Four terms get used interchangeably in most B2B teams. ICP. Buyer persona. Target account. Signals.
They mean different things. Using them wrong is why segmentation is a mess in most CRMs, why messaging misses, and why "we know our buyer" doesn't translate to pipeline.
Here's what each one is and how they fit together.
Quick definitions
- ICP (Ideal Customer Profile). The profile of the company you sell to best.
- Buyer Persona. The profile of the person who buys within those companies.
- Target Account. A specific, named company you're actively working.
- Signal. An observable event indicating a company or person is interested, in-market, or ready.
Each one solves a different targeting problem. Read on.
What is an ICP?
The ICP is a description of the company you sell to best.
What's in it: firmographics (industry, size, geography, revenue, funding stage), technographics (tools they use), and situational factors (recent hires, expansions, category shifts).
What it's for: filtering the market. Deciding who's addressable and who isn't. Everything downstream (accounts, personas, campaigns) inherits from the ICP.
What it's not: a job title. A single company. A "sweet spot" you feel about.
Example: "US-based B2B SaaS, 50 to 500 employees, Series A or B, using HubSpot, recently hired a Head of RevOps." That's an ICP.
Most "ICPs" we audit are actually three or four different ICPs stapled together. Tighten this and everything downstream gets easier.
When to update: every 6 to 12 months, or whenever your best-fit customer segment shifts.
What is a Buyer Persona?
A buyer persona is a description of the person who buys within your ICP companies.
What's in it: role, seniority, responsibilities, pain, buying authority, day-to-day workflow, decision criteria.
What it's for: shaping messaging, sequences, and offers. Different personas need different framings, even for the same product.
What it's not: an ICP (that's the company). A fictional avatar with a stock photo (that's cosplay). "The CEO" as a persona (too broad).
Example: "Head of RevOps at a Series B SaaS company. Reports to CRO. Owns HubSpot admin. Primary pain is broken lead routing between marketing and sales. Spends 40% of the week in reporting." That's a persona.
One ICP company usually has 3 to 5 personas involved in the buying decision. Message each one differently or lose the deal.
When to update: every 6 to 12 months, or when the buyer's context changes.
What is a Target Account?
A target account is a specific, named company you've decided to sell to. Not a category. A proper noun.
What's in it: company name, tier (T1/T2/T3 by priority), buying committee mapped by role, current stage of engagement.
What it's for: focused outbound and ABM. When you're running plays on 100 to 500 named accounts, not the whole market.
What it's not: an ICP (that's the category). A persona (that's the person). A CRM record you haven't done research on.
Example: "Stripe, Notion, Ramp, Vanta." Four target accounts inside the ICP "US B2B SaaS, Series B+."
Target accounts only matter when there's a play running on them. A CRM full of "target accounts" that get one cold email and no follow-up isn't ABM. It's a list.
When to update: quarterly at minimum. Accounts stop being targets when they buy from a competitor or no longer fit the ICP.
What is a Buying Signal?
A signal is an observable event that indicates a company or person is interested, in-market, or ready to buy.
What's in it: the event itself (website click, email reply, hiring update, product usage), the account and contact it maps to, and its strength, scope, and recency.
What it's for: telling you when to act on your ICP, persona, or target account. Without signals, you're guessing at timing.
What it's not: a lead. A conversion. A guarantee. Signals indicate readiness. They don't confirm it.
Signal categories:
- Inbound. Website visits, content downloads, form fills, email replies.
- Outbound. Positive replies to cold outreach, meeting requests, LinkedIn engagement.
- Account. Hiring updates, funding announcements, leadership changes, tech stack shifts.
- Product. Feature usage, trial activity, integration installs.
Example: "Tier 1 target account viewed pricing page 3 times in 48 hours, and Head of RevOps (the DICE Champion) accepted the LinkedIn connection request yesterday." Two signals, correlated, high strength, high recency.
Signals don't cause action on their own. They're inputs. You need a trigger (a rule that combines signals into a decision) and an action (what happens next). This is Layer 2 of the AllBound Blueprint — signals get captured, prioritized, then routed through triggers into actions.
Also Read: Signal-Led Outbound: How B2B Teams Can Find Buyers Who Are Actually in Market
How they fit together
The four concepts stack, in order.
- ICP defines the addressable market inside your total market.
- Target Accounts are the specific companies inside the ICP you're actively working.
- Buyer Personas are the people inside those accounts who influence or decide.
- Signals tell you when any of the above is worth acting on.
Miss any layer and the ones above break. No ICP, target account list is random. No personas, messaging is generic. No signals, timing is always wrong.
The 5 mistakes we see repeatedly
1. Treating "ICP" as job title. ICP is a company profile. "VP Sales" is a persona, not an ICP.
2. Personas built from feelings, not from buyer conversations. Talk to 10 real buyers before you finalize the persona. Skip demographics. Focus on how they actually make decisions.
3. Target account lists built from firmographic filters and never validated. A 3,000-account list isn't a target account list. That's a filtered database. Real target account lists are 100 to 500 companies, scored, prioritized, and actively worked.
4. Tracking signals with no triggers. Most CRMs capture signals (page visits, email opens). Almost none convert them into decision rules. Signals sitting in a report are useless. They need to fire an action.
5. Confusing "our ICP" (companies that fit our profile) with "our best customers" (companies who bought). Not always the same set. Your best customers are the sample you learn from. Your ICP is the pattern you extract.
Get the segmentation right first
Bad segmentation breaks everything downstream. Outbound falls flat. Sequences underperform. ABM programs miss the accounts that would have converted. Signals fire, but no one acts on them.
Getting these four right, and connecting them, is the foundation of any GTM system that scales past feast-or-famine pipeline.
Work with Leadle
If your ICP is too broad, your personas are guesses, your target account list is a filtered database, or your signals aren't converting to pipeline, we can help.
FAQs
What's the difference between an ICP and a target account?
ICP is the profile of the type of company you sell to best. Target account is a specific, named company that fits the ICP and you're actively working. ICP is the category. Target account is the company. You need the ICP first to build a target account list that isn't random.
Can you have multiple ICPs?
Yes, but only if you have separate GTM motions for each. If sales, marketing, and outbound all run the same play, pick one ICP. Two ICPs with one motion means neither gets served well.
How many buyer personas should a B2B company have?
One per role that meaningfully influences the buying decision. Most mid-market B2B companies have 3 to 5 personas per deal, covering decision maker, influencer, champion, and evaluator.
How do you build a target account list from scratch?
Start with the ICP filter. Layer signals (recent hires, funding, tech stack changes, expansion). Score accounts by fit and timing. The right size range is 100 to 500 accounts. More than that and you're not doing ABM, you're doing outbound with a slightly smaller list.
Is ICP the same as TAM?
No. TAM (Total Addressable Market) is every company that could theoretically buy. ICP is the subset of TAM you actually sell to best. TAM might be 50,000 companies. Your ICP might be 3,000.
What are the most important buying signals to track?
Depends on your motion. For outbound: hiring updates, funding announcements, leadership changes, tech stack shifts. For inbound: pricing page visits, demo requests, comparison content downloads. For product-led: feature usage, integration installs, trial activity. Signals matter only if you have a trigger that acts on them.
