Your account list is the program: how to build one you can defend

Everything downstream of the account list is a multiplier on it. Better copy multiplies it, better design multiplies it, a better agency multiplies it, and a multiplier applied to a fuzzy list of five thousand companies returns a fuzzy result more expensively. This is the part of an account-based program that almost nobody buys help with and almost everybody gets wrong, so here is the procedure we run, in the order we run it, including the step that has to come before the list exists.

Write the disqualifier before you write the list

Not after. Before. If you build the list first, the disqualifier gets written to fit the list, and at that point it is a justification rather than a test.

A usable disqualifier has one property: a single person can check it from outside the company in about five minutes. If checking it requires a discovery call, it is not a disqualifier, it is a qualifying question, and it belongs to sales rather than to the list. Disqualifiers that pass the test:

• No named owner for this category. If nobody holds the budget line, there is no deal to be had regardless of fit.
• An incumbent under contract with more than a year to run, where you can see the contract term or the deployment date.
• Below a headcount, revenue or transaction-volume floor you can observe from public filings, hiring pages or product tiers.
• A geography or a compliance regime you cannot serve today.
• A buying motion that does not match yours. Self-serve companies with no procurement function are a real disqualifier for a product that needs a security review.

Disqualifiers that fail the test, and that we have all written anyway: "not innovative enough", "bad culture fit", "probably not ready". Those are feelings with a job title.

Write five to eight of them, in a document, with a date. The date matters because the disqualifier changes as you learn, and you want to be able to see when it changed and what you removed from the list because of it.

Build the list from closed-won, not from a database

Open your last twenty to forty won deals. For each, write down only what was observably true before the first conversation: headcount band, what they sold, the stack you could see, whether they were hiring for the function that owns your problem, what had recently changed. Externally visible facts only, because those are the only ones you can use to find more of them.

The patterns that repeat across ten or more of those accounts are your real ideal customer profile. Patterns appearing twice are anecdotes. This is slower than buying a filtered export and it is the single highest-return day of work in the whole program, because everything after it inherits the definition.

Then, and only then, go to the database and find companies matching the repeating pattern. Run every one of them against the disqualifier before it is allowed onto the list.

Count from your constraint, not from the market

The market size is not an input. Two constraints are.

The first is human. A team of two to three marketers with the right tools can run effective ABM programs for 50 to 200 target accounts, and that source is clear that the usual gap is intelligence infrastructure rather than execution headcount. If you have one person, you are at the bottom of that range, not the top.

The second is money. Paid media for account-based programs runs $5,000 to $50,000 and up per month depending on your target account list size. List size is the variable that sets the spend. Doubling the list does not double the cost of the list, it doubles the cost of everything you do to it.

And the list is not a list of companies. It is a list of people. One 2026 analysis puts the fintech buying committee at roughly 7.4 stakeholders with three named veto holders. Take even the low end of that across 200 accounts and you are committing to research and messaging for well over a thousand named humans. That number, not the account count, is what tells you whether your list is honest.

One sentence per account, with a date

For every account on the list, write a single sentence saying why it is there, and date it. "Hiring three people for the team that owns this problem, posted last month." "Announced an expansion into a market our compliance module covers." "Their incumbent's contract was signed publicly in Q1 three years ago."

The sentence is the test. If you cannot write one, the account is not on the list. If the sentence is "matches our ICP filter", the account is not on the list, because that sentence is true of every row in the export and therefore distinguishes nothing.

This sounds like bureaucracy for its own sake and it is the opposite. It is the artifact that makes the monthly review possible and the thing that survives whoever leaves the company.

Tier by what you will actually produce

Sort the list into the tiers the industry already uses: 1:1, 1:few and 1:many. Agency pricing is structured around exactly this split, which is a decent signal of how much work each one is.

Tier according to what you have committed to make, not according to how much you want the logo. If the 1:1 tier has fifteen accounts and you have capacity to build custom material for six, you do not have a fifteen-account 1:1 tier. You have six, and nine accounts getting 1:few treatment while someone feels guilty about it.

Review monthly, and cap the additions

Three questions, once a month, in a meeting with marketing and sales in the same room.

What changed at this account? This is where intent data earns its place. First-party intent from your own properties is the highest quality signal available, fully attributable and in your control, and the practical implementation is a scoring rule rather than a dashboard: that source gives the worked example of a pricing page visit at 30 points, an intent topic surge at 20, a blog view at 5, with 50 points inside seven days triggering an alert, and recommends narrowing to 5 to 15 intent topics that actually correlate with purchase.

Which accounts now fail the disqualifier? They were added months ago under conditions that have changed. Remove them and note why. This is the step everyone skips, and skipping it is how a 60-account list becomes a 400-account list without a decision ever being made.

Which accounts have had no human response in 90 days? Down a tier, or off. Silence at that length is data.

Then the cap: remove as many accounts as you add. Not a law of nature, a forcing function. Without it the list only grows, because adding is easy and removing feels like giving up.

What a five-thousand-account list actually costs

We want to be careful here, because we have not seen a published study putting a price on list bloat, and we are not going to invent one. What follows is our argument built on the sourced numbers above rather than a finding.

Media spend scales with list size. Per-account research does not compress, so a list twenty-five times larger is twenty-five times the reading or it is no reading at all. Nothing in the sources above suggests anyone runs account-based quality at five thousand accounts; the working figure is 50 to 200 for a small team. And a large list pushes you toward volume sending, which is where the 2026 sender gates start charging you in domain reputation.

Put together: a five-thousand-account list is usually demand generation wearing an ABM label, at ABM prices. The tell is that nobody can say why account 3,412 is on it.

Worth knowing that the bar has moved. Seventy-one percent of organisations now run ABM programs, so a named account is being approached by people who also read the same playbook. The list is what separates you from them, because it is the only part they cannot copy.

What we got wrong

We used to build the list first and write the disqualifier afterwards, which meant the disqualifier was always shaped to keep the accounts we had already fallen in love with. The order in this post is the fix, and we arrived at it the slow way.

If you want us to pressure-test a list you have already built, tell us what you are trying to fix.