HubSpot lead scoring limits, and what to do about each one

HubSpot rebuilt lead scoring in 2025. The old HubSpot Score property was retired in stages: no new legacy scores from May 1, no edits from July 1, and from August 31 the old scores stopped updating and froze at their last value. In its place is a proper scoring tool with separate fit, engagement, and combined scores, decay, and reporting.

The new tool is a real improvement, and for many teams it is enough. It also has limits, and they show up quickly once you score more than one motion, sell to buying groups, or move fast. This guide covers each limit, who it affects, the workaround inside HubSpot, and where that workaround stops working.

Everything below comes from HubSpot’s own documentation on building lead scores, managing them, and how the scoring tool works. HubSpot changes this product often, so check your own portal before you act on any of it.

First, check that your old score is not still in use

If you scored leads before the change, start here. Any workflow, list, report, or alert that reads the legacy HubSpot Score property is now reading a number that no longer moves. Nothing warns you. New leads get no score, qualification workflows stop triggering, and a list of top-scoring leads quietly stops changing.

Search your workflows, lists, and reports for the old property, and repoint each one to a score from the new tool. It is the fastest fix in this guide, and on many portals it is the most important.

What the native tool does well

It is worth being clear about this before listing the limits.

• Fit, engagement, and combined scores for contacts and companies, and combined scores for deals.
• A nine-band grid for combined scores, A1 to C3, with high, medium, and low bands for fit and engagement on their own.
• Decay on engagement, so old activity counts for less.
• Positive and negative points, and limits on how many points each group can add.
• A score history chart on each record that shows the events that changed the score.
• Test tools that show how points are added or subtracted rule by rule for a single record, and a preview of how scores spread across your records.
• Reports on score distribution and trends over time.

If you sell one product, through one motion, to buyers who mostly act alone, this may be all you need. The limits below matter most when that stops being true.

The limits, and what to do about each

1. It needs a Professional or Enterprise plan

The scoring tool is available on Marketing Hub and Sales Hub Professional and Enterprise. Free and Starter portals do not get it.

• Who it hurts: smaller teams, and teams that bought HubSpot for its CRM and have not upgraded.
• Workaround: a custom number property you calculate outside HubSpot, in a spreadsheet or another tool, and import back on a schedule.
• Where it breaks: the score is only as fresh as your last import, and nobody can see why a record scored what it did.

2. Which records you can score depends on which hub you own

Contact scores need Marketing Hub. Deal scores need Sales Hub. Company scores work with either.

• Who it hurts: sales-led teams on Sales Hub alone, who cannot score contacts natively, and marketing-led teams who cannot score deals.
• Workaround: score at the company level, which both hubs support, and use contact properties such as role and seniority in lists and routing rather than in a score.
• Where it breaks: you lose the ability to tell reps which person to call, which is often the question they care about most.

3. The number of scores is capped by plan

HubSpot says the types and number of scores you can create depend on your subscription. Its knowledge base does not list the counts, and third-party guides report different numbers, so check the limit in your own portal.

• Who it hurts: teams with several segments, products, or motions that each deserve their own model, such as a mid-market fit score and an enterprise fit score, or separate scores for new business and expansion.
• Workaround: decide which scores earn a slot. One fit and one engagement score per major segment usually matters more than a score for every product.
• Where it breaks: once you run separate motions, such as product-led and sales-led, one model has to serve both, and it will be wrong for one of them.

4. Decay is stepped and monthly

Decay reduces a score by a percentage every 1, 3, 6, or 12 months, where HubSpot counts a month as 30 days. It is set on the event criteria in an engagement score.

• Who it hurts: fast motions. In inbound and product-led sales, a pricing visit from three weeks ago is already stale, and monthly steps are too coarse to reflect that. Stepped decay also means a score can drop sharply on one day rather than fading.
• Workaround: HubSpot lets you use a workflow to reset an engagement score to zero on certain triggers, such as a set period with no activity. Pair that with the shortest decay interval. Points from segment membership criteria are not included in a reset, and the reset only runs while the score is turned on.
• Where it breaks: a reset is all or nothing. You cannot give different signals different half-lives, so a demo request and a blog visit fade at the same rate.

5. Every score has three bands

Fit and engagement scores are labeled high, medium, or low. Combined scores use the nine-band A1 to C3 grid.

• Who it hurts: teams that want more than three tiers, such as a top-priority band above A, or that route by finer score ranges.
• Workaround: the raw number is still stored on the record, so lists and workflows can use score ranges directly instead of the band labels.
• Where it breaks: your reports and your routing now use different definitions of “high,” and someone has to keep them in step.

6. Account roll-ups are basic

When a score draws on associated records, such as contacts at a company, HubSpot can combine their values by sum, average, minimum, or maximum.

• Who it hurts: anyone selling to buying groups. Sum rewards one very active contact as much as five moderately active ones. None of the four options can tell you whether you have reached the economic buyer, the champion, and the technical reviewer, or only one of them.
• Workaround: a workflow that sets a “roles engaged” property on the company when contacts in key roles take action, then scoring that property as a fit or engagement factor.
• Where it breaks: it is a lot of workflow logic to maintain, it does not decay, and it breaks when job titles are inconsistent, which they usually are.

7. AI scores need Enterprise and enough history

AI-built scores need Marketing Hub Enterprise, and a minimum sample of 50 contacts, with at least 25 that converted and 25 that did not.

• Who it hurts: everyone below Enterprise, and teams with low deal volume, where 25 conversions of the right kind can take a long time to collect. Fifty records is also a minimum for the model to run, not a guarantee it is accurate.
• Workaround: start with a rules-based score you set by hand, which you can explain and test, and revisit AI scoring once you have a larger history of wins and losses.
• Where it breaks: if reps cannot see why an AI score is what it is, they will treat it like any other black box. Check how much of the reasoning your team can actually read before you route on it.

8. It can only score what is already in HubSpot

Every rule reads a HubSpot property or a tracked event. Signals that live elsewhere, such as product usage, public records, third-party intent, or data in a warehouse, have to be brought into HubSpot as properties or events first.

• Who it hurts: product-led companies, teams selling into regulated or public-data-heavy industries, and anyone whose best signals live outside the CRM.
• Workaround: sync the signals you need into custom properties or events, through an integration or a data tool.
• Where it breaks: every new signal is a new sync to build and maintain, and stale or missing syncs silently lower scores.

Test before you trust

Whatever you build, use HubSpot’s test tools before you switch the score on. Check a few records you know well, rule by rule, and look at the distribution preview. In portals with many records, the preview is based on a sample, around 20% of eligible contacts, so treat it as a guide rather than a count.

Then check the ranking the way a rep would. If the top 20 records are not the ones your best rep would call first, the weights need work before anyone sees the score.

When to look past native scoring

Native scoring is the right place to start. It is worth looking further when:

• you run more than one motion or segment and the score limit forces them to share a model
• your deals involve buying groups and a sum or average of contact scores does not reflect who you have reached
• you need scores to stay current as your records change through the day
• your strongest signals live outside HubSpot
• your reps do not trust the score because they cannot see why a record ranks where it does

That gap is what BoostIdeal was built for. BoostIdeal reads what is in HubSpot or a CSV you upload, so outside signals have to be brought in first. It scores leads, contacts, and accounts, can score one record using the data on another, such as a contact by the account behind it, and runs several models side by side so you can compare them. Scores recalculate hourly by default, and in near real time with webhook sync. Every score and the reasons behind it write back into HubSpot, so a rep can open any record and read why it ranks where it does.