Churn risk indicators for customer success: the signal outside your product, before the renewal

Churn risk indicators are the signs that a customer may not renew: falling product usage, a low Net Promoter Score, rising support tickets, a sponsor who leaves and late payments. Software that reaches the customer’s own customers, such as an email, help desk or payments platform, has one more outside the product: a welcome email that stops arriving, or emails that start coming from another email platform. Obsession keeps what each account sends its own customers, dated with the email or screenshot, by API, MCP or a table, so that indicator reaches the health score before the renewal call.

Churn risk indicators are the signs that a customer may not renew. The usual ones are product usage falling, a low Net Promoter Score (NPS, the 0 to 10 survey of whether a customer would recommend you), support tickets piling up, a sponsor leaving and invoices paid late. Customer success teams combine them in a customer health score and work the accounts that turn red before the renewal.

Some software reaches the customer’s own customers: email and SMS platforms, loyalty and reviews apps, help desks and AI support, checkout and payments, messaging, identity and onboarding. For these vendors there is one more indicator, and it sits outside the product. It is what the account’s customers stop receiving, such as a welcome email that no longer arrives, texts that go quiet or emails that start coming from a new email platform.

Gross revenue retention is the share of last year’s recurring revenue a team keeps before any expansion. Each point of it is worth $500,000 a year on a $50m book, $100,000 on a $10m book and $20,000 on a $2m book (example books). An outside indicator can show weeks before the renewal call, which is the time a customer success manager needs to save the account.

The churn risk indicators customer success teams track today

Most customer success teams score the same families of indicators, and the platforms that build health scores describe them in close agreement.

Gainsight’s guide to health scores calls a health score “a predictive metric” of the likelihood of renewal, built from usage, support, sentiment and engagement data. ChurnZero’s definition splits the inputs into quantitative factors, such as product usage and support history, and qualitative ones, among them relationship quality and competitor risk. Vitally suggests 4 metrics: product setup, product usage rate, NPS and the CSM pulse, a customer success manager’s own rating of the account.

CustomerGauge adds a quieter sign it calls the absence of signal: an account, often a passive one, that stops engaging altogether and churns without warning. Common Room’s guide adds signals from outside a team’s own tools, such as a champion joining a new employer and an account hiring to grow.

Customer health score indicators, and how the tools weigh them

A customer health score turns those indicators into one number or colour per account. As of 11 November 2026, each platform’s own pages describe the weighting this way.

• Gainsight gives example weights of usage 40%, support 25%, sentiment 20% and executive engagement 15%. Its 0 to 100 scale reads 71 to 100 as healthy, 31 to 70 as at risk and 0 to 30 as critical, and it recommends 4 to 6 signals.
• ChurnZero suggests 5 to 7 factors on a 100-point scale, each worth at least 10 and at most 20 points.
• Vitally’s example is product setup 20%, product usage rate 40%, NPS 20% and CSM pulse 20%, with weights set by “which metrics best predict the outcomes you care about”.
• HubSpot’s customer success workspace builds a score from groups of events or properties, each adding or subtracting points, with score decay that lowers a score as the event behind it ages. It needs Service Hub Professional or Enterprise and a Service Seat.

Every one of these inputs comes from your own product, your help desk, your surveys or the account’s people. Usage shows what your product did for the account. It can’t show what the account’s customers received from anyone else, and a move to a competitor can show there first.

Churn risk signals outside your product: what the account’s customers receive

The outside indicator is what an account’s own customers receive from it, seen the way one of those customers sees it. Sign up, leave a basket, opt in to texts, ask the chat a question, and keep every message with the time it arrived. It sits behind the account’s sign-up, in what search engines can’t reach: the web behind the login. When your product sends those messages or runs that page, a change there is a change in how the account uses you.

A stopped message has 2 readings, and the difference decides the call. If the emails stopped and nothing replaced them, the account has broken something, and the customer success manager who brings the screenshot and the fix strengthens the renewal. If they start arriving again from another email platform, the account has moved that work to a competitor, weeks before anyone says so. An email’s headers, the lines of technical information at the top of every email, name the platform that sent it and tell the 2 readings apart.

Email and SMS platforms

The account’s welcome series, browse and basket reminders and texts are your product’s output. Watch for a triggered flow that stops, such as an abandoned cart email that no longer follows a basket, and for texts that go quiet after an opt-in. Watch hardest for emails whose headers name a different sending platform. That change can sit beside steady usage, because an account can keep its campaigns on one platform while it moves its flows to another.

Loyalty and reviews apps

The loyalty programme’s emails after sign-up and the reviews on product pages are what the account’s customers see of you. A points email that stops after sign-up, or product pages whose reviews disappear or reappear in another app’s widget, both show on a screenshot taken as a shopper.

Help desks and AI support

The account’s chat and support email answer its customers through your product. Ask the same question each week and time the reply against the account’s own help page. A reply that takes 31 hours when the page promises 24, or a chat that stops answering, is a change in how the account runs support.

Checkout and payments

Walk the account’s checkout up to the payment step and screenshot it each week. When your payment option leaves the page, your volume report shows the drop, and the screenshot shows what replaced it.

Messaging, identity and onboarding

The account’s sign-up sends a verification email or a login code by text. A code that arrives late, stops arriving or comes from another provider shows on a new sign-up, logged with its arrival time.

How to add the outside indicator to a customer health score

Add it as its own group, beside usage, support, sentiment, relationship and commercial. Weight it by how much of your product the account’s customers see. For an email and SMS platform that is most of it; for a payments provider, it is the checkout page.

Set the group’s rules before the first score.

• A change of email platform overrides the total. Set the account red whatever its other scores, because a steady usage figure can hide a flow that has moved.
• A message is missing only once its window closes. Choose the window per message, such as 1 hour for a basket reminder or 24 hours for a welcome email, and score it late or missing after that.
• Findings fade once fixed. Score decay of the kind HubSpot’s workspace applies to ageing events fits here: once the new email confirms a fix, the finding’s weight fades over the weeks after it.
• Keep the screenshots in your CRM. Every finding carries its date and its screenshot or email on the account’s record in your CRM, so the renewal conversation starts from the same page. The guide to CRM data enrichment fields sets out a field for each.

A weekly churn risk routine for customer success

1. Pull the renewals. Every account renewing in the next 120 days, largest annual recurring revenue (ARR) first.
2. Read what each account’s customers got this week. New emails and texts, anything past its window, the platform behind each email, and the checkout and chat where your product sits.
3. Score the group. Green, amber or red by the rules, with a change of email platform overriding the total.
4. Call the reds within 2 working days. Open with the screenshot and the date. Where the account broke something, offer the fix; where it moved work to a competitor, ask what drove the move while there is time to answer it.
5. Hold the ambers to the next touchpoint. Check again once the window closes, and bring the screenshots to the next quarterly business review; the QBR template has a slide for it.
6. Report on Friday. The accounts that changed colour, the renewal ARR now red, and what each call found.

The same reading finds expansion. An account that never set up texts, or sends no browse reminder, has a gap your product fills, and the screenshot shows its team what their customers are missing.

How to check what an account’s customers receive, by hand

A customer success team can start this week with a spare inbox and phone number per account.

1. Pick the accounts

Start with renewals in the next 120 days and the accounts with the most ARR. For each, write down which of its customer-facing steps your product runs: the welcome email, the basket reminder, texts, the chat or the checkout.

2. Give each account its own inbox and number

Use an address your company owns, kept for that account and nothing else, so every message traces back to it. Add a phone number for texts where the account sends them.

3. Sign up the way any customer does

Join the newsletter, create an account, leave a basket and stop before payment, and opt in to texts. Never enter details that aren’t true.

4. Log every message with its time

Record each email and text with its arrival time and subject in one sheet per account, and set the window for each. A pass: every expected message logged inside its window.

5. Read the sending platform from the headers

In Gmail, More, then Show original opens the full header, which shows where an email came from. Note the servers in its Received lines and the domain in its DKIM signature, a signature the sending domain adds to each email. A change from one month to the next marks the week the account’s emails started coming from a different platform.

6. Repeat every week, and after any change

Sign up again each month to see the welcome series from the start, and again the week after any relaunch or migration you hear about.

By hand, the check takes about 15 minutes a week per account, covering the inbox log, a share of the monthly sign-up and the header check. Multiply your accounts by 15 and divide by 60 for hours a week, then multiply by your team’s cost an hour and by 52.

The same sum gives $78,000 a year for 100 accounts and $15,600 for 20. By hand, the check holds for the 10 to 20 largest renewals; a whole book needs an inbox and number per account, kept open for months, with every original email saved.

Obsession collects what companies show only to customers, step by step: their screens, emails and texts, member prices, trial limits, quotes and cancel offers, each dated with a screenshot. Teams and AI agents look up any company through the API or MCP, and a company not yet covered is added on request. For customer success, it collects the churn signals outside your product for every account on your book. That covers each email and text an account sends its customers, its sign-up and basket walked daily, and the sending platform behind each email with the week it changes. Each fact lands dated on the account’s record in your CRM, so you can see what it does to your renewals, retention and net revenue retention.

Most customer success and sales teams already use similar data about their accounts, from contact databases and intent tools to Exa, which calls itself “The world’s data for AI agents”. Its page for GTM teams offers company and people data with cited sources, such as hiring, funding, job changes and product launches, in seconds. Exa’s pages don’t describe keeping what a company sends the people who sign up. An agent can call both, and the comparison with Exa sets out which questions each one answers.

Churn risk tools compared

Each option reads a different source of churn risk. We make Obsession, so every other option is described from its provider’s own pages as of 11 November 2026. The sales page shows how the same facts serve account briefs and battlecards.

For the indicator outside your product, use Obsession. It keeps what every account on your book sends its own customers, dated, with the sending platform behind each email, for as long as you follow the account. Gainsight, ChurnZero, Vitally and HubSpot turn your product, support and survey data into a health score and its playbooks. Common Room adds job changes and hiring at the account’s company. The outside indicator becomes one more group in whichever health score you run.

Limits and next steps

• It fits products the account’s customers touch. The outside indicator says the most where your product sends the account’s emails and texts or runs its checkout, chat or sign-up. Back-office software, such as payroll or accounting, leaves little for a customer to see.
• Every check stops before payment. Order confirmations and shipping texts sit with the account’s own data and yours.
• It shows what changed, and the account knows why. A stopped flow or a new platform opens the conversation; the reason comes from the call.
• People stay with your CRM. Sponsor changes come from your own records and public profiles, and Account watch tracks no individual staff.

The same changes tell sales which prospects are changing platform, which the guide to buying signals examples covers for sales development teams. Start with the 10 largest renewals in the next quarter, sign up to each this week, and read the headers before the next forecast call. Sales intelligence from behind the login shows the same view for every role on a revenue team, with an example for each.

Frequently asked questions