How to define an ICP that a lead engine can actually execute
Most ICP documents are positioning statements, not targeting instructions. Here is how to write an ideal customer profile precise enough for a data engine to source against every day.
10 min read · Updated September 8, 2026
Key takeaways
- Filter companies first, then people, then timing. Mixing the layers produces vague lists.
- Name job functions and seniority rather than exact job titles, which vary wildly between companies.
- Exclusions matter as much as inclusions: existing customers, current pipeline, competitors, wrong regions.
- Review the ICP monthly against what actually booked meetings, not against what looked good on the slide.
Why most ICP documents cannot be executed
A typical ICP describes who a company wants to sell to. An executable ICP describes how to recognise those buyers in data.
Statements like "ambitious mid-market companies that care about efficiency" cannot be turned into a query. Nothing in that sentence maps to a field on a record.
The fix is not more detail, it is different detail. Every line in an executable ICP should be checkable against a company or person record without a judgement call.
Layer one: the company filter
Start with the account, because targeting the right person at the wrong company wastes the best contact data you will ever buy.
Keep this layer tight enough to be meaningful and wide enough to sustain your monthly volume. If you need one thousand leads a month, a filter that matches three hundred companies will starve.
- Industry or sector, expressed as categories rather than single keywords.
- Headcount band, which is usually a better proxy for buying behaviour than revenue.
- Geography, at country and region level, including anywhere you cannot legally or practically serve.
- Business model signals: sells to businesses or consumers, field teams or desk teams, single site or multi site.
- Technology or operational context where it genuinely changes the pitch.
Layer two: the person filter
Target function plus seniority, then list example titles. Titles alone break across company sizes and countries.
A head of revenue operations at one company is a sales operations manager at another and a commercial director at a third. If your ICP lists only exact titles you will miss most of your market.
Define the function, the seniority band, and what that person owns. Then add example titles as illustrations rather than as the rule.
Decide explicitly whether you want the economic buyer, the champion, or both, because the messaging and the call outcome differ sharply.
| Field | Weak version | Executable version |
|---|---|---|
| Function | Sales leaders | Owns outbound pipeline targets |
| Seniority | Senior people | Director level and above, reports to CRO or CEO |
| Titles | Head of Sales | VP Sales, Sales Director, Head of Revenue, CRO |
| Company size | Mid-market | 50 to 500 employees |
Layer three: timing signals
Fit tells you who to call. Timing tells you who to call first. Hiring activity for roles your product supports, new leadership in the function you sell to, expansion into a new market, or a funding event are all reasons a conversation lands this month rather than next quarter.
Timing signals should reorder your list, not replace your filters. A perfect trigger at a company that will never buy is still a wasted dial.
Layer four: exclusions
Exclusions are the part teams skip and then regret. Every list should exclude current customers, live opportunities, closed-lost accounts inside their cool-off window, competitors, partners you do not want to confuse, and regions you cannot support.
Exclusions also protect brand. Nothing undermines a rep faster than cold-calling an existing customer with a first-touch pitch.
Testing and tightening the profile
Run the profile for a defined period, then compare three cohorts: records that were reachable, records that led to a conversation, and records that led to a qualified meeting. The gap between the first and third tells you whether your problem is data or targeting.
If reach is high and meetings are low, the ICP is wrong. If reach is low, the data or the segment is wrong. Fixing the wrong one of those is the most common outbound mistake.
Adjust one layer at a time so you can attribute the change. Monthly is a sensible cadence for most teams.
How the ICP builder works in LeadNumbers
During signup you add your website and answer a short set of questions covering industries, headcount bands, regions, buyer functions and seniority, example titles, and anything to exclude. Your site is used to ground the profile in what you actually sell.
That profile drives every daily batch. You can edit it at any time in the portal, and the next business day's delivery follows the updated version, so testing a new segment does not require a new contract or a new list purchase.
Frequently asked questions
How specific should an ICP be?
Specific enough that two people reading it would build the same list, and wide enough that the segment can sustain your monthly lead volume. Function plus seniority plus firmographics plus exclusions is usually the right level.
Should I target job titles or job functions?
Target function and seniority, and use titles as examples. Titles differ between companies and countries, so a title-only filter silently removes large parts of your market.
How often should I update my ICP?
Review monthly against what actually produced meetings, and change one layer at a time so you can tell which change worked.
Can I run more than one ICP?
Yes. Agencies and multi-product teams commonly run separate profiles per client or per segment, each with its own delivery destination.
See the data before you decide
Five sample leads in your ideal customer profile, with mobile or direct dial and verified work email. No card required.