Order queue open · Verified lists delivered within 24–48 hours
ICP definition guide: the 6 dimensions of an ideal customer profile, how to validate from closed-won data, a scoring template, and how to map ICP criteria to Quarvio filters.
Priya Nair
B2B growth marketer, ex-Apollo user · Updated June 24, 2026
Last updated: June 2026 · Priya Nair, B2B growth marketer
TL;DR — 5 things to know before reading
Most ICP definitions I see in B2B companies are written by a product or marketing team based on who they think the product is for. The result is an ICP that describes a plausible customer rather than an actual customer pattern derived from real purchase data.
The difference between a hypothetical ICP and a data-derived ICP shows up immediately in outbound performance. A team targeting their hypothetical ICP might be generating 4% reply rates and booking one meeting per 25 sequences sent. A team targeting a data-derived ICP — one validated against 30 closed-won deals — often sees reply rates two to three times higher because they are targeting companies that genuinely match the pattern of companies that buy.
This guide covers how to build a data-derived ICP from scratch, score it, and translate it into a filtered contact list.
These two frameworks are frequently used interchangeably, which produces targeting errors that compound through the entire outbound motion.
An ICP describes a type of company. It answers: which organizations are most likely to buy our product, retain it at high rates, and expand their use over time? The ICP is firmographic: industry, size, geography, maturity stage, technology environment.
A buyer persona describes an individual within that company. It answers: who within the ICP company is most likely to initiate, champion, or approve the purchase? The persona is role-based: title, seniority, functional accountability, information sources, evaluation criteria.
Both are necessary. The ICP tells you which companies to target. The persona tells you which person within those companies to contact. Targeting the right persona at the wrong company wastes the outreach. Targeting the right company profile at the wrong individual stalls at the wrong level.
The sequence is ICP first, persona second. Define which companies, then define who within those companies.
Step 1: Pull 20 to 30 recent closed-won deals. If you have fewer than 20 closed-won deals, use all of them. You need a pattern sample, not a statistically representative dataset. Pull from your CRM or deal tracking system.
Step 2: Document the firmographic attributes of each company. For each closed-won deal, record:
Step 3: Find the pattern. Look across your 20 to 30 deals. Which industry appears in 60% or more of them? What company size range do most fall within? What triggers — hiring events, funding, new leadership, compliance deadlines — preceded the conversation in the majority of cases?
Step 4: Document the ICP pattern. The ICP is not the average of all your deals. It is the pattern that appears most consistently across your best deals. “Best” is defined by average contract value, retention rate, and expansion revenue. Your ICP should reflect the profile of customers who buy, stay, and grow — not the profile of customers who buy once and churn.
The most specific ICP definition wins. “Financial services” is a broad category that includes retail banking, insurance, asset management, fintech, and credit unions — segments with different regulatory environments, buyer personas, and purchasing processes. An ICP that specifies “regional commercial banks with $500M to $5B in assets” is specific enough to be operationally useful.
Industry specificity enables copy specificity. When you know you are targeting compliance teams at regional commercial banks, you can open cold email with BSA/AML monitoring references that are immediately credible. When you are targeting “financial services” broadly, your copy has to be generic enough to apply to all of them, which makes it specific enough for none.
Company size (employee headcount, revenue, or both) is a reliable ICP filter because company size determines:
For most B2B products, company size correlates directly with deal size and retention rate. Identifying the size band where your closed-won deals cluster is one of the most reliable ICP dimensions to derive from data.
Geography affects ICP through regulatory environment, labor markets, and sales motion. A US-centric product selling into GDPR-regulated European markets faces different compliance requirements than the same product sold domestically. A solution requiring significant onboarding support may have a geographic constraint if implementation capacity does not scale to certain markets.
For most outbound campaigns, geography is a filtering dimension rather than a behavioral one — you target the geography where your sales motion works, not where your product could theoretically work.
This is the bridge between ICP and buyer persona. The ICP defines the company; title and seniority define the entry point for the conversation. The relevant questions are:
For a sales tool, the initiator is often a sales operations manager; the approver is the VP of Sales or CRO; the champion is the SDR team lead. Your outbound should target the initiator and the approver, since the champion rarely purchases independently.
Technology stack is an ICP dimension when your product integrates with, replaces, or complements specific tools. If your product requires a Salesforce integration, companies running HubSpot as their CRM are a lower-quality ICP fit regardless of other attributes. If your product replaces a point solution that larger companies use, the presence of that point solution is an ICP indicator.
Technology stack data is harder to obtain from a contact list but can be researched for high-priority accounts using tools that surface technology adoption signals.
The pain trigger is the specific event or condition that causes a company to move from “might evaluate this someday” to “evaluating this now.” Common triggers include:
Identifying the trigger in your closed-won data tells you what event to look for in outbound targeting. Outreach timed to a relevant trigger outperforms cold outreach that arrives with no contextual reason for the conversation.
A scored ICP allows you to rank companies from most to least likely to buy. The scoring system is simple: each dimension has a maximum score, and you assign a score based on how closely the target company matches your ICP criteria.
| Dimension | Max score | How to score |
|---|---|---|
| Industry match | 30 | 30 = exact sector match; 15 = adjacent sector; 0 = different sector |
| Company size | 20 | 20 = within ideal headcount range; 10 = one tier outside range; 0 = two or more tiers outside |
| Geography | 15 | 15 = primary target market; 8 = secondary market; 0 = unsupported market |
| Title match | 20 | 20 = ideal title; 10 = adjacent title; 0 = wrong function |
| Technology stack | 10 | 10 = confirmed compatible stack; 5 = likely compatible; 0 = incompatible |
| Pain trigger present | 5 | 5 = known active trigger; 0 = no trigger signal |
Total: 100 points. Companies scoring above 70 are priority outbound targets. Companies between 50 and 70 are secondary targets. Companies below 50 are outside ICP for this campaign.
According to Woodpecker’s 2025 cold email benchmark study, top-quartile cold email senders achieve reply rates of 15 to 20% compared to an 8.5% average. The primary differentiator is targeting quality — senders who apply a structured ICP filter before building their outreach lists consistently outperform those who target broad categories.
Quarvio contact lists are filterable by the firmographic dimensions that make up most ICP definitions. The translation from ICP criteria to Quarvio filter is direct:
| ICP dimension | Quarvio filter | Notes |
|---|---|---|
| Industry | Industry category | Filter to specific sectors rather than broad categories |
| Company size | Employee count | Select the headcount range that matches your ICP |
| Geography | Country and region | Select primary target markets |
| Title | Job title | Filter by specific titles and title variants |
| Seniority | Seniority level | Filter by C-suite, VP, Director, or Manager |
A verified buyer on sales engagement platforms on G2 described the ICP targeting improvement:
“Once we defined an actual ICP from our closed-won data and used it to filter our contact lists, our campaign reply rates nearly doubled. The list was smaller but every contact matched a pattern we knew converted.”
— Verified buyer on sales engagement platforms on G2
Unused credits are valid for 12 months, which means you can purchase an ICP-filtered list at the start of a campaign cycle and use remaining credits on a follow-on initiative without repurchasing.
ICP definition informs every element of your outbound copy. Once you know the industry, company size, and role you are targeting, you can:
The ICP does not just tell you who to contact. It tells you what to say. Instantly handles the sequence delivery, inbox rotation, and A/B testing; Inframail provides the warmed infrastructure. The quality of what lands in the prospect’s inbox is determined by how well your ICP definition informs the copy.
For LinkedIn outreach to the same ICP, Aimfox manages connection campaigns and follow-up messaging that run alongside your email sequences.
| Need | Tool | Notes |
|---|---|---|
| ICP-filtered contact lists | Quarvio | Industry, title, size, geography filters |
| Email inboxes | Inframail | Microsoft 365 inboxes, auto DNS setup |
| Sequence delivery and A/B testing | Instantly | Warmup, inbox rotation, reply tracking |
| LinkedIn outreach | Aimfox | Connection campaigns alongside email sequences |
What is the minimum number of closed-won deals needed to derive an ICP?
Twenty deals is the practical minimum for pattern recognition; fewer than that makes it difficult to distinguish ICP signal from random variation. If you have fewer than 20 closed-won deals, supplement the analysis with sales conversations that progressed significantly but did not close — the companies where the problem resonated but purchase timing was wrong are nearly as informative as closed-won deals. If you are pre-revenue, derive your ICP hypothesis from competitive customer case studies and customer interviews, then validate it with your first 20 customers.
How often should an ICP be updated?
Annually for most B2B companies, and any time you notice a significant shift in your closed-won pattern. If your last three months of closed-won deals cluster in an industry segment that did not appear prominently in your original ICP, that is a signal to investigate whether the ICP needs updating. ICP drift happens in both directions: sometimes companies outside your original ICP start buying at high rates; sometimes companies inside your original ICP stop converting. Quarterly closed-won analysis takes one hour and catches drift early.
Is an ICP the same as a target account list (TAL)?
No. The ICP is a criteria definition — the pattern of attributes that predict purchase likelihood. A target account list is the application of those criteria to a specific set of named companies. The ICP is the template; the TAL is the output of applying the template to available company data. The ICP drives the TAL construction, which then drives contact identification and outreach.
Can a company have more than one ICP?
Yes, when a product addresses meaningfully different problems for meaningfully different customer types. Multiple ICPs are legitimate when each has a distinct profile derived from distinct closed-won deal patterns and when they require different outreach approaches. Multiple ICPs become a problem when they are created by committee to avoid excluding potential customers rather than derived from actual purchase data. When in doubt, run separate campaigns for each ICP hypothesis and let reply rate and pipeline data determine which is real.
Build your ICP list with verified contacts
Translate your ICP criteria directly into a filtered contact list on Quarvio. Industry, title, company size, and geography filters — one-time purchase, credits valid 12 months, unused credits returned.
