How do you build effective customer health scores in HubSpot?

Service Hub Tips, Tricks & Best Practices
1

Hi everyone!

We’ll soon be starting with HubSpot Service Hub, and one of the areas we’re currently exploring is Customer Health Scores. I’d love to hear how others have approached this, especially if you have a slightly complex product/licensing setup.

Our situation looks something like this:

  • We offer multiple SaaS products, with different versions/platforms available.
  • A single customer can have multiple licenses, including multiple licenses for the same product or licenses across different versions.
  • We sell through different marketplaces, which means licensing models can vary quite a bit.
  • For example, on the Atlassian Marketplace, customers generally need to purchase a license covering all users in their Atlassian instance, even if only a small percentage of those users actually use our app. As a result, license size isn’t a reliable indicator of adoption or actual usage.
  • As an EU company, we also have privacy considerations around tracking individual users without the appropriate consent. Our usage data is therefore primarily aggregated at license/account level rather than individual user level.

This makes some of the more straightforward health-score signals, such as number of seats vs. active users, difficult or potentially misleading for us.

We’re now trying to determine what a meaningful health score could look like in HubSpot. We’re considering signals beyond pure usage, such as engagement with our team, support activity, product adoption at an aggregated level, renewals, training/onboarding participation, feedback, etc., but we’re still very much in the design phase.

Has anyone here dealt with a similar setup?

I’d especially love to hear how you approached health scoring when license size doesn’t reflect actual usage, or when you have multiple products/licenses associated with the same customer.

Thanks!
Margaux

3

Hi @Margauxx and welcome! :orange_heart: It’s a pleasure to have you on the HubSpot Community!

Thanks for sharing this use case, that’s exactly why the Community is here for!

  • How are your products and licenses represented in HubSpot currently?
  • Also, is the score primarily intended to predict renewals or help your team prioritize outreach?

Those details will help the Community suggest the best structure.

One approach would be to separate the customer-level health score from product/license-level health.

Here are some resources that can help:

Now, I’d love to put you in touch with our Top Experts: Hi @danmoyle, @Edouardbrunetot and @EugeneAtKiwi do you have best practices to share, please?

Thanks so much and enjoy the day! :sun:

Bérangère

4

Hi Bérangère,

Thanks for the warm welcome!

Our data is still being set up, so I can’t yet answer how licenses/products are set up.

We would like to use the health scores mostly to predict renewals and boost onboarding/adoption.

Thanks for the help!
Margaux

5

Welcome to the Community @Margauxx! +1 for what @Berangere offered. I’d just add a few ideas to think about.

Don’t let seat count drive the score when it’s structurally disconnected from usage, which sounds exactly like your Atlassian Marketplace situation. Measure against an expected usage baseline instead (activation milestone, the customer’s own historical level, or a comparable cohort) rather than active users ÷ purchased seats.

For aggregate-only data, focus on meaningful actions instead of individual tracking:

  • Activation: has the license completed core setup (first workflow run, first integration, etc.)
  • Breadth: is it using multiple core features, not just one
  • Recency and trend: when did it last do something meaningful, and is that trending up or down
  • Consistency: steady use across weeks, or isolated spikes
  • Outcome volume: completed workflows, transactions, resolved items, not people-level data

On roll-up logic for multi-license customers, don’t just average every license score. A few patterns that work better:

  • Critical-product floor: overall health can’t be “healthy” if a strategic license is at risk
  • Revenue-weighted roll-up: weight by recurring revenue or renewal value, not seats
  • Worst-signal plus context: show both the overall score and the lowest-scoring license so CSMs know where to look

A couple of guardrails to keep in mind, too: separate “new and not yet activated” from “at risk,” since those look similar in raw data but mean very different things. And don’t let missing or unavailable telemetry (a real risk with marketplace and privacy constraints) quietly default to a red score.

Finally, advice I share often with clients is to start small: 4-6 factors you can actually influence, then validate against real renewal and churn outcomes before adding complexity.

Hope that’s helpful!

6

One thing that isn’t seat-dependent and works fine with aggregate-only data: support signals at the account level. Ticket volume trend over the last few months, how often the same issue gets reopened or re-contacted within a week, and whether tone across successive conversations is getting more frustrated - none of that needs per-user tracking, just account ID on each conversation. For renewal prediction, we look at contacts clustering near contract/renewal dates, and any mention of alternatives or switching, which tends to show up in support transcripts first

this could work alongside the approach Dan suggests rather than instead of it. Usage tells you whether they’re using the product; support tells you whether they’re happy using it, and those two can diverge quite a bit, especially early in onboarding

7

Hi there @Margauxx I recently went down the hubspot health score route and did not have very good experience, i found it very confusing and restrictive to set up and hard to work out how scores were calculated to be able to trust it.
I moved to using Claude to run a fortnightly task to calculate health scores looking at about 5 connectors including useage, sentiment, calls analytics, it works well so would recommend having a go at this

8

Thanks Nathan, would you be open to share a bit more details outside of this forum?

9

Hi @NathanJones and all,

Thanks for sharing this feedback with the HubSpot Community. A good starting point is to test individual records and review the score history so you can see exactly which rules are contributing to each score. HubSpot also supports engagement, call, ticket, sentiment, NPS, timeframe, frequency, aggregation, and score-decay criteria.

If you need to combine broader signals such as CRM activity and call transcripts, the Customer Health Agent may be worth checking. An external AI workflow can be a great option too, just worth keeping data privacy in mind when working with sensitive customer information.

Thanks,
Bérangère

10

HI Margauxx, sure no problem. email is nathan@lentune.com