August 10, 2026
You inherited a lead scoring model nobody's touched since HubSpot rolled out the fit/engagement split in 2024, and sales stopped trusting the "hot lead" flag. Nobody flagged it as broken, reps just started ignoring it and building their own workarounds instead, which is usually the first real sign something's wrong.
This is the audit to run before that distrust becomes permanent. We’ll explore six technical checks that catch most broken scoring setups, the account-level rollup most teams miss entirely, and how to triage a quick fix from a setup that needs a full rebuild.
We run this exact audit on portals that have been live for years, and the same handful of settings are almost always the culprit. Cross-check your criteria against this lead scoring worksheet first. It's a fast way to confirm the model still means what you think it means before you go looking for what broke it.
A HubSpot lead scoring audit checks six things: property groups versus event groups, the fit and engagement split, cross-object rules, decay settings, group point caps, and exclusion lists instead of negative scoring. Most broken scoring setups fail two or three of these, not all six, which is why a focused audit usually finds the fix fast.
Signs Your HubSpot Lead Scoring Is Already Broken
A few signs show up before anyone officially declares the scoring model broken. Reps stop acting on the "hot lead" flag and build their own manual process instead. Scores climb and never come back down, because there's no decay built in. One channel or event type can push a contact's score past the threshold on its own, because there's no cap on how many points a single group can contribute. Nobody remembers the last time the model was checked against actual closed-won data.Your thresholds don't come close to a clean 100-point scale per score type. If fit tops out at 40 or engagement runs past 300, someone changed the math along the way without anyone recalibrating what 'hot' actually means.
The Technical Audit Checklist
Run through these six checks, in roughly this order.
- Property groups versus event groups. Static attributes like job title and company size belong in a different group than behavioral events like page views and email clicks. Scoring both the same way blurs fit and intent together. Behavioral events should also carry a frequency and a timeframe, not just a flat yes or no. A pricing page viewed twice in the last 30 days signals something a single visit two years ago never will.
- Fit and engagement split. Confirm the model still uses two distinct scores. It's common for someone to quietly recombine them into one number over time, usually without meaning to.
- Cross-object rules. Check whether scoring accounts for activity happening at the company or deal level, not just the contact record. A buying signal on the company object should be able to move a contact's score.
- Decay settings. Scores should drop when engagement stops. Without decay, every contact just accumulates points indefinitely, and stale contacts end up looking as hot as active ones.
- Group point caps. Set a ceiling on how many points a single property group or event type can contribute, so one channel, like repeated email opens, can't flood a contact past the threshold on its own.
- Exclusion lists instead of negative scoring. Disqualifying contacts (competitors, students, unsubscribes) should be excluded outright rather than assigned negative points. Negative scoring rarely nets out cleanly against genuine engagement.
The Gap Most Teams Miss: Account-Level Scoring
Contact-level scoring alone misses how B2B buying actually works in healthcare or enterprise SaaS, where five or six people from the same account touch your content before anyone books a call. An account-level rollup catches buying committee activity that no single contact's score will ever show on its own.
Who Owns This Once It's Fixed
A scoring model without an owner drifts back to broken within a year. Assign one person or team responsible for changes, set a recurring cadence for reviewing thresholds against actual closed-won data (quarterly is reasonable for most teams), and use HubSpot's preview-distribution feature to test how a change would reshuffle your current contact list before it goes live. When thresholds come up for review, a simple grid beats a spreadsheet argument every time, fit on one axis, engagement on the other, with color zones marking hot, warm, and cold. It's the fastest way for whoever owns this to see if the model still lines up with reality.
Quick Fix or Full Rebuild? How to Triage What's Broken
Not everything on the checklist above requires a rebuild. Missing decay or missing point caps are usually a quick fix with a few changes to settings inside the model you already have. A missing fit and engagement split, or a model still scoring everything at the contact level with no account rollup, usually means the underlying structure needs to be rebuilt, not patched. Run the checklist first, then decide which bucket you're in before committing to either path.
FAQ
How do you audit HubSpot lead scoring?
Check six things: whether property groups and event groups are scored separately, whether fit and engagement remain two distinct scores, whether cross-object rules pull in company and deal activity, whether scores decay over time, whether point caps exist per group, and whether disqualifying contacts are excluded outright instead of negatively scored. Most broken setups fail two or three of these, not all six.
What is HubSpot's preview-distribution feature for lead scoring?
It shows how a proposed scoring change would reshuffle your current contact list before you publish it, how many contacts would move above or below a given threshold. Running a change through preview distribution first catches unintended shifts, like flooding sales with newly "hot" leads, before they hit a live pipeline.
Should lead scoring roll up to the account level or stay at the contact level?
For B2B buying committees, like healthcare or enterprise SaaS, account-level rollup usually matters more than any single contact's score. Individual buying signals from five different people at the same company can each look unremarkable alone. Combined at the account level, they show real intent that contact-level scoring alone would miss.
Relevant Blog Posts
Top 26 HubSpot Hacks for Marketing Pros
If you've invested in HubSpot Marketing Hub, you know the promise: powerful growth, streamlined processes, a happy team. But let's be real – there's a TON of info out there on HubSpot. Sorting...
Marketing Collateral: The Ultimate Guide (With Examples)
When we’re talking about marketing, things can get kind of buzz-wordy. Marketing collateral. Landing pages. Audience personas. Ever feel like you need a guide to walk you through these things?...
Using Content for Top-of-Funnel Marketing (Plus Examples)
What does your content have to say? Does it answer your audience’s most burning questions? Does it bring them value after they read, watch, or listen to it? Delivering value via content is especially...
