---
title: Health Scores
slug: health-scores
description: Create health scores from any numeric data, set thresholds, build clustered scores, and learn how the Global Health Score is calculated and refreshed.
createdAt: 2024-08-02T08:39:27.601Z
---

*Avoid customer churn by identifying risks and opportunities early, with scores that reflect what healthy means for your business.*

# Overview

Health Scores turn your data into a clear signal: **Good**, **Average** or **Bad**. Each account can have several health scores covering product usage, engagement, satisfaction, support or revenue, plus a **Global Health Score** that combines them. You decide what each score measures and where the thresholds are.

![](https://api.archbee.com/api/optimize/16HhozMLfMR_uwZBLzFrU/cLV_xRRd0-qoAm-9G4qnq_image.png)

While Segments group your customers, Health Scores help you:

- See the overall health of every customer and of your whole portfolio at a glance.
- Score any numeric data: events, attributes, metrics, and calculated metrics.
- Follow how health evolves and compare it with previous periods.
- Trigger automations, prioritize your day, and report on risk.

# Types of health scores

![](https://api.archbee.com/api/optimize/16HhozMLfMR_uwZBLzFrU/8rakkGWd0xU2XZPlKGQZp_image.png)

| **Type**                   | **Use it when**                                                                                                                                                                                                                                                                                                                   |
| -------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |
| **Standard health score**  | You want to score a single data point: an event frequency, an attribute or a metric (for example logins in the last 30 days).                                                                                                                                                                                                     |
| **Clustered health score** | You want to combine several health scores into one, for example an “Adoption” score built from feature usage scores. Clustered scores can also include other clustered scores, so you can build multi level frameworks such as “Customer Health” = “Engagement Health” + “Adoption Health”. Read more in Clustered Health Scores. |
| **Global Health Score**    | Created by default. It is the weighted average of your other health scores and can be enabled or disabled.                                                                                                                                                                                                                        |

# Creating a health score

Open *Health Scores* from the main menu and click *New Health Score*, use the **+** shortcut on any page, or **duplicate** an existing score and adjust it.&#x20;

![](https://api.archbee.com/api/optimize/16HhozMLfMR_uwZBLzFrU/RVuJ3a9MuVEtkXYxdsiNx_image.png)

Then define:

| **Setting**                               | **What it does**                                                                                            |
| ----------------------------------------- | ----------------------------------------------------------------------------------------------------------- |
| **Name and description**                  | Clear names help your team understand the logic behind the score.                                           |
| **Event, attribute or metric to measure** | The data the score is based on.                                                                             |
| **Time frame to measure**                 | Only for events: the period over which events are counted (for example logins in the last 30 days).         |
| **Restrict to segments (optional)**       | Calculate and show the score only for companies or people in specific segments.                             |
| **Impact on Global Health Score**         | None, Low, Medium or High.                                                                                  |
| **Score distribution**                    | The worst and best values, and the thresholds for Bad, Average and Good.                                    |
| **Distribution preview**                  | How your companies would be distributed before you save.                                                    |
| **Show in Customer 360 / Show in Portal** | Whether the score appears on the Customer 360° Profile and whether it can be shared in the Customer Portal. |

![](https://api.archbee.com/api/optimize/16HhozMLfMR_uwZBLzFrU/VVwTs-Lb4s_AHcnjGKx66_image.png)

## Defining the score distribution

Set the **worst value** (the result that clearly signals a problem) and the **best value** (the result of a healthy customer), then move the sliders to define the Bad, Average, and Good intervals. If a higher value means worse health (for example, days since last login, or number of open tickets), set the score up in reverse; CustifyAI summaries interpret reversed scores correctly.

:::hint{type="warning"}
**Observation:&#x20;**&#x41; threshold value belongs to the interval on its right. With thresholds at 3 and 7, a score from 0 to less than 3 is red, from 3 to less than 7 is yellow, and 7 or more is green.
:::

:::hint{type="info"}
**Custify Tip:&#x20;**&#x45;xample: a **Login Frequency** score counts login events over the last 30 days, with 0 logins as the worst value and 10 as the best. Between 3 and 6 logins the score is Average. It is restricted to the **Paying Accounts** segment because the signal only matters for active paying customers.
:::

:::hint{type="warning"}
**Observation:&#x20;**&#x48;ealth scores work with numbers only. To score a text or dropdown value (for example, a renewal likelihood of “High” or “Low”), first convert it into a number: use a playbook with conditions that sets a numeric attribute for each value, then build the health score on that attribute.
:::

# Clustered health scores

A clustered health score combines several health scores into one, for example, a **Product Usage** score built from four feature usage scores. Go to *Health Scores > New Health Score > Select Health Score Type* and pick one of three methods:

| **Method**        | **How it calculates**                                                                                                                                                              | **Example**                                                                                    |
| ----------------- | ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | ---------------------------------------------------------------------------------------------- |
| **Impact based**  | A weighted average using Low, Medium and High impact, like the Global Health Score. Each score starts with its Global Health Score impact, and you can change it for this cluster. | Product Usage = four usage scores, each with its own impact                                    |
| **Custom impact** | You set an exact percentage weight for each score.                                                                                                                                 | Courses rated 20%, quizzes completed 30%, courses completed 50%                                |
| **Point system**  | You assign points to the Bad, Average and Good result of each score. Points can be negative, so one bad signal can pull the whole score down.                                      | Logins per user: Bad takes away 20 points, Good adds none, because healthy logins are expected |

- A cluster needs at least two health scores.
- For Impact-based and Custom impact clusters, the scale is fixed from 0 (worst) to 100 (best). For the Point system, the worst and best values are the sums of the lowest and highest points you assigned.
- Turn on **Ignore nulls** in an Impact-based cluster so missing values are left out of the calculation instead of pulling the score down.
- Clusters can include other clusters, for multi-level frameworks.
- New clusters have no impact on the Global Health Score by default. Set an impact if you want them to count.
- You can change the method of an existing cluster by editing it.

# How the Global Health Score is calculated

The Global Health Score is the weighted average of your health scores, using the impact you set on each one: Low counts once, Medium twice and High three times. For example, with score A (low), B (medium) and C (high):

:::hint{type="info"}
**GHS = (A + B × 2 + C × 3) / 6**
:::

:::hint{type="danger"}
**Important:&#x20;**&#x41; health score with impact **None** is multiplied by zero and does not count in the Global Health Score. Scores restricted to a segment only count for companies in that segment.
:::

![](https://api.archbee.com/api/optimize/16HhozMLfMR_uwZBLzFrU/uD0HNALXlOCRw3yWovx_Z_image.png)

# Calculation Frequency

Health scores are recalculated once a day, around midnight in your account’s time zone, no matter whether they are based on events, attributes or metrics. Each score on the Customer 360° Profile shows when it was **last calculated**, so you always know how fresh the data is.

:::hint{type="warning"}
**Observation:&#x20;**&#x41; new health score, or one you just edited, needs the next daily calculation before values appear or change. If a calculated metric used in a score updates more often (for example, hourly), the score reflects its value at the time of the daily calculation.
:::

# Following health across your portfolio

- **Health Scores list:** see how many customers have Good, Average and Bad scores. The numbers are clickable and open the matching customers.
- **Health score dashboard:** click a score name to see the split, statistics and a histogram of current values.
- **Customer List View:** add health scores as columns, sort and filter by them. In Kanban view, each card shows the health score you choose in *Settings > Company Board Card Design*.
- **Customer 360° Profile:** the sidebar shows the scores for the account, each with its evolution chart.
- **Insights:** build widgets such as MRR by Global Health Score, or healthy versus unhealthy customers per CSM.
- **Customer Portal:** share selected scores with your customer, with a customer friendly display name. Show the current value with its evolution, or the history only, without revealing today’s value.
- **Standard reports:** upcoming renewals and the Cashflow Forecast can be split by Global Health Score, so you see how much revenue is at risk.
- **CRM sync:** export health scores, including the Global Health Score, to HubSpot or Salesforce. In Salesforce, turn on *Export Health scores from Custify as Account properties in Salesforce* in the integration settings; Custify creates the fields for you.

![](https://api.archbee.com/api/optimize/16HhozMLfMR_uwZBLzFrU/n2EnPeDOtcnY49G6Nc4SH_image.png)

# Useful health score recipes

## Usage trends

When customers differ a lot in size or plan, comparing them with each other is not fair. Compare each customer with their own past instead:

1. Build a calculated metric that compares two periods, for example logins in the last 7 days against logins in the 7 days before. The result is the change in percent: (this period divided by the previous period, times 100) minus 100.
2. Build a health score on that metric. For example, a drop larger than 50% is Bad, a drop between 50% and 20% is Average, and anything better is Good.

Read more in [Usage Trends](https://kb.custify.com/usage-trends).

## Alerts when health drops

Create a calculated metric that compares the current Global Health Score with its value a few days ago. Then start a playbook when that metric falls below your threshold, for example, a 15-point drop, to create a task for the CSM or send a Slack alert. Playbooks can also start directly when a health score becomes Bad.

## Rolling up health from child to parent entities

Health scores of child entities do not affect the parent automatically, so each level can have its own scores and thresholds. To roll health up, build a calculated metric on the parent with a child entity function (sum, average, minimum, maximum, or count), optionally filtered by segment or tag, then create a health score on that metric restricted to parent companies. Read more in Rolling up health scores from children to parents.

# Managing health scores

- **Reorder** scores by drag and drop; the order is kept on the Customer 360° Profile.
- **Edit**, **duplicate** or **delete** scores. Deleted scores are removed from every profile.
- **Tag** scores to organize them (create health score tags first in *Settings > Tags*).

# Going beyond classic health scores

- **AI Sentiment and Risk scores:** CustifyAI Conversation Analysis scores the tone and risk of customer emails and selected notes. The scores are company attributes, so they can feed health scores, segments and playbooks. Read more in CustifyAI.
- **Calculated Metrics:** score ratios and formulas such as active users out of total users, days since last message, or the average completion of Customer Objectives.
- **Single-threading risk:** the Stakeholders Map flags accounts where most communication goes through one contact, and stores it as an attribute you can score.
- **Signals:** count open bad signals per account and include them in your Global Health Score.

# Related articles

- [Clustered Health Scores](https://kb.custify.com/clustered-health-scores)
- [Health scores are green, but the Global Health Score is red. Why?](https://kb.custify.com/health-scores-are-green-but-the-global-health-score-is-red-why)
- [Rolling up health scores from children to parents](https://kb.custify.com/rolling-up-health-scores-from-children-to-parents)
- [Calculated Metrics](https://kb.custify.com/calculated-metrics)
- [Health Scores: Best Practices](https://kb.custify.com/health-scores-best-practices)

