Build a Renewal Forecast You Can Defend

    You are a revenue operations analyst. Build a renewal forecast from my book that survives being questioned.

My accounts renewing this period: {{list_with_value_and_date}}
What I know about each: {{health_usage_relationship_notes}}
Historical renewal rate: {{if_known}}
Typical notice period: {{notice}}

Part 1 — Categorise every account by evidence, not by feeling. Use these definitions and hold to them:

- **Committed** — someone with authority has said yes, or the contract auto-renews and the notice window has passed. Nothing else counts as committed. Not "they love us." Not "no reason they'd leave."
- **Likely** — active use, engaged champion, no adverse signals, positive last conversation, and someone with authority has been spoken to this quarter.
- **At risk** — one or more specific adverse signals. Name them per account.
- **Likely loss** — they have told us, or the signals are unambiguous.
- **Unknown** — I have not had a real conversation with this account this quarter.

Then tell me how much revenue sits in Unknown. This is the most useful output of the whole exercise. A forecast built on accounts nobody has spoken to is not a forecast, and Unknown should be treated as at-risk until it is worked, not quietly rolled into Likely.

Part 2 — The numbers.
- Gross retention: what renews of what is up, before any expansion. This is the honest measure of whether customers stay.
- Net retention including expansion and contraction, stated separately from gross, because expansion masking churn is the most common way a retention problem stays hidden for a year.
- A range: worst case if every at-risk account leaves, expected case, best case. Say which assumption drives the spread.
- Concentration: what share of the renewing revenue is in the largest three accounts, and what the forecast looks like without them.

Part 3 — Work the book.
- Rank by revenue at risk, which is value times probability of loss — not by value alone, and not by how worried I am
- For each at-risk account: the one action, the owner, and the date, working back from the notice deadline rather than the renewal date
- Which accounts I should stop spending time on because the outcome is already set
- Which Likely accounts are actually Unknown wearing a nicer label

Finish with the two accounts that would most damage the number, and what I do about them this week.