Tiering Your ICP: Where to Point Your Best Sellers First

Start with four or five criteria you can verify at scale, not a wish list of traits. Firmographic fit (headcount and revenue bands that match your closed-won accounts), a real trigger event (a funding round, a new VP of Sales, a tool your prospects just adopted), a warm path in (an existing customer reference or a mutual connection), and stated intent (a pricing-page visit, a specific asset download) cover most of what actually correlates with a deal closing.

Give each criterion a point range and add them up to a single score out of 100. Leave out anything you can't check consistently across the whole list, like "culture fit": a criterion nobody can score the same way twice isn't a criterion, it's a guess.

A workable scoring model usually checks these criteria:

• Firmographic fit: headcount and revenue bands that match the accounts you have actually closed, not a broad filter.
• A real trigger event, such as a funding round, a new VP of Sales or a tool the prospect just adopted.
• A warm path in, like an existing customer reference or a mutual connection who can vouch for you.
• Stated intent, such as a pricing-page visit or a specific asset download.
• Only criteria you can verify the same way across the whole list, which rules out subjective traits like culture fit.

Three tiers are usually enough. A high-scoring Tier A gets multi-channel outreach (email, a call, and a LinkedIn touch), a named rep who owns the account, and exec-to-exec outreach when your buyer's seniority supports it. A middle Tier B gets a standard two-channel sequence that any rep, or an AI SDR, can run without customization. A low-scoring Tier C gets a single light-touch channel, mostly automated, so you're not spending real hours on accounts that probably won't close soon.

The point of the exercise is inversion: your heaviest human effort should land on your smallest, highest-scoring group, not get spread evenly across the whole list because that feels fair.

Say your list has 500 accounts and a scoring model that runs from zero to 100 points. In that hypothetical split, a tight Tier A might hold 60 accounts scoring above your cutoff, a much larger Tier B might hold 180, and the remaining 260 fall into light-touch Tier C.

That shape, a small highest-scoring group and a large bottom one, is normal and useful: it tells you exactly where a fixed number of rep hours should go instead of spreading them evenly across all 500 and hoping. It also gives you a second number worth tracking: how many accounts moved up a tier since the last re-score, which tells you whether the pipeline of newly-qualifying accounts is actually flowing or the same 60 have sat in Tier A for months.

Scores decay. An account that scored low six months ago because it lacked a trigger event can jump into Tier A the day it announces a funding round or a leadership change, and it should move the same week, not wait for an annual list refresh.

Build a lightweight monthly re-score into the process, plus a rule that flags specific triggers (funding, a key hire, a product launch) for an immediate re-check outside that cycle. A tiering model that only runs once a year is really just a one-time sort, not a system.

The most common failure is scoring by gut feel instead of documented criteria: a rep decides an account "feels" like a good fit and it gets Tier A treatment regardless of what the scoring sheet says. That defeats the purpose, since the whole point is to remove that inconsistency.

The second failure is giving Tier C zero touches instead of a light one. Without any signal from that tier, you never learn whether the scoring model correctly identified them as low-value or just missed something, and a model nobody checks against reality tends to drift further wrong over time.