Most ICP documents are written with good intentions and used almost never.
They live in the strategy deck. They get presented in Q1 planning. They describe a market: mid-market financial services, growth-stage SaaS, North American manufacturers with over five hundred employees. The team agrees that sounds right. Then campaigns get built, lists get bought, and six months later the pipeline is full of accounts that look close enough but never close fast.
The problem is not the campaigns. It is that the ICP was never tested against the criteria that make it operational. A profile that fits thousands of companies is not an ICP. It is a market segment with a title slide.
There are five tests every ICP must pass before it is ready to drive a campaign, a target list, or a sales motion. An ICP that fails any one of them will not produce reliable pipeline.
Test 1: Pain
The first test is whether the pain your ICP shares is acute enough and expensive enough to drive action. Not a vitamin-level nuisance but an aspirin-level problem. The distinction matters because vitamins get purchased when budgets are comfortable and deprioritized when they are not. Aspirin gets purchased when the pain is present, regardless of what else is happening.
The practical test: can you tie a dollar number to every pain point in the ICP? Not a vague statement about inefficiency or missed opportunity, but a specific estimate of what it costs the account to leave this problem unsolved. If you cannot, the pain is not yet specific enough to build messaging from.
What good looks like
A Series B SaaS company losing an estimated four hundred thousand dollars a year in sales capacity because their rep targeting is too broad. That is a number a sales rep can open a conversation with.
Test 2: Profile
The second test is whether the firmographic, technographic, and organizational profile is narrow enough to disqualify ninety percent of the addressable market. The discipline here is subtraction. Most ICP documents add variables: vertical, headcount, revenue range, geography. They rarely ask whether each variable actually knocks out accounts that will not close.
The practical test: cross-check your current ICP variables against the last twenty closed-won accounts. Variables that do not appear in at least eighty percent of those wins come out. Variables that appear in eighty percent but are not yet in the ICP go in. The profile that emerges from that exercise is closer to reality than the one you started with.
What good looks like
Series B to Series D, five to thirty million ARR, HubSpot or Salesforce already deployed, headquartered in North America or Northern Europe, headcount growing above twenty percent year-over-year. Specific enough that a researcher can build the list by Friday.
Test 3: Trigger
The third test is whether each named trigger is instrumented to a data source your team already has. Triggers are what turn the ICP from a static list into a live pipeline. They detect when an account that fits the profile enters the buying window. Without them, the ICP tells you who to target but never tells you when.
The practical test: for every trigger you list, name the data source. A new CMO hired means a LinkedIn Sales Navigator alert. A Series B closed in the last sixty days means a Crunchbase signal. A job posting for Head of RevOps means a Wellfound search. If you cannot name the data source, the trigger is theory, not infrastructure. Remove it until every trigger on the list is actually detectable.
What good looks like
Five named triggers, each tied to a specific tool in your stack, each producing a signal that a rep or a sequence can act on within forty-eight hours.
Test 4: Economics
The fourth test is whether the unit economics work for this segment specifically, not on average across the customer base. Company-level NRR and CAC figures hide the segments that are quietly dragging the numbers down. The ICP is supposed to define the segment where the economics are best. If you have never run the numbers by cohort, you do not know whether you have found that segment or just described the average.
The practical test: pull CAC payback, NRR, and ACV for the cohort that matches your stated ICP. If CAC payback is above twelve months in that cohort, the segment is too expensive to acquire at scale. If NRR is below one hundred and ten percent, you have found a leaky bucket dressed up as a target market. Fix the leak before scaling acquisition, or re-segment around where the retention math actually works.
What good looks like
ICP cohort NRR at one hundred and twenty-five percent, CAC payback at nine months, Sales Magic Number above one. That is the segment you lean into.
Test 5: Density
The fifth test is whether the universe of qualifying accounts is large enough to build a SaaS business on, and whether those buyers cluster in communities you can systematically reach. A segment that passes the first four tests but contains only three hundred companies globally is a services business, not a SaaS business. Distribution math has to work before you spend the first marketing dollar.
The practical test has two parts. First, count the named accounts that fit the full profile. The minimum viable number is one thousand globally, with five thousand or more as the working target for most SaaS motions. Second, identify where those buyers congregate: the three events they attend, the three publications they read, the Slack or Discord communities where they ask questions, the five podcasts they listen to. If you cannot name those watering holes, you do not yet have a distribution strategy.
What good looks like
Five thousand named accounts in the profile, plus a mapped community footprint that tells you exactly where to show up before you spend on paid.
The diagnostic
Run these five tests against your current ICP document. Give each one a pass or a fail. Three or more failures and the document is decoration, not infrastructure, and the campaigns you build from it will reflect that.
The five-test framework does not replace the traditional ICP approach. Vertical, headcount, revenue range, and geography are still the right starting point. But they are the profile test, not the whole picture. An ICP that passes only the profile test tells a researcher who to put on a list. An ICP that passes all five tells a sales rep who to call, when to call them, what to say, and whether the math will work when they close.
That is the difference between a market segment and a target.
If you'd like to see how all five tests work together in practice, we recently hosted a session called 5 Things That Separate an ICP That Fails from One That Scales. It walks through each test in depth: what good looks like, how to fix it when it fails, and how to run the full diagnostic against your own ICP in under sixty seconds. Watch the recording below:
