HubSpot lead scoring adds and subtracts points on contacts, companies and deals using rules you write. Fit criteria score who a record is; engagement criteria score what it has done. It needs a Professional or Enterprise plan. The scores are only as good as those rules, and HubSpot doesn't check them against the deals you won.
That last sentence is the whole post, really. Everything else is how the tool works, checked against HubSpot's own knowledge base in November 2026, so you can set it up properly and know where its edges are.
Which HubSpot plans and records can you score?
Lead scoring needs a Professional or Enterprise seat on Marketing Hub or Sales Hub. It isn't on Free or Starter.
What you can score depends on the hub:
| Record | Which hub | Score types available |
|---|---|---|
| Contacts | Marketing Hub only | Fit, engagement or combined |
| Companies | Marketing Hub or Sales Hub | Fit, engagement or combined |
| Deals | Sales Hub only | Combined only |
So a sales team on Sales Hub Professional can score companies and deals, but not contacts. A marketing team on Marketing Hub Professional can score contacts and companies, but not deals. That catches a lot of people out.
Fit, engagement and combined scores
HubSpot splits scoring into two questions, and it's worth keeping them apart in your head.
Fit scores a record on its properties: job title, industry, company size, country. It answers "is this the kind of company or person we sell to?"
Engagement scores a record on what it has done: page views, email clicks, form fills, meetings. It answers "is this record active right now?"
Combined does both. A combined score creates three properties: the total, the engagement part and the fit part. On deals, combined is the only option.
When you set thresholds, fit and engagement scores get High, Medium and Low labels. Combined scores get a grid from A1 to C3. The letter is fit (A is high fit, C is low) and the number is engagement (1 is high, 3 is low). So an A3 is a great-looking company that has gone quiet, and a C1 is a busy record that looks nothing like your customers.
That grid is the most useful thing in the tool. Most teams look at the total and ignore it.
How the points work
Every score has at least one score group. Groups hold your criteria and add up to the total.
- Criteria can add or subtract points, and points can be decimals. Subtract more than you add and the score goes negative.
- You can cap each group, and cap the overall score. Once a record hits the cap, nothing more is added or removed.
- Several criteria on the same property or event are joined with AND. For OR logic, HubSpot suggests building a segment or a workflow and scoring membership of that instead.
- Event criteria can be filtered by how often something happened or when (in the last 30 days, before a date, and so on).
- You can build more than one score per object, say one per region or per product.
- When you edit a score, HubSpot re-evaluates every record against the new criteria, history included. Check what workflows and reports use the score property before you change it.
Score decay
Engagement and combined scores can decay, so old activity fades. You pick an interval of 1, 3, 6 or 12 months, where a month is 30 days. Decay is linear, applies to each event separately and applies to historical data too. It doesn't work on segment membership criteria.
If you score engagement without decay, someone who clicked everything in 2024 can still look red hot today.
Does HubSpot lead scoring use AI?
Yes, in one place. On Marketing Hub Enterprise, HubSpot can build a contact fit or engagement score for you.
You pick a start and end lifecycle stage and a timeframe. HubSpot's own example is Marketing Qualified Lead to Sales Qualified Lead within 30 days. It needs at least 50 contacts, with 25 that made that move and 25 that didn't. It then recommends criteria, and you review, edit and switch the score on.
Two things to notice. It's contacts only, not companies or deals. And it learns from lifecycle stage changes, not from which deals you actually won.
What happened to the HubSpot score property?
The old "HubSpot score" property stopped working on 31 August 2025. Any workflow, list or alert still pointing at it stopped updating that day. If your lead routing quietly broke last year, that's the first place to look. Rebuild it in the lead scoring tool and repoint the workflows.
What HubSpot lead scoring can't see
The tool is solid. It runs your rules reliably and cheaply. The problem is upstream of it.
The fit criteria are a guess until you check them. Somebody decides that SaaS companies with 50 to 200 staff in the UK are worth 20 points. Maybe that's right. But nothing in HubSpot compares that rule to the companies you've actually closed. On companies and deals, which is where most sales teams score fit, there's no AI option at all. The rules stay exactly as good as the person who wrote them.
Engagement isn't fit. A company can open every email and still be a terrible customer. Activity tells you someone is interested. It doesn't tell you they look like the people who buy, stay and pay. Mixing the two into one total hides exactly the records you need to see: the high-fit company that's gone quiet, and the busy one you'll never close.
Rules don't move unless you move them. Your market shifts. The customers you won this year may not look like the ones you won two years ago. A manual score only changes when someone sits down and edits it, and that rarely happens.
One property at a time. Criteria score properties individually and join with AND. What actually separates good customers is often a combination, like a certain size in a certain industry, and that's awkward to express in points.
None of this is HubSpot doing anything wrong. It's a scoring engine, not a way of working out who your best customers are. That part is your job.
How to check your fit criteria against your wins
You can do a rough version of this today, in HubSpot or a spreadsheet.
- Export your closed-won deals with their company properties: industry, employee count, country, how they found you. Aim for at least 20. Below 10 there's no real pattern to find.
- For each fit criterion in your score, count how many of those wins actually have it. If you give 20 points for "SaaS" and only 4 of your 30 wins are SaaS, that rule is costing you.
- Do the same for your closed-lost deals. A trait that's common in both is telling you nothing.
- Reweight. Put the most points on the traits your wins share and your losses don't.
- Put a date in the diary to do it again next quarter.
If you'd like a worksheet for step 2, the ideal customer profile template has an evidence column for exactly this. There's a longer method in how to analyse your closed-won deals.
I once opened my weekly lead allocation and found a one-person mobile pizza van in it, at an agency selling B2B services worth years of that van's revenue. It had passed every check, because the checks measured volume, not fit. The pizza van was easy to spot. The leads that really hurt are the ones that score well and look right but don't match who actually buys. That story is here.
What if you're on HubSpot Starter?
Then you can't use HubSpot lead scoring at all. It starts at Professional, and moving up isn't a small step.
| HubSpot plan | Published price (third-party guides, 2026) | Lead scoring? |
|---|---|---|
| Sales Hub Starter | About $15 a seat a month, billed annually | No |
| Sales Hub Professional | About $100 a seat a month, billed annually, plus a $1,500 one-off onboarding fee | Yes, rules you write |
| Marketing Hub Professional | From about $890 a month for 3 seats, plus a $3,000 one-off onboarding fee | Yes, rules you write |
| Marketing Hub Enterprise | From about $3,600 a month | Yes, plus AI contact scores |
Check HubSpot's pricing page for today's numbers in your currency. They change, and they vary by region and deal.
So a five-person sales team going from Starter to Sales Hub Professional pays roughly $425 a month more, plus $1,500 up front. What it gets for that is a points system it still has to design, check and maintain itself.
Plenty of small teams quite reasonably decide that isn't worth it. They're the teams Telepath was built for first.
HubSpot lead scoring vs Telepath
They aren't the same kind of tool, so this isn't a like-for-like fight. HubSpot runs the rules you give it. Telepath works out what the rules should be from what you've actually won. Here's where they differ.
| HubSpot lead scoring | Telepath | |
|---|---|---|
| HubSpot plan needed | Professional or Enterprise | Starter, Professional or Enterprise |
| Price | Per seat, plus onboarding fee | One price for the whole team (see pricing), no per-seat pricing, no onboarding fee, and a £1 first month for new subscribers (shown in your currency on the pricing page) |
| Where the fit criteria come from | You write them | Worked out from your closed-won deals |
| Learning from outcomes | AI scores on Marketing Hub Enterprise only, contacts only, trained on a lifecycle stage change | Every customer, built from the deals you actually won |
| Minimum data | AI scores need 50 contacts (25 that converted, 25 that didn't) | 10 closed-won deals; 30 or more gives a sharper picture |
| Companies | Fit score from your rules | Companies with enough data ranked by how much they look like your wins, with the reason shown |
| Open deals | Combined score from your rules | A T-Score on deals with enough data, showing which traits matched and which were missing |
| Thin data | Scores whatever it has | Leaves the score blank and says what's missing |
| Writes to HubSpot | Its own score properties | One deal property (plus fit properties on companies if you turn that on); never changes your existing records |
| Try before you pay | No | Free ICP report from HubSpot or a CSV, no card |
The honest version: if you're already on Professional and your rules are good, HubSpot will run them well and you may not need anything else. If you're on Starter, or you're on Professional and suspect your rules are a guess, that's where Telepath earns its place.
And they work together. Telepath writes its score into HubSpot as a normal property, so you can use it in views, lists and workflows next to an engagement score, or use what it finds to rewrite your HubSpot fit criteria.
You can see what your won deals have in common before deciding anything. Run a free ideal customer profile report from HubSpot or a CSV. No card, no sales call.
HubSpot lead scoring FAQs
Is HubSpot lead scoring free?
No. It needs Marketing Hub or Sales Hub at Professional or Enterprise. Free and Starter don't include it.
Can I score leads on HubSpot Starter?
Not with HubSpot's own tool, which starts at Professional. You can keep a manual fit property and update it with Starter's simpler automation, or use an app that writes a score into HubSpot. Telepath works on Starter and writes its score to your deals as a custom property.
Can HubSpot score deals?
Yes, on Sales Hub Professional or Enterprise. Deals get combined scores only, with no separate fit or engagement score and no AI option. There's more detail in HubSpot deal scoring.
What do A1 to C3 mean in HubSpot lead scoring?
They're the threshold labels for combined scores. The letter is fit (A high, C low) and the number is engagement (1 high, 3 low). A1 is a great fit that's active. C3 is a poor fit that's quiet.
How many contacts do I need for an AI lead score?
At least 50, with 25 that made the lifecycle move you chose and 25 that didn't. You also need Marketing Hub Enterprise.
What's the difference between lead scoring and fit scoring?
In HubSpot's terms, fit scoring is one part of lead scoring: points for who a record is, as opposed to engagement points for what it does. The harder question is whether your fit points match the customers you actually win. That's what the checklist above is for.
Sources: HubSpot Knowledge Base, "Overview of the lead scoring tool" (updated 23 October 2026); Weidert Group on the retirement of the legacy HubSpot score property; HubSpot prices as published by Docket (Sales Hub, verified April 2026) and TinyCommand (Marketing Hub, citing HubSpot's pricing page, June 2026). Checked 3 November 2026.