Lead scoring has a reputation problem. Plenty of companies have a scoring model somewhere in their CRM, but sales ignores it, because experience has taught them that a high score doesn't mean a good lead. When that happens, every lead gets treated the same, and the most valuable prospects wait in the same queue as everyone else.
A scoring model sales actually trusts is built on two separate questions, is transparent about how points are earned, and is recalibrated with real outcomes. Here's how to build one, with example criteria you can adapt.
Two questions, two scores
The most common mistake is mixing everything into one number. A lead from a perfect-fit company who has barely engaged and a student who has downloaded every guide you've published can end up with the same score, for completely different reasons. Instead, score two things separately:
- Fit score: how closely does this organisation match your ideal customer profile? Built from firmographic, demographic and enriched company data.
- Engagement score: how strongly is this lead showing buying intent? Built from marketing interactions and, if you have a product-led motion, product usage.
A lead becomes an MQL when both scores pass agreed thresholds. High fit with low engagement means nurture; high engagement with low fit might mean a self-serve route; high on both means call them now.
Example fit criteria
Fit scoring rewards the attributes of your best customers and penalises the signals of poor fit. The values below are illustrative starting points:
| Criterion | Points |
|---|---|
| Business email domain | +10 |
| Team size or seats required above 50 | +30 |
| In a target industry | +15 |
| Decision-maker job title | +20 |
| Interested in multiple products | +10 |
| Team size below 10 | −10 |
| Personal email domain | −25 |
Negative scores matter as much as positive ones. They stop obviously poor-fit leads from reaching sales just because they've been busy on your website.
Example engagement criteria
Engagement scoring weights actions by how strongly they signal intent. A pricing page visit means more than opening a newsletter.
| Action | Points |
|---|---|
| Demo or enterprise enquiry | +20 |
| Started checkout | +20 |
| Pricing page visit | +15 |
| Gated resource download | +15 |
| Requested sales support | +10 |
| Resource or article page visit | +10 |
| Marketing email interaction | +5 |
| No activity for more than 30 days | −20 |
The inactivity penalty is important. Engagement should decay over time; a burst of interest six months ago isn't intent today.
Add product-level scoring for trialists
If businesses can sign up for a free trial, some of your best leads are already in the product. Their usage is the strongest intent signal you have, so add product-specific engagement scores alongside the global one. For example:
| Product A | Product B | Product C |
|---|---|---|
| Connected their data (+10) Installed mobile app (+10) Installed desktop app (+5) Completed first key action (+5) | Created first project (+10) Connected an integration (+10) Installed mobile app (+10) Installed desktop app (+5) | Uploaded first file (+10) Shared with a colleague (+10) Installed mobile app (+10) Installed desktop app (+5) |
Product-level scores do more than qualify leads. They show which products a customer has activated, which powers smarter segmentation and cross-sell, and they identify the key product events you'll need for later lifecycle automations.
Setting the MQL threshold
There's no universal right number. Start by scoring your historical leads against the model and looking at where your past customers would have landed. Set the initial threshold so that the leads above it convert meaningfully better than those below it, and so the volume is something sales can actually handle within the SLA.
Then treat the threshold as a hypothesis, not a fact.
Recalibrate with sales
A scoring model earns trust by being right, and by changing when it isn't. After the first two or three months:
- Compare score distribution with outcomes. Do higher-scoring leads really convert to opportunities and customers at higher rates?
- Review rejection reasons. If sales rejects lots of MQLs as "too small", your fit criteria need adjusting.
- Look for missed winners. Customers who never reached MQL point to signals your model doesn't capture.
- Adjust together. Change point values and thresholds with sales in the room, so they understand and own the model.
Common pitfalls
- Scoring everything. Dozens of tiny scoring rules make the model impossible to understand. Focus on the signals that genuinely predict conversion.
- Never subtracting points. Without negative scores and decay, every long-standing contact eventually becomes an MQL.
- Hiding the logic. Show sales why a lead scored highly. Transparency is what turns a number into a reason to pick up the phone.
- Ignoring the handoff. A perfect score is wasted if the lead then waits a week. Pair scoring with a clear marketing-to-sales handoff and SLA.
Lead scoring is a core part of the Acquire stage in our lifecycle marketing framework. If you'd like help designing a model for your business, book a free consultation.