Account Scoring for ABM: Rank Your Target Accounts
Six months into an ABM program, pipeline looks thin, and the postmortem keeps finding one cause. Reps spread themselves across 400 accounts. Marketing ran one generic play against all of them. Nobody could say which 40 accounts deserved real effort, because the target list was assembled in an afternoon from a wish list of familiar logos. Every downstream decision inherited that guess.
Account scoring replaces the guess with a ranking. You score whole companies on how well they fit your best customers, whether their buying group shows signs of a purchase, and how deeply they already engage with you. That ranking then drives everything: which accounts get a named owner, which get segment campaigns, which sit in nurture, and how much budget each layer earns.
This guide covers the full loop: how account scoring differs from lead scoring, how to select target accounts step by step, a worked model with weights and thresholds, ABM tiers with resourcing for each, setup in HubSpot and Salesforce, and benchmarks for judging whether any of it worked. Expect to build version one in a spreadsheet within a week, and expect to recalibrate it.
What account scoring is, and how it differs from lead scoring
Classic lead scoring ranks individual people. Someone downloads a guide, opens three emails, hits your pricing page, and their points climb until a threshold routes them to sales. For high-volume products with a single decision-maker, that mechanic works.
B2B deals with six to ten people in a buying group break it. Your economic buyer might be a VP who never fills out a form. A junior analyst who downloads every asset you publish may have zero budget authority. Score people one at a time and you misjudge the company they work for, in both directions.
Account scoring rolls signals up to the company level. Individual behavior still gets tracked, but the score that drives action describes an account across three dimensions:
- Fit: firmographic and technographic match with your ideal customer profile. Would this account stay, expand, and refer if they bought?
- Intent: research activity, on your site and across the wider web, suggesting the account is in or near a buying window.
- Engagement: how many people from the account interact with you, how often, and how recently.
Some teams call this account based lead scoring: familiar scoring mechanics, applied to companies. The change of unit matters more than the name. Three people from one account each opening two emails is a stronger buying signal than one enthusiast opening six, and only an account-level model can see that.
One more distinction worth naming. Lead scoring typically answers "is this person ready for sales?" Account scoring answers two separate questions: "should this company be on our target list at all?" (fit) and "should we act on it now?" (intent plus engagement). Keeping those apart saves you from chasing hot signals at companies that could never become good customers.
Start with the ICP, or the model scores noise
Every weight in your model traces back to your ideal customer profile. Skip that step and you end up scoring what is easy to measure rather than what predicts revenue.
Pull 18 to 24 months of closed-won deals. Filter for the ones that renewed, expanded, or referred, and set aside deals that closed but churned fast. Then look for what your best customers share: industry, employee range, revenue band, region, tech stack, the trigger events that preceded their purchase. A written ICP definition with explicit disqualifiers is the deliverable, and it becomes the fit half of your scoring model almost verbatim.
Disqualifiers deserve as much attention as positive attributes. "Under 20 employees", "region we cannot serve", "runs a platform we do not integrate with" each become negative points later. Teams that skip them build lists that look big and convert badly.
How to select target accounts for ABM: a step-by-step framework
Account selection is where scoring meets reality, and where the biggest worry lives: what if we bet a quarter of effort on a list that was wrong? A repeatable process lowers that risk. Six steps.
Step 1. Build the universe. Export every account from your CRM, then extend it with an enrichment source so you cover companies that fit your ICP but have never touched you. In our experience the addressable universe runs several times larger than the CRM suggests. Dedupe aggressively; parent-subsidiary confusion will poison scoring later.
Step 2. Score for fit first. Apply firmographic and technographic criteria to the whole universe. This pass is cheap, stable, and filters out the majority. An account that fails fit never advances, no matter how loudly it engages.
Step 3. Layer intent and engagement on the survivors. Now the expensive signals earn their cost, because you only apply them to accounts worth winning. First-party behavior comes from your own analytics and CRM. Third-party intent comes from providers, if you use one.
Step 4. Hold a joint review with sales. Marketing brings the ranked list. Sales brings context the data cannot see: an account in a lawsuit, a champion who just left, a competitor locked in for two more years. Reps veto or add accounts with a stated reason, recorded so vetoes stay honest. A list sales helped build is a list sales will work; skipping this workshop is the most common cause of ABM lists that die in a spreadsheet.
Step 5. Cut to capacity. Decide list size from your team's real selling capacity. If each rep can genuinely run one-to-one plays on 15 accounts, five reps give you a 75-account ceiling for your top tier. A 300-account "priority" list is a fiction that reps quietly ignore.
Step 6. Set a review cadence. Quarterly works for most teams. Accounts move between tiers as signals change; a few get removed entirely. The list is a living artifact, and treating it as one is half the discipline.
Data sources that feed the score
Four families of data, in rough order of how much you should trust them.
Your CRM and marketing platform. Closed-won history, current pipeline, contact coverage per account, email and meeting activity. This is first-party, verified, and free. It anchors everything.
Website and product analytics. Account-level visits (via reverse-IP identification or known-contact stitching), pricing page sessions, demo requests, trial behavior if you have a product-led motion. High-trust intent, because it happened on your property.
Technographics. What tools an account runs, from enrichment vendors or your own integrations data. Valuable when your product displaces or integrates with something specific. Refresh at least twice a year; stacks change.
Third-party intent. Providers detect surging research on topics relevant to your category and attribute it to companies. Useful reach into accounts that have never visited you, and also the noisiest input in the stack: attribution from IP to company is probabilistic, topics are broad, and a spike may be an intern writing a term paper. Weight it below first-party signals. Our guide to intent data in B2B covers vendor evaluation and accuracy in depth.
Most teams can field a solid version one with only the first two sources. Add technographics when fit criteria demand it, and third-party intent once the basic model has proven itself.
A worked scoring model: criteria, weights, thresholds
Here is a complete example you can adapt. Every number below is illustrative; your closed-won data should reset them. The shape, though, transfers across most B2B teams: fit and intent carry the bulk of the weight, engagement breaks ties, negative points keep the list honest.
| Category | Criterion | Points (illustrative) |
|---|---|---|
| Fit (cap 50) | Target industry | +20 |
| Fit | Employee count 200 to 2,000 | +15 |
| Fit | Runs a platform we integrate with | +10 |
| Fit | Trigger event: funding round or new exec in buyer's function | +10 |
| Intent (cap 35) | First-party: 2+ pricing page sessions in 30 days | +25 |
| Intent | First-party: demo request or ROI calculator use | +20 |
| Intent | Third-party topic surge, verified two weeks running | +10 |
| Engagement (cap 25) | 3+ contacts from account active in 30 days | +15 |
| Engagement | Webinar or event attendance by a buying-group role | +10 |
| Engagement | Sales meeting held in last 60 days | +10 |
| Negative | Employee count under 20 | -25 |
| Negative | Region outside coverage | -30 |
| Negative | Competitor or student email domains only | -15 |
| Negative | No activity in 90 days (decay, reapplied quarterly) | -10 |
Four design rules make a model like this hold up in production.
Cap each category. With the caps above, a perfect-fit account showing early interest scores around 60, while a poor-fit account with screaming intent tops out near 35. That ordering is deliberate. Intent without fit produces meetings that go nowhere; the cap encodes the lesson so no single loud signal can dominate.
Score negatives. Without subtraction, every account drifts upward over time and your thresholds slowly lose meaning. Disqualifiers subtract hard. Decay subtracts gently and on a schedule, so an account that engaged heavily six months ago and went silent stops outranking one with modest recent activity. Recency beats volume.
Set thresholds from distribution, then translate to capacity. Score your whole list, sort it, and look at natural breaks. With this example model, 70+ might mark your top tier and 40 to 69 the middle, but the honest cutoff is wherever your capacity math says it is. If 70+ yields 200 accounts and your reps can work 75, raise the bar.
Keep version one explainable. Eight to twelve criteria a skeptical rep can understand beat a 40-variable model nobody can audit. When a rep asks "why is this account Tier 1?", the answer should take one sentence. Complexity can come later, after the simple model has earned trust.
ABM tiers: Tier 1, Tier 2, Tier 3 and how to resource each
Raw scores mean nothing to a seller on a Tuesday morning. ABM account tiers translate them into plays, owners, and budget. The three-tier structure maps onto the classic ABM delivery models: one-to-one, one-to-few, one-to-many.
Tier 1: one-to-one. Roughly your top 5 to 10% of accounts (illustrative; capacity decides). High fit plus live intent. Each account gets a named owner, a researched account plan, custom content such as a bespoke business case, multithreaded outreach across the buying group, executive sponsorship for the biggest bets, and its own ad budget with account-specific creative on LinkedIn. Resourcing is heavy by design: several hours of marketing effort and sustained rep attention per account per month. This tier only works small.
Tier 2: one-to-few. The next 20 to 30%. Strong fit, early or moderate intent. Accounts get clustered by industry or use case into groups of 10 to 25, and each cluster gets a semi-custom play: an industry-specific landing page, a tailored webinar, ad campaigns personalized at segment level, SDR sequences referencing the cluster's shared pain. Per-account cost drops an order of magnitude versus Tier 1, which is the point.
Tier 3: one-to-many. The rest of the qualified list. Good fit, little current signal. Programmatic treatment only: always-on retargeting and paid social against the account list, newsletter nurture, no rep time. Tier 3 is a holding pattern with a tripwire; when an account's intent or engagement score jumps, it gets promoted and a human looks at it.
Two operating rules. Movement between tiers is the system working, so build promotion and demotion into your quarterly review. And resource Tier 1 first, letting lower tiers absorb whatever remains; teams that spread budget evenly end up running mediocre one-to-many everywhere while calling it ABM.
Operationalizing in HubSpot and Salesforce
A model that lives in a spreadsheet dies of neglect within a quarter. Move it into your CRM once the logic stabilizes.
In HubSpot. Store component scores as custom company properties (fit, intent, engagement) and compute them with workflows or calculated properties. HubSpot's target accounts feature gives tiers, buying-role tracking, and account activity a home; scoring-property types and some ABM features gate by subscription level, so check your plan's current limits before designing around them. Build active lists per tier to drive ads audiences and nurture enrollment, and surface tier on the company record so reps see it without hunting.
In Salesforce. Add custom fields on the Account object: component scores, a total, and a Tier picklist. Flows can aggregate contact-level engagement up to the account, though many teams find it simpler to compute scores in their marketing automation or data layer and write results back. Report win rate, deal size, and cycle length grouped by Tier; those reports become your recalibration input. Route Tier 1 accounts to named owners with assignment rules.
Platform specifics change often in both ecosystems, so verify any load-bearing feature against current documentation. And whichever platform you use, automate three things: score refresh on a schedule (weekly is plenty), tier-change alerts to account owners, and decay. Manual scoring survives contact with reality for about six weeks.
ABM benchmarks: how to measure ABM success
"Is it working?" arrives around month three, usually from whoever signed the budget. Account based marketing benchmarks are slippery because programs differ in deal size, cycle length, and tier structure, so treat published numbers with suspicion and your own baseline as the yardstick. Measure in four layers, per tier wherever possible.
Coverage. Do you have the right people identified in each target account? Track contacts per buying-group role and data completeness for Tier 1. Low coverage caps everything downstream, and it is the cheapest problem on this list to fix.
Engagement. Share of target accounts with meaningful activity in the last 30 or 90 days, buying-group breadth per account, and account score trend. Engagement is your leading indicator; pipeline lags it by weeks or months in most B2B motions.
Pipeline. Opportunity creation rate per tier, average deal size per tier, and pipeline velocity, which combines opportunity count, win rate, deal size, and cycle length into one number you can trend. Velocity per tier is the most persuasive chart an ABM program produces, because it shows whether focused effort actually moves deals faster.
Revenue. Win rate per tier, closed-won revenue from target accounts versus a comparable non-target cohort, and expansion revenue over time.
The patterns below are illustrative, offered so you know what "the model is working" looks like. Your own quarter-one baseline should replace them immediately.
| Metric | What a working model shows (illustrative) |
|---|---|
| Tier 1 engagement rate | Clearly above Tier 3, often by a multiple |
| Win rate by tier | Tier 1 > Tier 2 > Tier 3, visibly ordered |
| Average deal size, Tier 1 vs untargeted | Meaningfully larger, since fit selects for it |
| Cycle length, Tier 1 | Shortening quarter over quarter as coverage improves |
| Coverage, Tier 1 buying groups | Approaching complete within two quarters |
The ordering test matters more than any absolute value. If Tier 1 win rate looks like Tier 3 win rate after two or three quarters, your weights are wrong, and the honest move is to rebuild the model against fresh closed-won data. Disappointing, and exactly what measurement is for.
Common mistakes that quietly break account scoring
Scoring on firmographics only. A fit-only model produces a static wish list that never tells you when to act. It also flatters familiar logos. Fit answers "who"; intent and engagement answer "when"; a usable score needs both halves.
Static scores. No decay, no refresh cadence, no tier movement. Within two quarters the list describes last year's market. Decay and scheduled rescoring are unglamorous and they are what keeps the model alive.
Marketing owns the list alone. Sales was absent from the selection workshop, distrusts the tiers, and works gut feel instead. Every hour spent building the model is wasted at that point. Run selection jointly, record vetoes with reasons, and revisit together; the broader discipline of sales and marketing alignment starts with co-owning this one artifact.
Treating a high score as a buying decision. A maxed score means "spend attention here now." Timing, authority, and budget still need human judgment. Reps who get told "this account is ready to close" by a dashboard learn to ignore the dashboard.
Anchoring on vendor benchmarks. Published ABM numbers come from unlike programs with unlike math. Baseline your own first quarter, then compete with yourself.
Skipping negative scoring. Most first models omit it. Scores that only rise stop ranking anything.
FAQ
What is account scoring in ABM?
Account scoring ranks target companies by combining fit (match with your ideal customer profile), intent (signals of active research), and engagement (how the buying group interacts with you) into a single account-level number. The ranking decides which accounts get one-to-one attention, which get segment plays, and which stay in automated nurture.
How is account scoring different from lead scoring?
Lead scoring ranks individual people; account scoring ranks whole companies and their buying groups. In multi-stakeholder B2B deals, one person's behavior rarely represents the group, so the account becomes the unit that predicts revenue.
How do I select target accounts for ABM?
Build the widest reasonable universe from your CRM plus enrichment, score everything for fit, layer intent and engagement onto accounts that pass, then hold a joint marketing-and-sales review to finalize the list. Cut list size to match rep capacity, and rescore quarterly so accounts can move in, out, and between tiers.
How many tiers should an ABM program have?
Three, for most teams: one-to-one, one-to-few, one-to-many. Two tiers force too much of the list into expensive treatment, while four or more blur into routing arguments. Split a tier only when it grows past what its playbook can serve.
What are good ABM benchmarks?
Your own first-quarter baseline, then improvement against it. Cross-company benchmarks mislead because deal size, cycle length, and tier structure vary so much between programs. The pattern to demand from your data: Tier 1 outperforming Tier 3 on engagement, win rate, and deal size, with pipeline velocity trending up per tier.
Can I run account scoring without buying an ABM platform?
Yes. Version one needs a spreadsheet, your CRM export, and your website analytics. Move scoring into HubSpot or Salesforce once the logic stabilizes, and consider a dedicated platform or third-party intent data only after the basic model has proven it predicts outcomes.
Ship version one, then let outcomes argue
A checklist before you build:
- Define your ICP from 18 to 24 months of closed-won data, disqualifiers included.
- Select accounts with the six-step framework, and cut the list to sales capacity.
- Score all three signal families, cap each category, subtract for disqualifiers, decay stale signals.
- Map scores to three tiers, each with a named play, owner, and budget share weighted toward Tier 1.
- Automate refresh and tier-change alerts in your CRM.
- Baseline coverage, engagement, pipeline velocity, and win rate per tier in quarter one; recalibrate quarterly.
Your first model will misjudge some accounts. A rough ranking your sales team trusts and works still outperforms a perfect model nobody opens, so ship it, watch which tiers actually close, and tighten weights against the evidence.
If you want a second set of eyes before committing a quarter of effort to the list, we run a 30-minute review of your target accounts and scoring logic and flag the weights most likely to mislead your reps. Get in touch and bring your closed-won export.