Account tiering (ABCDX): grade accounts by ICP fit before you reach out

ABCDX tiering grades every target account by how well it fits your ideal customer profile, so effort goes where deals are most likely. A is the best fit, B a good fit, C a partial fit, D a weak fit and X an account you will not contact. The grade is decided before you write a word.

Abmatic published a post on this method on 6 October 2026, "ABCDX Segmentation: How to Grade Accounts by ICP Fit for ABM", and describes it as grading "every account A, B, C, D, or X by ICP fit so ABM spend goes where deals close" (Abmatic blog, read 25 October 2026). The scoring sheet and the actions below are our own version, written for outbound teams that use LinkedIn, email and WhatsApp.

Why grade accounts before you write to them?

Because the same message to a perfect fit and a poor fit gives you two different results, and an average of the two hides which was which. Grading first lets you do three things.

• Put the most careful, personal effort where a win is most likely.
• Stop spending sends on accounts that will never buy.
• Learn which criteria matter, because every outcome is tied to a grade.

It also fixes a common failure. A team writes to everyone who matches a job title, then sees a low reply rate and blames the message. Often the list was the problem.

What does each letter mean?

The X tier is the most important and the most often missing. Competitors, current customers, people who said no, companies in a country you cannot serve and anyone on a suppression list belong here, so nobody writes to them by accident.

Which criteria should you score?

Take them from your ideal customer profile. Choose four to six that you can check from public data, and split them into must haves and nice to haves.

• Must haves are yes or no. If an account fails one, it is X or at most D. Examples: operates in a country you can serve, has the role that buys, is the right size band.
• Nice to haves earn points. Examples: uses a tool that pairs with yours, hired for the relevant role recently, has a recent trigger such as a funding round or a new office.

A criterion you cannot check without a call is not a criterion. Drop it.

What does a scoring sheet look like?

This is a made up example for a company that sells scheduling software to dental clinics. The criteria and points are illustrative, not a recommendation.

Cut offs, set before scoring: 7 or more points is A, 5 to 6 is B, 3 to 4 is C, 1 to 2 is D, and any failed must have is X. The maximum is 8.

Score two hundred accounts and you will see the shape of the list. If almost everything is B or C, your criteria are too loose to separate anyone. If almost everything is X, your must haves are too strict or your list source is wrong.

What do you do with each tier?

A accounts. Treat them as a short list. Read the company's recent news, find two or three people, and write a line that could only apply to them. On LinkedIn, a post engaged with or a recent job change is an opening; see pre invite engagement. Review these messages before they go, and let a person take any reply that touches price or scope.

B and C accounts. Run the standard sequence with one personal line. The difference between B and C is priority and volume, not message quality.

D accounts. Either leave them, or run a small test. Keep it honest: if D accounts reply as well as A accounts, your criteria do not predict anything and you should change them.

X accounts. Keep them on an exclusion list and leave them out of every import. As a second line of defence, Sellinger checks the team's suppression list before every send on email, LinkedIn and WhatsApp, and refuses the send when the contact is on it.

How do you run tiers in Sellinger?

Keep the tier as a column in your sheet. When you import the leads from CSV, an extra column can be mapped to a custom field, so the grade travels with the lead. Run A accounts in their own campaign with approval mode on, so every message waits for a person, and B and C in a campaign that sends automatically within the daily limits. The lead scoring post shows a simple model for ranking people inside an account.

How do you check that the tiers work?

After a month, compare the grades against outcomes. For each tier, count the people contacted, the replies, the positive replies and the meetings held, with the base next to each number.

• If A beats B beats C, the tiers predict something. Keep them.
• If A and C look the same, the criteria are not doing any work.
• If D beats B, you have the criteria wrong, and that is useful to know.

Do not draw conclusions from a handful of accounts. With 20 accounts in a tier, one reply changes the rate by five points. Wait until the numbers are large enough to trust, or report them as small and say so, as the Lamothermic case study does.

Where does tiering go wrong?

• Tiering by title only. A title is not a fit. Include the company.
• Scoring on guesses. Use data you can check.
• Never regrading. Companies grow, merge and change tools. Regrade quarterly.
• Treating X as a score. X is a decision. Do not let it creep up to D.
• Over engineering. Four criteria and three cut offs beat a model nobody updates.

A tiered list is a short document that says who you will and will not write to, and why. Once it exists, the message is easier to write, the results are easier to read and the next quarter's list is easier to improve.