Lead Scoring

Lead scoring is a model that assigns numeric scores to leads based on firmographic fit and behavioral intent signals, so sales teams know which leads to call first and which to let warm up longer.

How It Actually Works

A lead score is typically a composite of two dimensions: fit and intent. Fit covers who the lead is: company size, industry, tech stack, job title, annual revenue. Intent covers what they have done: visited pricing pages, downloaded a comparison guide, opened four emails in a week, searched a relevant keyword on G2.

Most teams build scoring in their MAP or CRM, assigning point values to each signal. A VP of Engineering at a 200-person SaaS company might start at 40 points for fit. Add a pricing page visit (15 points) and two opened sequences (5 points each) and that lead clears the threshold for an SDR call. A lead with the same behavior but from a 15-person company in a vertical you don’t sell into stays in nurture.

More sophisticated setups layer in predictive scoring, where a machine learning model trains on closed-won data to weight signals automatically rather than relying on manual point assignments.

Why It Matters for B2B Revenue Teams

Without scoring, SDRs default to recency: whoever just filled out a form gets the call, regardless of fit. That burns capacity on leads that will never close while real buyers go cold.

Good scoring changes the sequencing math. If your SDR team can work 80 leads per week and you have 400 in queue, scoring tells you which 80 are worth the effort this week. Teams with a calibrated model consistently see higher connect-to-opportunity rates, sometimes 2x to 3x compared to unscored queues, simply because reps spend time on the right names.

Scoring also feeds lead routing. A high-score enterprise lead routes to an AE directly. A mid-score SMB lead routes to an SDR sequence. A low-score lead with high intent but poor fit goes to a low-touch nurture track. Without scores, routing logic breaks down fast.

Common Mistakes

The most common mistake is building a model once and never touching it again. Signals decay. A pricing page visit meant something different in 2021 than it does now that buyers do more research before ever hitting your site. If your win rates are dropping but score distributions look the same, your model is stale.

Second mistake: weighting demographic fit too heavily over intent. A Fortune 500 logo with a perfect ICP profile but zero engagement is not a hot lead. Engagement signals often predict near-term pipeline better than firmographics alone.

Third: not connecting scoring thresholds to actual pipeline outcomes. Teams set a score threshold of 80 because it felt right, not because they analyzed what score range historically converts to opportunities. Run the analysis. Anchor your thresholds to data, not instinct.

Finally, lead scoring only works if your underlying data hygiene is clean. Scores built on incomplete or stale records produce false confidence. Garbage in, garbage out applies here more than almost anywhere else in the stack.

How It Connects to Adjacent Concepts

Lead scoring sits at the intersection of data enrichment (which populates the fit fields your model needs), lead routing (which acts on the score), and CRM architecture (which determines whether scores are visible, actionable, and tied to the right objects). If any one of those three breaks, your scoring model produces numbers that nobody acts on.

At Phi, scored leads feeding into an instrumented routing layer is a baseline assumption when we build GTM pods, not a nice-to-have.