Account scoring model: how to build one reps trust

An account scoring model is a repeatable way to rank the companies in your territory by how worth working they are right now. It combines how well each company fits your ideal customer profile with evidence of timing and engagement, then turns those parts into one score that a rep can read and explain.

Most reps already score accounts in their heads. A model writes that judgment down so it is applied the same way to every account, and so you can check later whether it was right.

Account scoring vs lead scoring

These two get confused, but they answer different questions.

• Lead scoring rates individual people, usually based on what they did: opened an email, visited a pricing page, filled in a form.
• Account scoring rates a whole company: whether it is the right kind of customer and whether something is happening there now.

For outbound work, account scoring usually comes first. A BDR picks the company, then decides who to contact inside it. A highly engaged person at a company that can never buy is still a dead end. A quiet company with a perfect fit and a fresh leadership change may be the best account on your list.

What goes into an account scoring model

Most useful models use three kinds of input. Keep them as separate parts of the score, even if you add them up at the end.

Fit

Fit is how closely the company matches the customers you can serve well. It covers facts like industry, size, region, the tools they use and how their team is set up. Fit changes slowly. If you have not written your fit criteria yet, start with how to define ICP fit.

Timing and signals

Timing comes from buying signals: observable changes such as a new leader, a hiring push, a funding round or an expansion. A signal is evidence that a conversation may be more relevant now, not proof the company will buy. The B2B buying signals guide covers which changes tend to matter.

Signals should lose strength as they age. A hiring push from this month is worth more than one from last year. See how signal freshness affects priority.

Engagement

Engagement is how people at the account have responded to your team: replies, meetings, event visits or website activity you can trust. Some teams fold it into timing. Others add it as a small bonus on top. Either way, check your data quality first, because engagement tracking is often noisy.

How to weight an account score

Weights decide how much each part counts. There is no correct set. Pick a reasonable starting point, write down why, and tune it later.

As one example, Syntri calculates a 0 to 100 priority score like this:

• Priority = 60% ICP fit + 40% signal strength
• Strike: 70 or higher
• Signal: 40 to 69
• Monitor: below 40

Fit gets more weight because it is the more stable foundation. Signals then lift good-fit accounts when something real is happening. Your team might choose different weights. What matters is that everyone uses the same ones and can see them.

A few rules help keep weights honest:

1. Disqualifiers override everything. If an account is outside your region or is a competitor, it scores zero.
2. Cap low-fit accounts. A loud signal should not push a poor-fit company to the top.
3. Avoid false precision. A score of 71 versus 69 is not a real difference. Treat tier boundaries as zones, not walls.

What makes an account score explainable?

A score is only useful if a rep can say why an account has it. “Score: 82” tells you nothing. Compare that with:

Fit 90: B2B, 400 employees, eight BDRs, uses an integrated CRM. Signal 70: new VP of Sales 12 days ago, four BDR roles posted. No outreach in 60 days.

That version lets the rep check the evidence, pick an angle and spot mistakes. To keep scores explainable:

• show the fit and signal parts next to the total;
• keep the source and date for every signal;
• list the top two or three reasons behind each score;
• let reps flag a score that looks wrong.

A worked example

Here are four fictional accounts scored with the 60/40 weights above.

Account	Fit	Signal	Priority and tier
400-person logistics software company	90	70	82, Strike
1,200-person manufacturing software company	85	10	55, Signal
60-person marketing agency	30	90	54, Signal
250-person healthcare billing company	50	15	36, Monitor

The logistics company is easy: strong fit plus a fresh signal. The interesting rows are the two in the middle. They land in the same tier for opposite reasons. The manufacturer fits well but nothing is happening, so the job is to confirm whether a quiet signal exists. The agency has a loud signal but weak fit, so the job is to check whether fit is really as low as the data says.

That is why the parts must stay visible. The total alone would treat them as the same account.

How do you know if your scoring model works?

Your first model is a guess. Calibrating it means comparing scores against what actually happened, then adjusting.

1. Pull recent outcomes. List accounts that turned into meetings, opportunities or deals, and accounts that went nowhere.
2. Score them as of the start. Use only what you could have known before outreach began.
3. Compare tiers. Do higher-tier accounts produce more real conversations than lower-tier ones? If the tiers look the same, the model is not helping.
4. Study the surprises. A won deal that scored low may point to a missing criterion. A high score that never converts may point to a weight that is too generous.
5. Change one thing at a time. Adjust a weight or a criterion, then check again with the next batch of outcomes.

If you only have a few deals, do not overreact. Look for clear patterns, talk to the AEs who worked them, and recheck each quarter. Once the model holds up, use it to tier your territory into groups that each get a different kind of work.

The rule to remember

An account scoring model should rank companies on fit, timing and engagement, keep each part visible, and change when outcomes prove it wrong. If a rep cannot explain a score in one sentence, fix the model before you trust it.

See how Syntri combines fit and signals in Signal Arena.