 5 criteria met, one-to-few campaigns.
• Tier 3: industry and size fit, secondary criteria do not, one-to-many programs.

Validate segment sizes against pipeline math. If Tier 1 returns 80 accounts and sales has capacity for 400, your criteria are too tight. If it returns 12,000 and you have 8 reps, your criteria are too loose. A Starr operating heuristic for Tier 1 is 50 to 100 accounts per rep, calibrated to hit 4 to 6 qualified meetings per rep per month. Adjust down for enterprise motions, up for velocity motions.

Intent without ICP is a smoke alarm in a stadium, loud and useless. Tiering is what turns the alarm into an address.

• When to use: You have a working ICP and enrichment coverage above 80% of your CRM.
• When not to use: Enrichment coverage is below 60%. Fix data first.

If enrichment coverage is poor, run a one-time append before tiering, then reassess monthly. Marketing ops owns tier maintenance, with a quarterly re-tier or sooner on territory changes.

Confirm every Tier 1 account is exportable to CRM and MAP as a static list before proceeding. Output: a tiered TAM account list with fit scores, feeding Steps 3, 4, and 5.

Step 3: Build Buyer Personas Anchored to Validated ICP Accounts

A persona is the named decision profile of a specific role inside an ICP-fit account. Build 3 to 5 personas, one per role that materially influences the purchase decision. Validate personas first against the Tier 1 and Tier 2 accounts from Step 2.

For each persona, document the role title and common variants, the demand state they typically enter in, the metrics they are measured on, the systems they already own, the 3 most common objections, and the proof points that move them. Source this from win-loss interviews with the last 10 closed-won and 5 closed-lost deals, weighted toward economic buyers and primary users. Sample size logic: 15 interviews minimum, 3 per role across both outcomes.

If your persona came from LinkedIn job descriptions and vibes, it is fan fiction. Stop treating personas as content themes. Treat them as routing logic. Each persona must reference the ICP firmographic profile from Step 1. A VP of Marketing at a 300-person SaaS company is a different persona than a VP of Marketing at a 5,000-person manufacturer. See our buyer persona development guide for the interview script.

If titles vary wildly across segments, normalize by responsibility, not by title string. A "Head of RevOps" and a "VP of Sales Operations" are the same persona if they own the same decision.

• When to use: You have a validated ICP and access to recent buyers.
• When not to use: You cannot get 10 win-loss interviews scheduled in 30 days. Fix that bottleneck first. If sales will not do interviews, start with call recordings and CRM notes, then escalate.

Confirm each persona is tied to an ICP tier and includes 3 validated objections from interviews before proceeding. Output: 3 to 5 firmographic-anchored personas, each with objections and proof points, feeding Step 4.

Step 4: Map the Buying Committee Across Demand States

B2B purchases typically involve 5 to 11 roles, based on our work across two decades of demand gen engagements. A single-persona strategy will lose to a buying-committee strategy every time. Map the committee for your top 3 ICP segments.

For each segment, identify the following roles:

• Economic buyer
• Technical buyer
• User buyer
• Influencer (often 1 to 2 roles)
• Procurement and legal gatekeepers

For each role, document which of the Demand States, our shorthand for the Ten Demand States framework, they typically occupy at the same point in the cycle. The economic buyer is often in a different Demand State than the user buyer on the same deal. That asymmetry is where most B2B campaigns fail.

If sales says they only ever talk to 1 contact, map the other 4 anyway by role. The committee exists whether you see it or not. Yes, this feels like overkill on the first pass. Do it anyway.

• When to use: Average deal size justifies more than 1 stakeholder conversation, or sales cycles exceed 60 days.
• When not to use: Transactional, single-buyer sales under 30 days.

Confirm each committee map names at least 5 roles, assigns a Demand State to each, and specifies 1 message and 1 asset per role before proceeding. Output: a committee map per segment, feeding Step 5.

Step 5: Activate Intent Signals Against the ICP and Committee Map

Configure your intent feed to score only against the Tier 1 and Tier 2 account lists from Step 2. Build the topic taxonomy from the 3 most common objections per persona in Step 3, not from the vendor's default topic library. For example, if a persona's top objection is integration risk, map intent topics for integration platforms, middleware, and migration into that persona's signal set.

Set thresholds:

• A single mid-topic spike is a marketing touch.
• A sustained multi-topic surge across 2 weeks is a sales-qualified account.
• A surge plus a committee hiring event is an immediate sales action.

Validation checklist, run monthly:

• Compare intent-flagged accounts against a matched control group of ICP accounts with no intent signal.
• Track pipeline created within 60 days.
• Look for a 1.5x lift over control. 1.5x is a Starr operating heuristic. Your baseline may differ. Below that, recalibrate topics or thresholds.

Intent data without ICP definition is an invoice with no pipeline attached. The signal that 14,000 accounts are researching your category is noise. The signal that 22 Tier 1 accounts are researching your category, and 3 just hired a new VP in the committee, is a sales action.

Activate across CRM, MAP, ad platforms, and sales engagement. Marketing ops owns intent rule governance with a 90-day review cadence. Re-score on any change to Tier 1, persona objections, or topic taxonomy.

• When to use: You have a tiered TAM, a committee map, and SLA (service-level agreement) alignment with sales on follow-up.
• When not to use: Sales cannot work flagged accounts within 5 business days. The signal decays.

Confirm intent-flagged Tier 1 accounts produce at least 1.5x pipeline conversion versus a matched control group within 60 days. Output: intent activation rules wired into CRM, MAP, ad platforms, and sales engagement, closing the targeting system.

Common Mistakes to Avoid

In Step 1, the most common mistake is writing the ICP from a strategy deck instead of from closed-won data. The aspirational ICP and the actual ICP are almost never the same, and you will spend a year wondering why campaigns underperform.

In Step 2, teams skip exclusion criteria. Inclusion criteria define who you want. Exclusion criteria define who you refuse to sell to. Without the second list, sales capacity leaks into accounts that look right on paper and never close.

In Step 3, teams build personas without firmographic anchoring. A persona that could apply to any company size is a role stereotype, not a B2B persona. It will not change a single campaign decision.

In Step 4, teams map a single decision maker and call it a committee. Map 5 roles minimum, even if you cannot name the specific humans yet. The role map alone changes how content and outreach get sequenced.

In Step 5, the failure pattern is treating intent as a lead source. Intent is an account-prioritization signal layered on top of an ICP. Treat it as a lead source and you will fill the pipeline with accounts that will never buy from you.

If your account list is political instead of mathematical, book a working session. The Starr Conspiracy will build the ICP, tiered TAM, firmographic-anchored personas, committee map, and intent activation rules with you, wired into your CRM and MAP. Start 4 to 6 weeks before planning. Every week you wait, you train the org to trust bad targeting.

Related Questions

What is the difference between an ICP and a buyer persona in B2B?

An ICP describes the company you sell to, built from firmographics like industry, size, geography, and tech stack. A persona describes the human roles inside that company who influence the purchase. ICP comes first, because personas inherit firmographic boundaries from the ICP. See the ICP glossary entry for the full distinction. A persona without an ICP anchor is a role stereotype.

How often should a B2B company update its ICP?

Review the ICP every 6 months against the most recent 2 quarters of closed-won data. Material rewrites are usually annual, triggered by product expansion, a pricing change, or a new market entry. If closed-won revenue concentration has shifted by more than 15% across firmographic attributes, the ICP needs a rewrite, not a refresh.

How many buyer personas should a B2B company have?

3 to 5 personas per primary ICP segment. Fewer than 3 and you are not modeling the buying committee. More than 5 and you are modeling individuals rather than roles, which makes personas unusable for campaign targeting. Build the minimum number that lets you assign a distinct message and asset to each materially different decision role.

Can intent data replace ICP definition?

No. Intent data without ICP filtering produces volume, not pipeline. The accounts researching your category are not all accounts that can buy from you. Intent is the activation layer that prioritizes within your ICP, not a substitute for defining who fits. See the buyer intent data glossary entry for more.

How do I validate that my ICP is correct?

Run the ICP filter against your closed-lost pipeline. If a high percentage of closed-lost deals match your ICP, the criteria are too loose or disqualification is happening too late. The strongest validation is a win rate inside the ICP at least 2x the win rate outside it. Success at the system level looks like higher win rate inside ICP, faster cycle time in Tier 1, and meeting rates that justify the sales capacity allocated. ICP, tiered TAM, personas, committee map, and intent rules together are the targeting system. Anything less is a list.