Customer Health Score, Computed by AI
By the end of this recipe, every customer account carries a health score and a Red / Yellow / Green status you can trust — computed from real usage and support data, updated as signals change, with a plain-language summary of who's at risk posted to your success channel. Health scoring is a read-and-narrate job: someone has to look across product usage, open tickets, survey scores, and renewal dates for every account and turn it into one honest read. Because your accounts, tickets, surveys, and notes all live in one workspace, Copera AI can read across them and do that scoring for you.
Readiness: Available now — ask Copera AI on demand to score your whole book of business, backed by automations that keep each account's signals fresh and alert a CSM the moment one slips. A single named teammate that rescores every account on its own schedule is rolling out in early access.
What it does
A customer health score answers one question your team asks constantly: which accounts are thriving, which need attention, and which are quietly heading for the exit? The signals are already there — they're just scattered across a usage board, a support board, a stack of survey responses, and last quarter's QBR notes. This recipe pulls them together, and it works in two layers:
-
An always-on signal layer (automations). Small automations keep each account's inputs current: a new or escalated ticket updates the account's support signals, a returned survey writes its CSAT onto the account, and a formula rolls those numbers into a live health score. When that score crosses your threshold, an automation flips the account to At risk and pings its CSM — so nothing drifts into danger unnoticed between reviews.
-
A portfolio-wide scored read (Ask Copera AI). On demand, Copera AI reads your accounts board and your support board — and, if you like, your QBR notes and account plans — scores every customer against a rubric you define, writes the score and a one-line reason back onto each account, and posts a summary of the Red and Yellow accounts to your success channel, @mentioning the CSM who owns each one.
Splitting it this way is what makes the score trustworthy. The formula and automations keep the numbers honest, account by account. Ask Copera AI adds the judgment a formula can't: it reads the actual tickets, weighs the qualitative notes, and turns the whole portfolio into a narrative a CS leader can read in ten seconds. Nothing leaves the workspace — every score and summary is internal, for your team, never sent to a customer.
Build it yourself
The portfolio-wide score — the read across every account, every ticket, and your account notes — is a read-and-narrate task, and the fastest way to produce it is to ask Copera AI (next section). What you build here is the layer beneath it: a board that holds the signals, a formula that scores them, and automations that keep everything current and raise a flag the moment an account slips. That way the score Copera AI writes is always grounded in fresh data instead of a stale snapshot.
One-time setup
Give each customer one row on an Accounts board with the columns a score needs to read: Active users and Usage trend (number), Open tickets and Escalations (30d) (number), Latest CSAT / NPS (number), Renewal date (date), Contract value (number), and CSM (people). Add a Health Status column (a status with options like Healthy, Watch, and At risk). New to columns? See Boards Overview.
Add a Health Score column of type Formula that combines your signals into a single 0–100 number — weighting each one however your business works (for example, reward usage and satisfaction, penalize open tickets and an approaching renewal with no activity). The formula recomputes itself whenever an input changes, so the number is always live.
On your Support / Tickets board, make sure each ticket links to its account (a link column pointing at the Accounts board). This link is what lets an automation roll a ticket's activity up onto the right customer.
Create a saved view on the Accounts board filtered to Health Status is At risk (or Watch), sorted by Contract value descending. This is your live triage queue — the automations below keep it populated, and it's the first thing to open when a score drops.
Automation 1 — keep support signals current
On the Tickets board, add an automation with a Row created trigger, so every new ticket kicks off the flow.
Add an Update linked row action pointed at the ticket's linked Account. Use it to refresh the account's support signals — for example, raise Open tickets and stamp a Last contact date. The account's health picture now reflects support load without anyone typing it in.
Repeat the pattern with a Field changed trigger watching the ticket's Priority (or Status), with a condition that it changed to Escalated. On the true path, Update linked row to raise the account's Escalations (30d). Escalations weigh heavily on health, so this keeps them front and center.
Open and escalated ticket counts are exactly the kind of support pressure that erodes health fastest. If you also track response and resolution deadlines, pair this with SLA Tracking so a repeatedly breached SLA becomes another signal your score can read.
Automation 2 — capture satisfaction scores
Publish a short CSAT or NPS form and add an automation with a Form submitted trigger. Every returned survey now flows in automatically.
Add an Update linked row (or Update row) action that writes the returned score into the account's Latest CSAT / NPS column and stamps a Survey date. The freshest sentiment is always sitting on the account, ready for the formula and for Copera AI to read.
Automation 3 — alert a CSM the moment an account slips
On the Accounts board, add an automation with a Field changed trigger watching a key input — say Open tickets or Latest CSAT / NPS — so the check runs whenever a signal moves.
Add a Formula condition that evaluates your Health Score against a threshold (for example, continue only when the score has dropped below 50). The formula condition branches the flow on the computed result, so only genuinely at-risk accounts move on.
On the true path, add a Set field action (Health Status → At risk) and a Send direct message (or Send notification) to the account's CSM. The owner hears about a slipping account the moment it happens — not only when the next scored review runs.
Everything above uses documented Automation blocks — see the full list in Automations. What automations deliberately don't do is roam across every board and every doc to synthesize one portfolio-wide read: an automation is scoped to the board and trigger it's attached to. That cross-source scoring and narration is the read-heavy job Copera AI is built for, next.
Or just ask Copera AI
This is the primary path to the actual scored read, and the one we recommend. Because it reads across your boards and documents in one pass, Ask Copera AI can do what no single automation can: score every account from every signal at once. Open Ask Copera AI and describe it in plain English:
"Read our Accounts board and our Support Tickets board. Score each customer's health from these signals: product usage and its trend, the number of open and escalated tickets in the last 30 days, the latest CSAT or NPS, and how close the renewal is. Use this rubric — Green 70–100, Yellow 40–69, Red below 40. Write each account's score and status back to its Health Score and Health Status columns with a one-line reason, then post a summary to the #customer-success channel listing every Red and Yellow account, its top reason, and @mention its CSM."
Ask Copera AI reads and searches your workspace data, so it opens both boards, reasons about each account against the rubric you gave it, and drafts the scores. Then it updates the rows — writing the score, the status, and a short reason onto each account — and posts a channel message to #customer-success. With the default Ask before acting permission mode, it shows you a review card before each write and before the post, with a Sources list so you can see exactly which rows it drew from. It only ever reads accounts and posts to channels your own account can already reach. See the full capability list in What Copera AI Can Do.
A few useful variations of the same ask:
- Bring in the qualitative signals. Add "…and read the QBR notes in the Customer Success folder to factor in relationship risk and open commitments." Copera AI reads your documents too, so the score reflects the account plan, not just the numbers — the kind of context a formula alone can't see.
- Keep it private first. Swap "post it in #customer-success" for "DM the summary to me" to get a personal read before anything is written back or shared.
- Plan the saves. Add "…then check my calendar and suggest a good slot this week to call the three biggest Red accounts." Copera AI can read your schedule and turn the score into a to-do list.
- Drill in. After it runs, ask "Why is Acme Red?" — it's already looking at the same accounts and will explain the score from the underlying tickets and usage.
To get this on a rhythm, just ask each Monday — dictate it hands-free with Bankai voice if you're heads-down — or save the prompt and reuse it. For a version that runs entirely on its own, see the next section.
Make it a standing teammate
AI teammates are rolling out gradually. If you don't see them in your workspace yet, they're on the way — everything in this guide can be built today as an Automation plus an on-demand ask to Copera AI.
When one is available, you can put a named teammate — a "Customer Health Analyst" — in charge of this whole recipe. On its own weekly schedule and under its own identity, it reads your accounts and support boards (and the account notes you point it at), scores every customer, writes the score and status back to each account, and posts the at-risk summary to your success channel — no one has to ask. A teammate's read tools are always on, and writing rows and posting to channels are opt-in per agent; it can't delete data or send anything outside your workspace, so the scoring stays a safe, internal briefing. Until then, the automations above plus the on-demand ask do the same job today. Learn more in AI Teammates.
Tips
Define the rubric once, and keep it. Write down your bands and weights — what counts as Green, Yellow, and Red — and put the same rubric in your formula and your Ask Copera AI prompt. Applying one definition every time is what makes scores comparable across accounts and trends meaningful over time. Drifting definitions produce numbers no one trusts.
Let the formula do the math, and the AI do the judgment. Use the formula health score for the deterministic, always-live number from your numeric signals. Use Ask Copera AI for what a formula can't reach — reading the actual ticket threads, weighing a tense QBR note, spotting an account that looks fine on paper but has gone quiet. Together they give you a score with both rigor and context.
Keep last-touch honest. Health scoring is only as good as the freshness of its inputs. Have CSMs log calls and check-ins as row comments or stamp a Last contact date, so "no meaningful touch in 30 days" actually means the relationship is cold — not just that no one updated the board.
Review the writeback, then trust it. Keep the default Ask before acting mode so you see each score on a review card before it lands on an account. After a couple of clean weekly runs, you'll know exactly what to expect — and you can always have it DM the summary to you first for a private read before the channel post.
Turn a Red into action, not just a number. A score that no one acts on is theater. Pair this with a follow-up motion: when the summary flags a Red account, use the stale-lead follow-up pattern to schedule the save, or let the alert automation open a "Save play" task for the CSM automatically.
Frequently Asked Questions
Can Copera AI calculate a customer health score?
Yes. Ask Copera AI reads your accounts board and your support board, combines each customer's usage and trend, open and escalated tickets, latest satisfaction score, and renewal timing, and computes a health score against a rubric you define. It writes the score and status back to each account and posts a summary of the at-risk accounts to your success channel. It runs on demand today and, by default, shows every change on a review card first.
What signals go into a customer health score?
Whatever you keep in your workspace. Common inputs are product usage and its trend, the number of open and escalated support tickets, the latest CSAT or NPS score, how close the renewal is, and time since the last meaningful touch. You decide the weighting and thresholds; Copera applies the same rubric to every account so scores stay comparable.
Can customer health scores update automatically?
The signals stay current automatically — automations refresh each account's ticket counts, survey scores, and usage flags as things happen, and alert the CSM the moment an account tips into at-risk. The full cross-board score is produced on demand by Ask Copera AI today; a standing teammate that rescores your whole book on its own schedule is rolling out in early access.
Does the health scorer contact customers?
No. It writes scores to your internal accounts board and posts a summary to an internal success channel or your own DMs. Nothing reaches a customer. Ask Copera AI always acts as you and within your permissions, so it only reads the accounts and posts to the channels your account can already access.
How is a health score different from churn prediction?
A health score is a present-state read you define from your own signals; it's the leading indicator that surfaces churn risk early enough to act on. Instead of a black-box probability, it's a transparent score you can trace straight back to the tickets, usage, and survey data behind it.
Do I need to write code?
No. Build the signal layer on the visual Automations canvas by connecting blocks, and produce the scored read by describing it to Ask Copera AI in plain English. You review every score and message before anything is written or sent.
Related
- AI Pipeline Health Report, Sent Daily — the sales-side sibling that reads a board and narrates risk the same way
- Answer Team Questions From Your Own Docs — another Support & Success recipe that reads across your workspace
- What Copera AI Can Do — the full list of what Ask Copera AI can read, reason about, and write
- Ask Copera AI — the assistant that reads your boards and scores every account
- Formulas — build the weighted health-score column
- Automations — the trigger, condition, and action reference behind the signal layer