ICP definition framework for outbound sales teams

"We sell to mid-market SaaS" is not an ICP. It's a category. An ICP is a buildable spec: when handed to a researcher, it should produce the same list every time. After running thousands of custom list-building engagements, we've found that the teams who get repeatable outbound results define their ICP across five layers, not one. This post lays out the framework and walks through three worked examples.

The 5-layer ICP framework

Each layer is independently filterable. A complete ICP carries at least one signal from each layer.

1. Firmographic. Industry, company size (employee count and/or revenue), geography, founding year. The classic top-of-the-funnel filters.
2. Technographic. Stack signals: tools the company uses or doesn't, integrations exposed, whether they have a public API. A proxy for buying readiness.
3. Behavioral. Activity signals: hiring spikes in a function, recent funding round, opening a new office, executive moves, acquisition activity, content footprint changes.
4. Financial. Health signals: profitability indicators, recent funding stage, growth rate proxies (employee growth quarter-over-quarter, revenue ranges).
5. Organizational. Buyer-side filters: the specific role you sell to, the seniority of that role, the reporting line, the size of the team that owns the function.

Why a single layer isn't enough

"VPs of Engineering at 200–1000 person SaaS companies" filters on firmographic and organizational. It returns ~15,000 records globally. That's not an outbound list — it's a market-sizing answer.

Add behavioral: "VPs of Engineering at 200–1000 person SaaS companies that have hired more than 10 engineers in the last 90 days." Now you're at ~1,800 records and you're talking to teams under stack-pressure right now. Hiring volume is one of the strongest predictors of buying for engineering tools, and it's a free signal.

Add technographic: "...and the company runs at least one Kubernetes cluster in production." Now you're at ~700 records and you're qualified at first contact. Reply rates on lists this tight commonly run 8–15% versus 1–3% on the unfiltered firmographic-only list.

Worked example 1: SaaS

Product: A developer-focused observability tool.

• Firmographic: SaaS or fintech, 100–800 employees, US/Canada/UK headquartered, founded after 2018.
• Technographic: Production Kubernetes (BuiltWith / job-postings signal); has at least one engineering manager listing on LinkedIn mentioning "on-call" rotation.
• Behavioral: Net engineering headcount up >10% YoY OR has posted at least 3 SRE/Platform-Engineer roles in the last 60 days.
• Financial: Series A through C, last raise within 24 months.
• Organizational: VP Engineering OR Head of Platform OR Director of Infrastructure; report to CTO.

This spec returns roughly 600–900 contacts in our data. It's also the kind of list a generic database can't filter — most SaaS-data tools don't expose the behavioral signal layer cleanly. Custom research is how we typically build it.

Worked example 2: Healthcare

Product: A telehealth platform for hospital systems.

• Firmographic: US-based hospital systems with 5+ facilities OR a single hospital with >500 beds.
• Technographic: Currently running Epic OR Cerner as the primary EHR.
• Behavioral: Has launched OR publicly announced a virtual-care or telehealth initiative in the last 18 months.
• Financial: Operating revenue >$200M, non-academic medical centers preferred.
• Organizational: Chief Medical Information Officer OR VP of Digital Health OR Director of Telehealth Operations.

This is a tight, ~150-record ICP — small in absolute terms, but it's the right small. Healthcare buying cycles are 6–18 months; spending list-building effort on the wrong 5,000 records wastes a year. See the healthcare data page for how we source these records and the Meridian Health case study for what a campaign at this scale typically produces.

Worked example 3: Financial services

Product: A compliance-monitoring platform for mid-sized investment firms.

• Firmographic: US-registered RIAs (Registered Investment Advisors) with $500M–$5B AUM, OR private credit funds with $1B+ AUM.
• Technographic: Uses Salesforce Financial Services Cloud OR Wealthbox OR Junxure as primary CRM.
• Behavioral: Has had an SEC examination in the last 36 months (public filing) OR has hired a Chief Compliance Officer in the last 12 months.
• Financial: AUM growth >15% YoY OR registered as a fiduciary under DOL/SEC.
• Organizational: Chief Compliance Officer OR Chief Risk Officer OR VP Operations.

Financial-services contact data is the hardest of these three to source — public registrations get you the company list cleanly, but reaching the right C-level decision-maker requires per-record research. See the financial services coverage page.

What "good" looks like in production

A well-defined ICP across all five layers typically returns between 300 and 5,000 addressable contacts. Below 300, the market may be too narrow for a sustainable outbound motion (consider account-based field marketing instead). Above 5,000, the ICP is probably too loose at one or more layers and reply rates will reflect that.

The most common failure mode we see is teams who define firmographic + organizational only and skip behavioral. That's the difference between "a list" and "a list of companies that should be interested right now." If you only adopt one new layer, make it behavioral.

How to operationalize this

Write the ICP down before sending it to research. The process of writing forces specificity. Then send it to us or any other list-builder; if the resulting list is bigger or smaller than the ICP predicts, that's a flag the spec is loose. Iterate until the count matches expectations.

If you'd like to test this framework against your product, request a free sample with whatever ICP you have today and we'll deliver a small slice plus notes on which layers we'd tighten.