Lead scoring helps your team decide who to contact first. Most models fail because they mix two different questions into one number: is this the right kind of company, and are they interested right now? Separating fit and intent makes scoring simpler and far more useful.
Fit: is this the right account?
Fit describes how closely a lead matches your ideal customer profile. It is based on attributes that change slowly:
- Industry and company size
- Location and market
- Tech stack and tools used
- Role, seniority and department of the contact
Intent: are they interested now?
Intent describes behaviour that suggests a buying need. It changes quickly and should decay over time:
- Visiting pricing, service or comparison pages
- Repeat visits in a short period
- Requesting a demo or replying to outreach
- Engaging with nurture emails or attending a webinar
Account-level scoring
B2B decisions are made by groups, not individuals. Several people from the same company may each show a little interest, which looks weak lead by lead but strong for the account as a whole.
Roll engagement up to the account so your team can see when a target company is warming up, even if no single contact has crossed a threshold. This is especially useful for account-based marketing.
Where scoring data comes from
Fit data usually comes from forms, enrichment tools and your CRM. Intent data comes from website tracking, email engagement, outreach replies and, for some teams, third-party intent providers that show which companies are researching a topic.
Check data quality before relying on it. Enrichment can be wrong, website tracking misses visitors who block cookies, and third-party intent data is a probability rather than a fact. Use each source for what it does well and do not let any single signal decide a lead's fate.
Combine them in a simple grid
Score fit and intent separately, then group leads into a grid:
- High fit, high intent: route to sales immediately
- High fit, low intent: target with outbound, ABM and nurture
- Low fit, high intent: review manually, sometimes a new segment
- Low fit, low intent: nurture lightly or ignore
This is easier to explain than a single score and makes the right action for each lead obvious.
An example scoring model
Keep the first version simple. For example:
- Fit: matches target industry (+20), target company size (+20), target role (+15), uses a relevant tool (+10)
- Fit: excluded industry or company size (disqualify)
- Intent: requested a demo (+40), visited pricing (+20), visited a service page twice in a week (+15)
- Intent: replied positively to outreach (+30), attended a webinar (+10)
- Decay: intent points reduce gradually after a few weeks without activity
The exact values matter less than the logic. Start with numbers that reflect what your team believes, then test them against real outcomes.
Fit and intent in practice: two examples
Consider two leads arriving on the same day.
The first is a head of sales at a 200-person software company that matches your ICP. They visited your pricing page twice and read a comparison guide. High fit, high intent: this lead should go to sales within hours, with a note explaining what they looked at.
The second is a marketing coordinator at a 12-person agency outside your target industries who downloaded three guides in one evening. Low fit, apparently high intent. A single score might rank this lead above the first because of the downloads. A fit and intent grid sends it to nurture instead, which is almost always the right call.
Examples like these are the best way to explain scoring to sales. Walk through a handful of real leads and show where each lands and why.
Set thresholds with sales, not for them
The threshold for sending a lead to sales should match the lead definition both teams have agreed. Start with a reasonable guess, then adjust based on how many scored leads sales accepts and converts.
Revisit thresholds whenever sales capacity changes. A team with spare capacity can work leads with lower scores; a stretched team should raise the bar so the best leads are never left waiting.
Let intent decay
A pricing page visit six months ago is not a buying signal today. Reduce intent points over time so scores reflect current interest.
Document the model in a short page that sales can read in two minutes: the fit criteria, the intent signals, the thresholds and what happens at each level. When everyone understands why a lead is prioritised, they are far more likely to act on it quickly.
Automate routing and alerts
Scoring only helps if it triggers action. Use marketing automation and your CRM to route high-fit, high-intent leads to the right owner and alert them in real time. Lower-intent leads can enter nurture flows automatically.
How to test your model
Every quarter, export leads that became opportunities and leads that did not. Compare their scores. If many opportunities scored low, your model is missing a signal. If many high scores went nowhere, it is rewarding the wrong behaviour.
Also ask sales which leads they wished they had seen sooner. Their answers often reveal the next criterion to add.
Scoring outbound leads
Outbound prospects start with high fit and no intent, because you chose them. Their score should rise when they engage, such as replying, visiting your site after an email or accepting a LinkedIn connection. Tracking this in the same model helps your team prioritise follow-up across both channels.
Choosing intent signals that matter
Not all behaviour is equal. Rank signals by how often they appear before real opportunities in your data. Typical strong signals include demo requests, pricing page visits, replies to outreach and repeat visits to service or comparison pages. Typical weak signals include single blog visits, newsletter opens and social media likes.
If you have little historical data, start with the signals sales trusts most and adjust after a few months of results.
Negative scoring
Scores should go down as well as up. Subtract points or disqualify leads for signals that suggest a poor fit or no buying intent:
- Personal email addresses where you sell to companies
- Job seekers visiting careers pages
- Students, competitors and existing customers in sales flows
- Long periods of inactivity
- Unsubscribing or asking not to be contacted
Rolling scoring out to sales
A scoring model fails if sales does not trust it. Involve sales in choosing criteria, explain how the score works in one sentence, and show the reasons behind each score in the CRM, such as "matches ICP, visited pricing twice this week". Start by using scores to prioritise, not to block leads, and let sales flag scores that look wrong.
After a month, review together: which high-scoring leads were good, which were not, and what was missing. That conversation improves both the model and the relationship between the teams.
Scoring and outbound timing
Scores are also useful for deciding when to start outreach. When a high-fit account crosses an intent threshold, automation can alert the account owner or start a relevant sequence. This is one of the clearest ways inbound signals can make outbound more timely and relevant.
Implementing scoring in your CRM
Most CRMs, including HubSpot and Salesforce, support scoring natively. Build fit and intent as two separate properties so both are visible, then create a third property or view for the combined priority. Keep the criteria documented next to the model so anyone can see why a lead scored the way it did.
When native scoring is not flexible enough, enrichment tools and automation platforms can calculate scores and write them back to the CRM. Our guide to Zapier, Make and n8n covers how to choose a tool for that kind of workflow.
Prioritising accounts, not just leads
For account-based teams, the most useful view is a ranked list of target accounts by combined fit and engagement. Review it weekly with sales and outbound: which accounts are warming up, who from the buying committee is engaged, and who should reach out next. That turns scoring from a background number into a weekly plan.
Common mistakes
- Too many criteria, so nobody understands the score
- Rewarding email opens heavily, which are unreliable signals
- Never reviewing the model against real outcomes
- Scoring individuals but ignoring account-level engagement
Review your model every quarter against closed-won deals. A scoring model is a hypothesis about what predicts revenue, and it should be tested like one.