How to Build an Account Scoring Model (Fit + Intent + Engagement)

By the Puetto team · Updated 2026

9 min read

An account scoring model turns your ideal customer profile into a single, objective number that ranks every target account — so selection is repeatable and evidence-led rather than political. The model used across mature account-based programs combines three weighted inputs: fit (how well an account matches your ICP), intent (whether it is researching your category now), and engagement (how it is interacting with you). Score each, weight them, and combine into one prioritization score.

The discipline matters because the alternative is a list of "dream accounts" someone picked in a meeting — which is wishful thinking, not ABM. A scoring model forces every account onto the same ruler, and the payoff is real: organizations with a strongly defined ICP have been found to win at materially higher rates than those targeting broadly (a widely cited figure compiled in Foundry's ABM statistics roundup puts the win-rate uplift around 68%). This article shows how to build the model. It assumes you already have a validated profile; if not, start with how to build a B2B ICP. Scoring is Step 2 of the broader ABM strategy framework.

The model at a glance

Three inputs flow into one weighted score, which sorts the account list into tiers and triggers a play:

Fit, intent, and engagement combine into one weighted score that maps to a tier and a play.

The three inputs

1. Fit — "should we sell to this account?"

Fit measures how closely an account matches your ICP across firmographic and technographic criteria: industry, size, geography, business model, and tech stack. It is the most stable input and usually the heaviest weighted, because a poor-fit account rarely becomes a good customer no matter how engaged it looks. The criteria behind fit are broken down in firmographic, technographic, and behavioral criteria explained.

2. Intent — "is now the moment?"

Intent measures whether an account is actively in-market: third-party research surges on your category, relevant hiring, trigger events. Fit answers whether you should pursue an account; intent answers whether now is the time. Intent data comes from providers such as Bombora, G2 Buyer Intent, 6sense, and Demandbase. The fit-versus-intent distinction is important enough to warrant its own treatment — see fit vs. intent.

3. Engagement — "are they responding to us?"

Engagement measures first-party interaction: website visits, content downloads, email and ad response, event attendance, meetings booked. It is the signal that an account is moving from aware to interested, and it lives in your CRM and marketing automation (HubSpot, Salesforce, Marketo).

Keep fit and readiness separate. A common mistake is folding intent into fit and producing one muddy number. Score them independently so you can tell a high-fit / low-intent account (nurture) from a high-fit / high-intent account (pursue now) — the actions are completely different.

Weighting the inputs

There is no universal split, but a defensible starting point for many B2B teams is to weight fit most heavily, then intent, then engagement — for example roughly 50% fit, 30% intent, 20% engagement — and then tune against your own closed-won data. The principle: weight the input that best predicts revenue in your pipeline. If your back-test shows intent surges precede your wins more reliably than firmographic fit, raise the intent weight. Treat the weights as hypotheses you validate, exactly as you validated the ICP in validating your ICP using closed-won data.

A worked example

Suppose a B2B SaaS company scores each input out of 100, then applies the 50/30/20 weights:

Account A — Fit 90, Intent 70, Engagement 40 → (0.5×90)+(0.3×70)+(0.2×40) = 45 + 21 + 8 = 74
Account B — Fit 50, Intent 95, Engagement 80 → (0.5×50)+(0.3×95)+(0.2×80) = 25 + 28.5 + 16 = 69.5

The model surfaces both as priorities but tells a different story about why: Account A is a strong-fit account worth a considered 1:1 approach, while Account B is hot but a weaker fit — pursue it, but watch that the fit gap does not become a retention problem later. That nuance is exactly what a single muddy number would hide.

From score to action — and to ROI

A score is only useful if each band maps to a clear action — a play, an owner, and a service level. This is where scoring connects to tiering: the ranked list is cut into tiers (1:1, 1:few, 1:many) that govern how much effort each account warrants. That mechanism is covered in ABM account tiering, and the question of how many accounts belong in the list at all is covered in how big your target account list should be. If sales cannot explain why an account scored "hot," adoption collapses — so keep the model explainable, not a black box.

Put numbers on it. Once accounts are scored and tiered, you can model the pipeline a tier is likely to produce. Our ABM ROI calculator lets you plug in account counts, conversion rates, and deal values to estimate return before you commit the spend.

Tools

You can build a first model in a spreadsheet from CRM exports plus an intent feed, and many teams should. As volume grows, native CRM scoring (HubSpot, Salesforce) handles fit and engagement, while predictive and intent platforms (6sense, Demandbase, MadKudu) model propensity and surface intent at scale. Enrichment providers (ZoomInfo, Clearbit, Cognism) keep the firmographic and technographic inputs current. As always, an explainable model in a spreadsheet beats an opaque one in an expensive platform.

Want the points laid out? For a concrete, adaptable rubric with a per-criterion point breakdown, see the 100-point account scoring rubric.

Keep it live

Re-score accounts on a regular cadence as intent and engagement shift, and feed every new closed-won and churned account back into the weights so the model sharpens over time. A scoring model is a living system, not a one-time spreadsheet — the same maintenance discipline that keeps the overall strategy sharp, and that ultimately feeds how you measure ROI.