How to score lead quality (without a data scientist)

"Lead scoring" has a scary reputation. It sounds like something that needs a data team, a machine-learning model and three months of meetings. It doesn't. The truth is that a genuinely useful lead score is something you can sketch out this afternoon with a spreadsheet and the CRM fields you already collect — no statistician required.

The reason it's worth doing isn't to win a tidy report. It's that once every lead carries a sensible number, you can hand that number to your ad platforms and let them chase the leads that actually become customers. Let's build the simple version.

What a lead score really is

Strip away the jargon and a lead score is just a number that says how much a lead is worth to you. Higher number, more worth pursuing. That's it.

The cleanest way to define that worth is a short formula: the probability the lead will close × what they're worth if they do. In shorthand that's P(close) × LTV — "the chance they buy" times "their lifetime value". Both halves matter. A lead that's very likely to buy but only spends €50 isn't worth much. A lead worth €50,000 that has a 1% chance of ever signing isn't either. You want leads that score well on both.

You don't need exact figures to start. Rough, honest estimates beat no estimate at all. The point is to replace "this lead feels good" with a number you can compare, sort and act on.

The fields that already predict value

Here's the reassuring part: the information that predicts a good lead is usually sitting in your CRM already. You're just not using it as a score yet. The fields that tend to do the heavy lifting:

• Deal size or stated budget — the single strongest signal of value when you have it.
• ICP fit — how closely the lead matches your ideal customer profile (the type of buyer you win and keep most often).
• Company size — headcount or revenue band; bigger firms often mean bigger, stickier deals.
• Industry — some sectors close faster and spend more with you than others.
• Lifecycle stage — a "requested a demo" lead is worth more than a "downloaded a PDF" lead.
• Source — a referral or a high-intent search behaves very differently from a cold giveaway entry.

You don't need all of them. Three or four reliable fields are plenty for a first version. If you're still weighing whether to chase more leads or better ones, our piece on lead quality vs lead volume is a good companion read.

Build a simple model

Now turn those fields into a number. The mechanism is deliberately boring: assign points or multipliers to each field and add or multiply them up. No machine learning anywhere.

For example, you might start every lead at a base value and adjust: enterprise company size ×3, a target industry ×1.5, a referral source ×2, a low-intent source ×0.4. A small business that came in via referral and asked for a quote lands high; an anonymous freebie download from outside your market lands low. Run a few real leads through it and sanity-check the order against your gut — if a lead you know was great scores low, fix the weights.

The goal you're aiming for isn't precision to the cent. It's spread. You want your scores to fan out — some leads at €40, some at €2,000, plenty in between — so that the leads are clearly distinguishable from one another. A model where every lead scores roughly the same is useless, because nothing downstream (including an ad platform) can tell the good ones from the bad ones. Spread is the feature. If you want a fuller walkthrough of building one of these, see the complete guide to value-based bidding for SMBs.

Common mistakes to avoid

A few traps catch almost everyone the first time:

• Using fields that "leak." It's tempting to score on data that only exists after a sale — final contract value, number of renewals, support tickets. Those make your model look brilliant in hindsight but are useless in practice, because that information doesn't exist at the moment a new lead arrives. Only score on what you know up front.
• Over-engineering. Twenty fields, decimal-place weights and a colour-coded rulebook will not beat four good signals. Complexity slows you down and makes the model impossible to trust. Start simple, ship it, refine later.
• Scoring on vanity signals. Email opens, page views and "engagement" feel like quality but rarely predict revenue. Plenty of people who never buy will happily open your emails. Weight things that move money, not things that move dashboards.

Put it to work

A score sitting in a spreadsheet doesn't grow revenue. The payoff comes when you feed that value back to the platforms where you buy traffic — Meta, Google and LinkedIn.

Out of the box, those platforms optimise for whatever you tell them counts — usually a raw form-fill, where a €40 lead and a €2,000 lead look identical. When you send the actual value of each lead instead, their bidding starts steering budget toward the sources, audiences and creatives that bring in your high-value leads, and away from the cheap junk. That's the entire idea behind value-based bidding, and it only works if every lead carries a sensible number — which is exactly what you just built.

This is the job PipeValue handles for you: it reads the real € value of each lead from your CRM and sends it to Meta, Google, LinkedIn and TikTok automatically, so the algorithms optimise for revenue rather than raw clicks — no manual wiring, no data team.

FAQ

Do I need a data scientist to score leads?

No. A useful first version is just a points table built from CRM fields you already have — deal size, company size, industry, source. You can build it in a spreadsheet in an afternoon. A data scientist can refine it later, but you don't need one to start getting value.

What does P(close) × LTV mean?

It's the clean definition of a lead's worth: the probability that the lead will become a customer, multiplied by what they're worth if they do (their lifetime value). A lead with an 18% chance of closing and a €3,600 lifetime value is worth about €650. It turns a gut feeling into a number.

Which CRM fields should I use?

Start with fields that are known at the moment a lead arrives and that genuinely predict value: deal size or budget, company size, industry, ICP fit, lifecycle stage, and lead source. Avoid anything that only exists after a deal closes, because that information isn't available when you need to score.

How do I send the score to my ad platforms?

You pass the lead's value back to Meta, Google and LinkedIn as a conversion value, so their bidding optimises for high-value leads instead of raw form-fills. Tools like PipeValue read the value from your CRM and send it automatically, so you don't have to wire it up by hand.