# A Simple Customer Health Score for Small SaaS Teams

> How small SaaS teams can build a simple customer health score using usage, adoption, support and relationship signals, without complex tools.

Source: https://comppound.com/blog/a-simple-customer-health-score-for-small-saas-teams (published 2026-07-19)

A customer health score is an estimate of how likely an account is to stay, expand or churn. Large companies build elaborate models with dozens of weighted inputs and a team to maintain them. A small SaaS team does not need any of that to get value. A simple customer health score built from a handful of signals you already collect is enough to show which accounts deserve attention this week.

It helps to be clear about the goal from the start. You are not trying to predict churn with precision. You are trying to notice trouble early enough to do something about it, and to notice healthy accounts early enough to ask them for more. A score that does those two things imperfectly is far more useful than a sophisticated model nobody trusts or updates.

## Why bother with a health score

When you have twenty customers, you can keep the state of each account in your head. Somewhere between fifty and a few hundred, that stops working. The founder who used to handle every relationship is now also hiring, fundraising and shipping, and the customers who complain loudly get attention while the ones quietly drifting away get none. A health score is a way to make the quiet ones visible.

It also lets you prioritize outreach when you cannot talk to everyone. If one person handles customer success for two hundred accounts, the score decides who gets a call on Monday. It helps you spot expansion opportunities, since accounts that are thriving are the ones most likely to add seats or upgrade. And tracked over time, it tells you whether the customer base as a whole is getting healthier or slowly eroding, which is often hard to see from revenue alone until it is too late.

## Choosing a few signals

Start with three to five signals you can measure reliably and update without manual heroics. Reliability matters more than cleverness. A signal that is accurate every week beats a richer signal that depends on someone remembering to fill in a field. The categories below are the usual places to look, and most teams end up choosing one or two from each of the first few.

### Usage

Usage is the strongest and most objective signal for most products. The simplest version is active users in the last 30 days compared with seats purchased. Login frequency is also useful, but compare it to the account's own normal pattern rather than to an absolute number, because a team that logs in weekly by design is not at risk just because another team logs in daily. Usage of core features matters more than raw logins. Someone who opens the app to look at a dashboard is less engaged than someone who builds reports in it, so define the one or two actions that represent real value in your product and count those.

### Adoption

Adoption measures how deeply the product is woven into the customer's work. Useful signals include how many key features the account uses, how many teammates have been invited, and how many integrations are connected. Each of these raises the cost of leaving. An account with one user and no integrations can switch tools in an afternoon. An account with fifteen users and three integrations would have to rebuild workflows to leave.

### Support

Support data tells you about friction. The number of tickets is ambiguous on its own, since an engaged account may file many tickets because it uses the product heavily. Sentiment and resolution are more telling. A string of frustrated tickets, or an issue that has stayed open for weeks, is a warning sign regardless of volume. Some teams also treat a sudden stop in tickets from a previously active account as a signal worth checking.

### Relationship

Relationship signals capture what the data cannot. Recent positive or negative feedback from a call, a reply to a survey, or a comment in a review all belong here. One of the most important signals in B2B is whether your champion, the person who bought or advocated for the product, is still at the company and still in the role. When the champion leaves, the account often loses its reason to stay. Engagement with your emails and release notes is a weaker signal, but a total lack of it can be informative.

### Billing

Billing signals are blunt but important. Failed payments, recent downgrades and an approaching renewal date all change how urgently an account needs attention. A renewal in six weeks turns a "monitor" account into a priority.

## Building the score (illustrative)

Once you have chosen signals, assign points to each level of each signal and add them up. Here is one way a team might do it, described in words rather than a grid so you can see the reasoning.

For active users compared with seats, the team gives 3 points when 70 percent or more of seats are active, 2 points for between 30 and 70 percent, and nothing below 30 percent. For core feature usage, weekly use earns 3 points, monthly use earns 2, and rare use earns nothing. Connected integrations earn 2 points for two or more, 1 point for one, and nothing for none. Support sentiment earns 2 for positive, 1 for neutral and 0 for negative. Finally, an active champion earns 2 points, an unknown champion status earns 1, and a champion who has left earns nothing.

That makes 12 points possible. The team calls 9 to 12 healthy, 5 to 8 worth monitoring, and 0 to 4 at risk. These thresholds are arbitrary at first. The point is to have a starting system you can test, and the weights give extra influence to usage, which for most products is the best early warning. Your product may differ. If your tool is used intensively once a quarter, weekly usage is the wrong yardstick and you should define frequency around the natural rhythm of the work.

Resist the urge to add more signals in the first version. Every signal you add makes the score harder to explain and harder to debug. If someone on the team asks why an account is at risk, you want to be able to answer in one sentence.

## Validating the score against history

A score you have not tested is a guess dressed up as a number. The cheapest validation is to apply the score retroactively to customers whose outcome you already know. Pull the accounts that churned in the last six or twelve months and calculate what their score would have been a month or two before they left. Do the same for accounts that expanded.

If churned accounts mostly scored low and expanded accounts mostly scored high, the score is doing its job. If churned accounts were scattered across all levels, something important is missing or mis-weighted. Look at the churned accounts that scored well and ask what they had in common. Often it is a signal you left out, such as a champion leaving or a key integration breaking. With a small customer base you will not have enough data for anything statistical, so treat this as a sanity check, not proof, and revisit it as more outcomes accumulate.

## Making the score actionable

A score that does not change anyone's behavior is just a column in a spreadsheet. Decide in advance what happens at each level, so the score triggers work rather than discussion.

For healthy accounts, the natural actions are to ask for a review, a referral or a case study, and to look for expansion opportunities such as additional teams or a higher plan. These customers are already getting value, so the conversation is about doing more together.

For accounts in the monitor band, the goal is to nudge them toward deeper adoption. Share tips about features they have not used, especially ones that other successful accounts rely on, and check in on the goals they had when they bought. A short email asking whether they are getting what they expected often surfaces issues before they turn into churn.

For at-risk accounts, the response should be personal. Someone should reach out directly, ideally the person who knows the account best, to understand what is going on. Offering a fresh onboarding session or training for new team members can help, as can investigating specific problems in the support history. Sometimes the conversation reveals that the account was a poor fit from the start, which is useful to learn too, because it tells sales and marketing something about who to target.

## Tracking it without special software

You do not need a dedicated customer success platform to start. A spreadsheet updated weekly from product analytics exports works for many early teams. If your CRM supports calculated fields, you can store each signal on the account record and compute the score there, which keeps it next to the notes and contacts the team already uses.

The important part is that updating the score is routine and cheap. If it takes someone half a day of manual work, it will slip the first busy week. Automate what you can, even if that just means a saved report you export on Mondays. As the customer base grows and the manual process becomes painful, that is the signal that a customer success tool is worth the cost.

## Reviewing on a rhythm

Set a weekly review that takes no more than half an hour. Look at accounts whose score dropped since last week, accounts currently at risk, and accounts with renewals coming up in the next couple of months. Assign an owner and a next step for each one that needs action.

Once a month, look at the overall picture. Is the share of healthy accounts rising or falling? Are certain customer segments consistently scoring lower? Those trends feed product and positioning decisions, not just customer success work.

## Watch changes, not just levels

A sudden drop often matters more than a low but stable score. An account that has always been light on usage may be perfectly content using the product for one narrow job. An account whose usage halved in two weeks is telling you something changed, perhaps a reorganization, a new manager, or a competitor trial. Track the week-over-week change alongside the current score and flag large drops even when the account is still nominally healthy.

## Combine the score with human judgment

Scores miss context, and the people closest to the account usually know things the data does not. Low usage might be seasonal, as with a tool used heavily during annual planning and lightly the rest of the year. High usage might mask a customer who is actively evaluating replacements and using your product hard while they migrate. Treat the score as a prompt for a conversation, and let the account owner override it with a note explaining why. Those overrides are also valuable feedback for improving the model.

## Common mistakes

The most common mistake is starting with too many signals. A score built from fifteen inputs is impossible to reason about, and when it is wrong nobody can tell why. Close behind is using weights that were never validated, so the score reflects someone's intuition rather than what actually happened to past customers.

Many teams also build a score and then attach no actions to it, which means it becomes a dashboard nobody looks at. Others rely entirely on the numbers and ignore what the account manager heard on last week's call. And a surprising number of teams never revisit the model, even as the product changes and new features become the real markers of value.

## Start simple and refine

Begin with three to five signals, a basic points system and clear actions for each level. Validate it against the customers you have already won and lost, review it weekly, and change one thing at a time as you learn which signals actually predict churn and expansion in your product. A simple health score will not catch everything, but it will make sure a small team spends its limited attention on the accounts where attention changes the outcome.

## Keep reading

- [A SaaS Onboarding Call Script for Small Teams](/blog/a-saas-onboarding-call-script-for-small-teams)
- [Choosing between in-house SEO vs agency SaaS for small teams](/blog/in-house-seo-vs-agency-saas-choosing-for-a-tenperson-team)
- [Brief Landing Page AI Citations for Small Teams](/blog/brief-landing-page-ai-citations-a-system-for-small-teams)

## Frequently asked questions

### What is a customer health score?

A score estimating customer likelihood to stay, churn or expand based on signals like usage, adoption and support.

### How many signals should a health score use?

Small teams can start with three to five reliable signals.

### Do I need customer success software for health scores?

Not initially. A spreadsheet or CRM can work for small teams.

## About Comppound

Comppound runs SEO and AI search for B2B SaaS companies: a person sets the plan and reviews samples and every high-risk page, and AI speeds up production.

- [See SEO Growth](https://comppound.com/services/seo-growth): Buyers compare before they book a demo. Are your pages there when they do?
- [Book a call](https://comppound.com/contact): Most of this works better when buyers can find you first