How to define your ICP using buying signals

An ideal customer profile (ICP) defines the companies most likely to buy, succeed and stay — described in three layers: firmographics (who they are), situation (what is true inside them) and buying signals (what indicates timing now). Most teams stop at the first layer, which is why their outreach is generic. The signal layer is what turns a static list into a prioritised one.

Why does a firmographic-only ICP produce generic outbound?

Ask most B2B teams for their ICP and you get something like "manufacturing companies, 50–500 employees, in Western Europe". That is a market description, not an ICP. Thousands of companies match it, most of them are not going to buy anything from you this year, and nothing in the description tells you what to say to them.

The consequences cascade: because the profile is broad, the list is broad; because the list is broad, the message must work for everyone, so it says nothing specific; because it says nothing specific, it gets ignored. Weak ICP work is the root cause behind most "our reply rates are terrible" problems — before copy, before deliverability.

The three layers of a usable ICP

Layer 1: Firmographics — who they are

Industry, headcount, geography, revenue band, business model. This layer exists to narrow the universe and make list-building possible. Necessary, insufficient.

Useful discipline: for each criterion, write down why it matters. "50–500 employees" should trace to something real — below 50 they lack the problem, above 500 procurement kills the deal, whatever your evidence suggests. Criteria you cannot justify are decoration.

Layer 2: Situation — what is true inside the company

This is the layer most teams skip. It describes the internal conditions that make your offer relevant:

• Processes they run manually that you automate.
• Tools they use (or lack) that your product complements or replaces.
• Team structures that indicate the problem exists — for example, a sales team with no one owning outbound.
• Constraints typical of their stage: scaling delivery, entering new markets, founder-led sales hitting its ceiling.

Situational attributes are researchable — from websites, job posts, tech-stack data, public filings — which means data enrichment can populate them at scale. And they are what real personalisation is made of: a first line that references the prospect's actual situation, not their city or a recent LinkedIn post.

Layer 3: Buying signals — why now

Buying signals are observable events that correlate with readiness to buy. Common categories:

• Hiring signals: job posts for roles related to your problem space. A company hiring its first SDRs is thinking about pipeline.
• Growth signals: funding rounds, new offices, market expansion announcements.
• Change signals: new leadership in the relevant function — new executives review their tooling and vendors.
• Technology signals: adopting or abandoning tools adjacent to yours.
• Behavioural signals: engagement with your content, website visits, and third-party intent data indicating research activity around your category.

Signals answer the question firmographics never can: why should this message exist this week? An account that matches layers 1 and 2 and fires a layer-3 signal is a genuine target account. The same account without signals belongs in a lower-priority track — still worth contacting, but with different expectations.

How do you build the ICP in practice?

A working sequence you can run in a week or two:

1. Analyse your best customers. Take the deals that closed fastest, expanded most, complained least. List what they had in common across all three layers — including which signals were visible before they bought, in hindsight.
2. Interview a few of them. Ask what was happening internally when they started looking. Their answers become your situational criteria and signal list.
3. Write the ICP as testable statements. Not "mid-size industrial companies" but: "manufacturers, 50–300 employees, exporting, with a sales team but no structured prospecting process, currently hiring commercial roles." Every clause is checkable against data.
4. Define the negative ICP. Who looks like a fit but consistently goes wrong — wrong stage, wrong economics, wrong expectations. Excluding them saves sending capacity and sales time.
5. Map signals to sources. For each signal, decide where you will detect it (job boards, funding databases, tech-stack tools, intent providers) and how often the detection runs.

Which mistakes should you avoid?

Three patterns undermine most ICP work:

• Defining the ICP as "whoever might buy". An ICP that excludes nobody prioritises nothing. If cutting a segment feels impossible, you have a market description, not a profile.
• Confusing past customers with ideal customers. Your customer base reflects who you happened to reach, including deals you should not have taken. Filter for the best outcomes, not the full list.
• Writing it once and filing it away. An ICP that is not wired into list-building, signal detection and messaging is a document, not a system. It should constrain what enters your sequences every single week.

Turning the ICP into an operating system

An ICP is only useful if the machine consumes it. In a signal-driven outbound system:

• Account sourcing filters on layers 1 and 2.
• Signal detection runs continuously and ranks the account list — signal-firing accounts enter sequences first, with messaging that references the signal.
• Reply data flows back into the profile: segments that consistently reply and convert tighten the ICP; segments that stay silent get cut.

This last loop matters most. Your ICP is a living hypothesis, and outbound is the cheapest experiment you can run against it. This is precisely how we structure the Outbound Validation Sprint: a defined ICP hypothesis, signal-prioritised targeting, and real reply data to confirm or correct it before scaling the system through our AI outbound systems.

If you are not sure whether your current targeting is ICP-driven or just list-driven, the outbound maturity diagnostic will show you in a few minutes — or book a strategy call and we will pressure-test your ICP together.