Account Scoring Model: How to Build One for B2B in 2026
TL;DR
An account scoring model ranks target companies by how likely they are to buy, using three inputs: fit, intent and engagement.
Start with a simple points model on a 0 to 100 scale, test it against your won and lost deals, and agree on what each tier triggers for sales and marketing.
Score the buying group, not one contact: three engaged people at one account say more than one very active person.
Move to predictive account scoring only when you have enough closed deals to train a model.
A score only creates pipeline when someone acts on it. Decide the action for each tier before you launch.
What Is an Account Scoring Model?
An account scoring model is a set of rules that gives each target company a score for how well it fits your business and how ready it is to buy. Sales and marketing use the score to decide which accounts get attention first.
Most account scoring models combine three inputs:
Fit: how closely the company matches your ideal customer profile, such as industry, size, region and technology stack. Fit changes slowly.
Intent: whether the company is researching your category right now, from third-party research data or your own website. See our guide to intent data for B2B.
Engagement: how people at the company interact with your brand: ad engagement, website visits, event attendance and replies to sales.
Fit tells you whether an account is worth winning. Intent and engagement tell you whether now is the time.
Account scoring starts after you have chosen which companies to pursue. Our guide to target account lists sales will work covers how to pick those accounts. This page covers how to rank them week to week once the list is live.
Account Scoring vs Lead Scoring
Lead scoring gives points to individual contacts. Account scoring gives points to the company and adds up what everyone at that company does.
Lead scoring | Account scoring | |
|---|---|---|
What gets a score | One person | One company |
Main question | Is this person ready to talk to sales? | Is this company ready to buy? |
Typical signals | Form fills, email clicks, job title | Fit, intent, and engagement across several people |
Best for | High-volume inbound with short cycles | Long B2B sales cycles with buying groups |
Common failure | One curious junior contact looks like a hot lead | Fit looks strong but nobody at the account is engaging |
The difference matters because B2B purchases involve several people. A typical B2B buying group includes 6 to 10 stakeholders, according to Gartner. One person downloading three reports is a weaker signal than three people from the same company each engaging once.
Account-based lead scoring combines the two. Contacts still get lead scores, but the account score decides priority, and the contact scores show sales who to talk to first inside a high-scoring account. For how lead stages fit in, see MQL vs SQL: B2B lead definitions that matter.
How to Build an Account Scoring Model: Step by Step
Step 1: Start From Your Won and Lost Deals
Pull your closed-won and closed-lost deals from the last 12 to 24 months. List what the won accounts had in common before they bought: industry, size, region, technology, and the activity you saw from them in the months before the deal opened.
Do this with sales in the room. A model marketing builds alone tends to reward the activity marketing can see, not the signals sales trusts.
Step 2: Choose Your Fit Criteria
Turn your ICP into scored criteria. Give the most points to the attributes most of your best customers share, and fewer points to nice-to-have attributes.
Add negative points or an outright filter for disqualifiers, such as a company that is too small, in an unsupported region, or locked into a competitor's multi-year contract.
Step 3: Choose Your Intent and Engagement Signals
Pick a short list of signals you can actually collect. Five good signals beat twenty noisy ones. Strong signals are usually close to a buying decision: pricing or comparison page visits, several people from one account engaging, a demo request, or a research surge on your category.
Give each signal a time window. Engagement from the last 90 days and intent from the last 30 days is a reasonable starting point. Older activity should drop out, or the score keeps rewarding accounts that went quiet months ago. For which signals tend to predict pipeline, see buying signals that predict pipeline.
Step 4: Weight the Categories and Set a Scale
Most teams use a 0 to 100 scale and split it between the three categories. Fit usually carries the most weight, because an account outside your ICP rarely becomes a good customer however active it is.
A common starting split is 50 points for fit, 30 for engagement and 20 for intent. Adjust it to what your deal history shows. If your won deals almost always had heavy engagement before the opportunity opened, give engagement more weight.
Step 5: Set Tiers and Agree on the Action for Each
A score without an action changes nothing. Agree on score ranges and on what happens in each, before you launch:
Tier | Score | Marketing action | Sales action |
|---|---|---|---|
A | 75 to 100 | Person-level ads and content to the full buying group | Account owner reaches out within a week |
B | 50 to 74 | Ads and content to warm the account | Watch for movement into Tier A |
C | Below 50 | Light-touch nurture | No active outreach |
Add one rule on top: an account with a fit score below 25 stays in Tier C, whatever its engagement. That stops a busy but poor-fit account from taking sales time.
Step 6: Test the Model, Then Review It Every Quarter
Before you launch, score your past won and lost accounts with the new model. Won accounts should score clearly higher than lost ones. If they do not, change the weights before sales sees a single score.
After launch, review the model every quarter with sales. Look at Tier A accounts that never turned into opportunities, and deals that came from Tier C accounts. Both show where the model is wrong.
Example Account Scoring Model
This example is for a B2B company selling to mid-sized industrial firms. Use it as a template and change the criteria and points to match your own deal history.
Signal | Category | Points |
|---|---|---|
Industry matches core ICP | Fit | 20 |
Company size in target range | Fit | 15 |
Located in a market you serve | Fit | 10 |
Uses a complementary system | Fit | 5 |
Category research surge in the last 30 days | Intent | 10 |
Pricing or comparison page visit in the last 30 days | Intent | 10 |
3 or more people engaged with your ads in the last 90 days | Engagement | 10 |
Senior decision-maker engaged in the last 90 days | Engagement | 10 |
Event attendance or webinar sign-up in the last 90 days | Engagement | 5 |
Reply to sales outreach in the last 90 days | Engagement | 5 |
Current customer of a competitor with a long contract | Disqualifier | -20 |
Maximum | 100 |
In this example, fit adds up to 50 points, intent to 20 and engagement to 30.
Two accounts show how it works:
Account 1: matches industry, size and region (45 fit points), had a pricing page visit (10 intent points) and four people engaged with ads, including a director (20 engagement points). Score: 75. Tier A.
Account 2: an ideal fit on all four criteria (50 fit points), but no intent and no engagement. Score: 50. Tier B: worth warming with ads and content before sales reaches out.
How Account Scoring Keeps Sales and Marketing Aligned
An account scoring model gives both teams the same priority list and the same definition of "ready". Marketing warms Tier B accounts until they move up. Sales works Tier A accounts while marketing keeps them engaged. Neither team has to argue about lead quality, because the tiers and actions were agreed upfront. For the full operating model (shared definitions, targets, data and meetings), see our sales and marketing alignment playbook.
Predictive Account Scoring: When to Use It
Predictive account scoring replaces the hand-set points with a machine learning model. The model learns from your won and lost deals which attributes and activities came before an opportunity, then scores every other account against that pattern.
It needs data. Demandbase, for example, recommends training its Pipeline Predict model on at least 50 accounts with qualified opportunities from the last 12 months (Demandbase). With fewer deals than that, a points model is easier to explain to sales and easier to fix when a score looks wrong.
Many teams run both: a points model that sales understands, and a predictive score that ranks accounts inside each tier. For how the models work and when they pay off, see our guide to predictive lead scoring for B2B.
Account Scoring Software
You can run a first model in a spreadsheet. Once it works, move it into the system sales already uses, so the score updates on its own.
Tool | What it scores | Approach | Best for |
|---|---|---|---|
Companies, contacts and deals | Rule-based fit, engagement or combined scores built in the lead scoring tool. Company scores are available in Marketing Hub or Sales Hub | Teams on HubSpot that want a points model inside the CRM | |
Accounts and contacts | Predictive account fit and buying stage (Awareness, Consideration, Decision, Purchase), trained on each customer's own deal history | Enterprise teams that want buying-stage prediction and intent data in one platform | |
Accounts | Qualification Score (likelihood to become a customer), Pipeline Predict (likelihood to become an opportunity soon) and engagement minutes | Enterprise ABM teams that want predictive scores next to account advertising | |
Leads, contacts and accounts | Predictive Customer Fit and Likelihood to Buy models. The Account Fit score is pushed to Salesforce accounts | B2B SaaS teams on Salesforce that want predictive fit scoring |
Features and plans may change. Always check the latest details on the vendor's website.
For more predictive scoring tools, with pricing and trade-offs, see our comparison of the best MadKudu alternatives for predictive lead scoring.
Common Account Scoring Mistakes
Scoring without an action. If Tier A and Tier C accounts get the same follow-up, the model has no effect. Agree on the action for each tier first.
Letting engagement outweigh fit. A very active account outside your ICP is still a poor customer. Cap low-fit accounts in the lowest tier.
Counting one person as the account. One junior contact reading everything does not mean the buying group is engaged. Give extra points for several people, and for senior roles.
Never letting points expire. Activity from six months ago keeps inflating scores. Use time windows.
Using too many signals. Twenty signals with small points are hard to explain and hard to fix. Start with five to ten.
Building it without sales. If reps do not trust how the score is calculated, they will work their own lists instead.
Where Hey Sid Fits
Hey Sid is not an account scoring tool. It is a managed ABM service that adds engagement data to your scoring model and acts on the accounts it surfaces.
Engagement data in HubSpot: Hey Sid's HubSpot integration writes ad engagement properties onto existing company records, so you can use them as criteria in a HubSpot company score.
Influenced pipeline: Revenue Reporting syncs company-level impressions, engagement and clicks into HubSpot and reports influenced pipeline and influenced revenue. That shows whether your Tier A accounts are turning into pipeline. See why influenced pipeline is the B2B KPI that matters.
Action on Tier A and B accounts: Always On runs ads to named decision-makers at your target accounts, Authority Builder keeps your experts visible on LinkedIn, and Precision Connect runs LinkedIn outreach to the same people. The sequence is designed so that by the time sales reaches out, the buyer has already seen your brand.
Risk Ident reported 2.5x shorter sales cycles and 40% higher engagement with this approach (client-reported). See the Risk Ident case.
Book a demo: heysid.com/demo
FAQ
What is an account scoring model?
An account scoring model is a set of rules that gives each target company a score, usually from 0 to 100, based on how well it fits your ideal customer profile and how much buying intent and engagement it shows. Sales and marketing use the score to decide which accounts to work first.
What is the difference between account scoring and lead scoring?
Lead scoring rates individual contacts. Account scoring rates the whole company and adds up the activity of everyone who works there. In B2B sales with buying groups, account scoring usually gives a more reliable picture of which deals are real.
What should an account scoring model include?
Three categories: fit (industry, size, region, technology), intent (research activity and high-intent page visits) and engagement (ad engagement, events, sales replies, and how many people at the account are involved). Add negative points for disqualifiers, and time windows so old activity drops out.
How do you weight an account scoring model?
Start with your won and lost deals. A common starting split is 50% fit, 30% engagement and 20% intent, then adjust so that your past won accounts score clearly higher than your lost ones.
What is predictive account scoring?
Predictive account scoring uses a machine learning model trained on your won and lost deals to score accounts, instead of points set by hand. It works best when you have enough closed opportunities to train on. With fewer deals, a rule-based model is easier to explain and maintain.
How often should you update an account scoring model?
Review it every quarter with sales, and sooner if your ICP, product or market changes. Check whether Tier A accounts are turning into opportunities, and whether deals are coming from accounts the model scored low.
Sources
Related: Predictive Lead Scoring for B2B | Target Account Lists Sales Will Work | Intent Data for B2B
