Ask your sales team what a lead score of 150 means. If the answer is a shrug — or worse, "we ignore those" — your lead scoring model isn't working. And you're in good company, because most Account Engagement (Pardot) scoring models were set up once, years ago, by assigning points that "felt right at the time."
Lead scoring is one of the highest-leverage features in Account Engagement. Done well, it tells sales exactly who to call this morning. Done badly, it's noise that trains everyone to distrust the system. Let's do it well.
Quick but critical distinction. Account Engagement gives you two separate measures.
Score measures interest — what a prospect does. Page views, form submissions, email clicks, file downloads. Behavior.
Grade measures fit — who a prospect is. Industry, company size, job title, location. Profile.
A high score with a terrible grade is a student writing a research paper on your website. A high grade with a low score is your dream account that doesn't know you exist yet. You need both, and they need to work together. I've written a full breakdown of scoring vs. grading here — this post focuses on the score side.
Three patterns show up in almost every broken model I audit.
Every action is worth something. When a page view is 1 point and a demo request is 50, but a prospect can rack up 300 points reading blog posts, volume beats intent. The blog-binger outranks the demo-requester.
Nothing ever decays. A whitepaper download from 2023 shouldn't count the same as one from last Tuesday. Without score degradation, your database inflates forever and "hot" stops meaning anything.
Nobody defined the threshold. A score is only useful if something happens at a certain number. If there's no agreed line where marketing hands a lead to sales, the score is trivia.
Pull your last 20–30 closed-won deals and look at what those prospects actually did before entering the pipeline. Not what you assume they did — what they did. Common patterns: visited the pricing page, submitted a contact or demo form, attended a webinar, returned to the site three or more times in a week.
Those are your high-value signals. Everything else is supporting noise.
Keep it simple enough to explain in one slide. A structure that works for most B2B teams:
Use automation rules to reduce scores for inactivity — a common approach is cutting the score meaningfully after 60–90 days without engagement. This keeps "hot" honest and stops your MQL queue from filling with ghosts.
This is the step everyone skips, and it's the one that matters most. Sit down with sales and agree: at what score, combined with what minimum grade, does a lead get routed? What happens next, and how fast? Write it down. An SLA everyone agreed to beats a score nobody believes in.
Once the threshold exists, make Account Engagement act on it: automation rule assigns the prospect, Salesforce gets a task or the lead hits an assignment queue, and sales sees why the lead scored what it did (activity history syncs with the prospect record). If the handoff depends on someone checking a report, it will break by February.
Scoring models aren't set-and-forget. Products change, content changes, buyer behavior changes. Put 30 minutes on the calendar each quarter: Are MQLs converting? Is sales working the leads? Which signals correlated with the deals that closed? Adjust point values based on evidence, not vibes.
The pattern I see over and over: companies who review scoring quarterly trust their model; companies who don't, ignore it. There's not much middle ground.
If your current model is beyond saving, don't tune it — rebuild it. It's faster. Score the five to eight signals that predict revenue, set one decay rule, agree on one threshold with sales, and ship it. You can add sophistication later; you can't add trust back once sales writes the scores off.
And if you'd rather have someone who's rebuilt dozens of these do it with you: scoring and grading architecture is a core part of our Account Engagement retainer. Just getting started with the platform? Apply for our free Pro Bono Quickstart — we set one company up every month, no strings attached.
Founder of Thompson Technology. Salesforce and Account Engagement consultant for B2B companies.
