Your Lead Score Is Probably Wrong. Here's What to Do About It.
This week I had four separate client conversations that touched on lead scoring. Four. Different companies, different industries, different HubSpot portals. And every single one of them had the same underlying question: why aren't we getting better leads?
The answer, in every case, wasn't "you need more leads." It was "you need to trust the ones you already have."
The Thing Everyone Gets Wrong
A lot of teams build a model that adds up job title + geography + a few clicks, and when the number hits a threshold, the lead gets flagged as an MQL and dropped on a sales rep's desk.
The problem? In every case I reviewed this week, the model was conflating two very different things: interested and qualified. Those are not the same question. And a single score can't answer both.
Here's a real example. A client building a lead scoring model for a B2B company noticed that a junior-level contact could download a white paper, pick up points for being in the right geography, and hit their MQL threshold... without sales ever wanting to talk to them.
That's not an MQL. That's a curious person with a PDF.
The model was tracking engagement (did they do stuff?) but not fit (are they actually who we want to talk to?). And mixing both into one number is where the model breaks down. Sales gets a lead that "scored well," takes one look, and ignores it. Do that enough times and the whole team stops trusting the system. Then everyone blames each other.
The Fix: Two Scores, Not One
What I landed on with several clients this week (and this is something you can build in HubSpot right now) is a two-score model. Instead of one number that needs to hit a threshold, you build two separate scores:
Fit score. Based on static, demographic data. Company size, industry, job title, geography, named target accounts. Things that don't really change. This answers: is this person who we want to talk to?
Engagement score. Based on behavioral data. Emails opened, pages visited, forms submitted, content downloaded. Things they actually did. This answers: are they paying attention to us?
A lead only becomes an MQL when they hit both thresholds.
That junior contact who downloaded a white paper? Solid engagement, low fit. They don't cross the line. A VP at a named target account who just joined your email list? High fit, low engagement. They don't get prematurely handed to sales either... but now you know to watch that record.
This is exactly how HubSpot's scoring tool is designed to work. It splits fit and engagement into two separate scores, each with a maximum of 100 points, and plots contacts on a matrix (A/B/C for fit, 1/2/3 for engagement). An A1 is your best lead. A C3 isn't worth the time right now. And a B1 (medium fit, high engagement) is a very different conversation than an A3 (great fit, not engaging). The matrix makes the nuance visible. A single number buries it.
We wrote a full breakdown of how to build this in HubSpot (Lead Scoring 101: A Practical Guide to Fit, Engagement, and the Handoff) if you want the step-by-step. But today I want to talk about what happened when one client actually used it.
Lead Scoring Tells You More Than WHO. It Tells You HOW.
Here's where it got interesting this week.
One of my clients, a manufacturing-focused agency, has been running lead scoring for a few months now. Their threshold sits around 32 to 33 points. Not a huge number. But it's calibrated to their ICP: established manufacturing companies, decision-maker-level contacts, specific firmographic signals.
When a contact hits that threshold, they don't get dropped into an email sequence.
Their head of business development picks up the phone and calls them.
And here's what he told me: "You wouldn't believe how many people say, 'Thank God you called. You're one of the few people who still picks up the phone.'"
That's the insight a lot of teams miss. Lead scoring isn't just about automating your outreach. It's about knowing when a lead is warm enough to justify a human touchpoint. And in some industries (especially manufacturing, professional services, healthcare), that touchpoint isn't an email. It's a call.
We've spent the last decade optimizing sequences and automations because human attention is expensive. But the pendulum has swung so far toward automation that a real phone call is now a differentiator. Direct mail is making a comeback. Conferences are packed again. The "old school" stuff works right now because everything else is automated.
Lead scoring is the thing that tells you when it's worth picking up the phone.
Three Quick Audits for Your Portal
If you're running lead scoring today, here's what to check:
Do you have one score or two? If it's one, it's combining fit and engagement in a way that makes each harder to trust. Separate them. HubSpot gives you two scores for a reason.
Does geography alone get a contact halfway to MQL? We see this all the time. Location is a supporting signal, not a primary one. If someone in the right zip code can score high without being the right person or doing anything meaningful, your model needs reweighting.
When a lead hits your threshold, what happens next? If the answer is "they get added to a sequence," ask whether some of them deserve a phone call instead. Especially if your buyers skew senior or traditional. The automation is there to handle volume. But the leads that actually close? They often close because a human showed up at the right moment.
The Gut Check
Here's a test we use with every client. Ask your sales team to describe what a "good" lead looks like using your scoring model. If they can't, or if their description doesn't match what the model is actually measuring, the model isn't built around their definition of good. Start there.
Lead scoring isn't broken as a concept. But one-size-fits-all scoring (one number, one threshold, one automated response) almost always ends up as a number nobody trusts. Split fit from engagement. Set real thresholds on both. And then figure out what the right action is when a lead crosses that line.
Sometimes it's an automated email. Sometimes it's a 90-second phone call from a human who says "hey, I saw you were checking us out, wanted to reach out personally."
Turns out that's what some people have been waiting for.
If you want help building a scoring model that your sales team will actually use, or if you've got one running and want a second set of eyes on whether it's doing what you think it's doing, that's what we do.