The ten questions a head of customer success asks about the base, which accounts are at risk, which are under-using what they bought, which renewals are inside the window untouched, which accounts are surging or silent on support, which have to call twice, which onboardings are behind the milestone that matters, which are over their seat cap, which cost more to serve than they pay, which churned last quarter and why, and what the health score got wrong, each with the computed signal that answers it and the list it produces.
A head of customer success has a book of accounts and a renewal date on each. The questions between now and the renewal are answered by signals the company's own data carries, each producing a list. This guide is the ten questions, the signals, and the validation that says whether to trust them.
| # | The question | The signal | Source | List ranked by |
|---|---|---|---|---|
| 1 | Which accounts are at risk? | Health score, with inputs | Ledger, CRM, tickets, pipeline | Score, then ARR |
| 2 | Which are under-using what they bought? | Seat utilisation and trend | Usage export, contract register | ARR at risk |
| 3 | Which renewals are inside the window untouched? | Renewal calendar, notice window, activity | Contract register, activities | Contract value |
| 4 | Which are surging or silent on support? | Ticket surge and silence against baseline | Ticket export | Revenue |
| 5 | Which accounts have to call twice? | Repeat contact rate against base | Ticket export | Revenue |
| 6 | Which onboardings are behind? | Days past the milestone that predicts renewal | Usage, onboarding log | ARR |
| 7 | Which are over cap? | Active over contracted seats | Usage, contract register | Expansion value |
| 8 | Which cost more than they pay? | Effort ratio against norm | Tickets, revenue | Effort hours |
| 9 | Who churned, and what preceded it? | Churned accounts with their signals two quarters prior | All of the above | ARR lost |
| 10 | Did the score work? | Validation: churn rate by score band | Score history, outcomes |
Validation first. Bottom band churned at 30 percent; top at 3. The score works. Read list one.
List 1. 4471: score 31; run rate down 24 percent; 71 days since contact; ticket surge 4.7 times; two deals slipped. Renewal in 88 days. Row one.
List 2. 4471 again: 34 percent seat utilisation from 71. Same account, same reason.
List 3. 2207: renewal in 40 days, inside notice, no renewal activity logged. Different account, urgent for a different reason.
List 6. 9034: 22 days since start, first integration not connected; the milestone's typical day is 14. The success manager's Tuesday.
List 7. 2210: 131 active on 120 seats, renews in 60 days. An expansion with the evidence attached.
Five lists, four accounts, one week.
| Churned last quarter | ARR | Signals two quarters prior |
|---|---|---|
| 3310 | $90,000 | Utilisation 28%; silence on support; score 24 |
| 2044 | $60,000 | Repeat contact 44%; effort ratio 5×; score 38 |
Both were on the lists. The question is whether they were worked, and the activity log says.
Health score read without validation. A colour code with decimals.
Renewal date as the only calendar. Discovered inside the notice window.
Usage not joined to contracts. The product team has usage; the CS team has dates.
Churn reviewed without the prior signals. The lists are never checked against what happened.
Covirage computes the ten signals from the usage, ticket, contract and CRM exports and produces the lists per success manager every week, with the validation on the score. The sales insights solution describes the setup, and the health score guide covers the first signal and its validation.
The health score, because it combines the others, and its validation, because it says whether to believe it. A score whose bottom band churned at ten times the top band's rate last quarter is the list; one whose bands churned alike is decoration, and the individual signals are read instead.
Utilisation, surge, silence, repeat contact and onboarding milestones lead the renewal by a quarter or more. Churn and the validation lag. The leading signals produce the lists; the lagging ones say whether the lists were right.
Ten rows per signal, ranked by ARR at risk or expansion value. A manager with forty accounts and ten signals still reads under a hundred rows, most of which overlap, and the health score's top ten is usually the week.