Lead scoring and qualification

What lead scoring and lead qualification are.

Lead scoring gives each lead a number or grade that estimates how likely it is to become a customer. Lead qualification is the decision that follows, where someone or something determines whether a lead meets your criteria and what should happen next. The two get mixed up, but scoring produces an estimate and qualification produces an action.

Most scores combine two kinds of information. Fit describes who the lead is, such as company size, role, location or the service requested. Intent describes what they've done, such as asking for a quote, returning to a pricing page or replying to an email.

When lead scoring makes sense.

Lead scoring is worth setting up when you get more leads than your team can follow up properly, when quality varies widely or when your ad platforms report conversions that sales says are worthless. It also needs a record of what happened to past leads, because a score can't be checked without outcomes to compare it against.

It usually isn't worth the effort when you get a handful of leads a month and someone can simply call all of them. It also won't help much if nobody records outcomes in the CRM, or if slow follow-up is the actual problem.

Rules-based or AI lead scoring.

Rules-based scoring adds and subtracts points according to criteria you've chosen. It's easy to explain and easy to change, and it works well when you already know what a good lead looks like. Its weakness is that the points reflect opinions until someone tests them against outcomes.

Predictive scoring uses past leads and their outcomes to find the patterns that separate those who bought from those who didn't. If your closed deals are few or your CRM is patchy, a predictive model will produce confident numbers with little behind them.

Large language models add something else, because they can read free text. A message that says "we need this for 40 staff before our March audit" tells you a lot, and a points table can't read it. An AI model can pull out those details, check them against your criteria and explain its reasoning for a person to review. It can also misread a message, so we test it on real past leads before anyone relies on it.

We often start with rules, add AI where free text matters and move to a predictive model only when the data can support one.

Sending good leads back to the ad platforms.

Ad platforms optimize toward the conversions you report. If every form submission counts the same, the platform will keep finding people who fill out forms, including those who never buy. Reporting which leads qualified gives it a better target.

Google Ads and Meta both accept this kind of data, through offline conversion imports and the Conversions API. We connect your CRM stages to those events and pick a stage that happens often enough for the platform to learn from and soon enough after the click to be useful.

Whatever you score is what the platform will chase, so a flawed score can steer spending in the wrong direction at scale. We check the score against real sales outcomes before feeding it back. We also send only the fields the event needs, and health details or other sensitive information stay out.

Reliable lead events have to come first. Our Quebec health clinic project shows that kind of foundation, with browser and server-side Meta events verified before launch. That project was tracking work and didn't include lead scoring.

Fairness, privacy and human review.

A score can repeat the biases in your past data. If your team historically ignored a certain kind of lead, a model trained on those outcomes may learn to ignore them too. We review which inputs a score uses and leave out protected characteristics and obvious stand-ins for them.

Scoring also uses personal information, so we agree on what can be read, stored and shared. Quebec's Law 25, for example, has specific requirements when a decision about a person is made solely by automated means. Those questions go to your qualified reviewer.

A low score should change the order of follow-up, and a person should be able to overrule it.

How we measure lead scoring.

The main test is whether high-scoring leads actually become customers at a higher rate than low-scoring ones. We review that by score band, using outcomes from your CRM.

We also look at follow-up speed on top leads, how often your team overrides a score and what share of platform conversions turn into qualified leads. On the media side, we compare cost per qualified lead with cost per form submission, since the two can tell very different stories.