Lead scoring

Lead scoring answers a practical question: with limited time, who do you start with? It usually rests on two axes. Fit measures how closely the prospect matches your ICP: industry, size, role. Engagement measures what they did: a visit, a reaction, a reply, a download.

Many teams score the two axes separately, and they're right to. A perfect but silent prospect and a very active but off-target prospect shouldn't be handled the same way. Merging them into one number hides the difference.

Marketing automation tools often use points-based scores: so many points for a leadership role, so many for a pricing page visit. There are also predictive scores, calculated by a model trained on your past deals. Those need a lot of history before they can be trusted, which is why most small teams start with a hand-built grid.

Start from your signed customers. Which traits separate them from prospects who never bought? Those are the traits you weight most. If all your best customers have between 10 and 50 employees, company size deserves more weight than industry.

Keep few criteria, each one verifiable. A criterion nobody can fill in from a profile or a website will never be filled in correctly.

Add knockout criteria: competitor, student, a country you don't serve. One of them is enough to set the score to zero.

Above all, link each score band to an action. Above a threshold, priority contact. Between two thresholds, normal queue. Below, no contact. A score that changes no decision is decoration. And keep the reason behind each score: a score with no explanation can't be corrected when it turns out wrong.

Check that high scores really convert better. Group prospects by score band and compare meetings, then signed deals. If the bands look alike, your grid separates nothing.

Be careful about what you expect from a fit score. In the MimikFlow 2026 Observatory, the relevance rating given before contact did not predict who would reply: between a rating of 6 and a rating of 9, the reply rate moved by about two points. The score tells you who is worth talking to. Replies depend on the message, the timing and where the prospect came from.

Source weighed much more in the same study: 74.6% replies from people who had reacted to a post, against 35.6% from strangers found through search. If you want to predict replies, add direct gestures to your grid.

False precision. A fifteen-criteria grid that tells an 87 from an 84 looks scientific without changing a single decision. Nobody handles those two prospects differently.

Points for actions unrelated to buying. Opening an email or looking at a careers page says nothing about wanting your solution.

A threshold set too high. By only contacting the "perfect" prospects, you empty your queue and drop buyers who didn't tick every box.

A grid that is never recalibrated. Your offer and your customers change. Review the grid when the bands stop standing apart in your results.

A score nobody understands. If salespeople can't see why a prospect scored high, they stop trusting the grid and go back to gut feeling.

Before any contact, MimikFlow reads every profile it finds and checks it against your offer, your ideal target and your exclusions. The profile gets a relevance rating, and the reason for that decision is kept. Profiles that are ruled out are not contacted.

Among the prospects kept, those who combine the right profile with a recent signal move to the front. In line with what the Observatory shows, the rating is used to choose who to talk to, not to promise a reply.