Sign in

Blog · Coverage and territory · Customer service

Silence and surge as churn signals: account coverage for customer service teams

How a customer service or support team rolls tickets up to the account and reads the two signals that predict churn, an account that went quiet and an account whose volume doubled, with proactive coverage of high-value accounts and a reconciliation to the account list and revenue.

The short answerRoll tickets up from the ticket to the account and compare each account to its own history. An account whose contact fell to nothing after a steady rate is silent; one whose volume doubled is surging; a high-value account with no proactive contact in the window is uncovered. Rank all three by revenue, assert that the accounts in the ticket roll-up match the account list, and give team leads the list every week.

A service desk measures response time to the minute and misses the account that stopped calling. Ticket data is organised by ticket; the signals that matter are organised by account. This guide sets out the roll-up from ticket to account and the three lists that come out of it every week.

The measures

Per account:

Silent = tickets in the window fell below a fraction of the account's own trailing rate Surging = tickets in the window above a multiple of the trailing rate, or tickets past SLA above a threshold Uncovered = high-value tier with no proactive contact in the window

Per team:

Accounts covered ÷ accounts assigned, and the three lists ranked by revenue

The rows you need

  • Tickets: one row per ticket with account, opened date, closed date, status, SLA flag, product. From Zendesk, Salesforce Service Cloud, ServiceNow, Freshdesk or an export.
  • Account list: accounts with tier, revenue, team and owner.
  • Proactive contact log: optional. Account, date, type.

Account identifiers only.

The roll-up

  1. Account by week: tickets opened, past SLA, proactive contacts.
  2. Account: trailing rate, current rate, silent flag, surging flag, uncovered flag.
  3. Team: counts of each, ranked by revenue. Assert that the accounts with tickets are on the account list, and count those that are not.
  4. Company: assert teams to the company, and the account list's revenue to the figure finance reports.

accounts with tickets ⊆ account list, and revenue(account list) = reported revenue

The first assertion catches tickets logged against accounts the company does not recognise, which are either unmapped customers or mistyped identifiers.

A worked example

One team, this week, three signals from a book of forty accounts.

Account Tier ARR Trailing tickets per month This month Signal
2291 1 $180k 12 0 Silent, eight weeks
0876 2 $60k 4 9, three past SLA Surging
1301 1 $95k new 0 Uncovered, onboarded Tuesday

Account 2291 raised a dozen tickets a month for a year and none for eight weeks, and no one has called. That is the line the team lead reads first. Account 0876 doubled after an upgrade and has tickets past SLA; that is the one that becomes an escalation on Friday.

Where it goes wrong

Silence measured against a global norm. A small account that always raised one ticket a quarter looks silent every month. Measure against the account's own history.

Accounts under several identifiers. Tickets logged against a contact's email domain, a subsidiary and the parent split one account into three. Map to the account the company bills.

Surge from a known cause. A release that broke something surges every account at once. Show the surge, and let the team mark the release so the list is not every account.

Proactive contact not logged. Calls that are not logged do not count. The measure only sees the log; the process has to feed it.

Every Monday, per team

Mapped once, the ticket export and the account list produce the three lists per team every week, reconciled to the account list and to revenue where it is loaded. Covirage builds this from the exports as they are. The customer service page describes the setup, and you can upload a sample ticket export and see the roll-up on your own rows.

Questions people ask

Why is silence a churn signal?

Because an account that used to raise tickets and stopped has usually stopped using the product, or moved to a replacement, before anyone cancels. Silence against the account's own history is the earliest signal in the service desk data.

What counts as a surge?

Ticket volume above a multiple of the account's own trailing average, typically double, or tickets past SLA above a threshold. Both point at an account in trouble, and the second is the one that ends in an escalation.

Do we need revenue data?

No. Tickets alone give the signals. Revenue ranks them, so the team lead's list opens with the silent account that pays the most, and reconciles the roll-up to the account list finance uses.