Renewal Forecast

When one person owns sixty accounts and twenty of them expire this quarter, giving all twenty equal attention does not fit into a calendar. Deciding what to do first requires knowing which ones are shaky, and for a long time that judgment was made by feel.

Judgment made by feel ran optimistic. Someone would walk a list of twenty accounts in a quarterly meeting, say each one looked likely to continue, and by the end of the quarter three had cancelled. The worse part was that nobody had been able to name which three in advance. When finance asked what next quarter's revenue would be, the answer had the same texture. "Probably no major problems" is not a number, so the company had nothing solid to plan the quarter against.

Renewal forecasting exists to turn that answer into one with evidence attached. It isn't about removing instinct — it's about writing down what the instinct was based on. Fill in items like weekly active users and the date of the last conversation, and the judgment moves out of one person's head and onto a table where someone else can challenge it.

One round starts by pulling the list and ends by checking the previous round against reality.

1. Pull the list. Every account expiring in the next 90 days, one row each with contract value, expiry date and owner. Accounts set to auto-renew stay on the list; billing continuing and a customer deciding to stay are two different events. Pagedeck added that rule after one auto-renewing account cancelled two weeks into its new term.
2. Fill in the evidence for each account. Pagedeck uses five items: weekly active users over the last three months, whether the core features are in use, the date of the last conversation with the budget owner, the count of open tickets, and the health score — several usage signals weighted into one number from 0 to 100. Blanks get filled with "unknown" rather than left empty, because four months of silence from a budget owner is itself the strongest signal on the row.
3. Sort into three grades: likely, uncertain, at risk. Each grade gets a written rule. At Pagedeck, a conversation with the budget owner inside two months plus steady active users makes it likely; one of the two missing makes it uncertain; being flagged as a red account or carrying more than five open tickets makes it at risk.
4. Weight the grades by contract value and total them. Multiply likely by 100 percent, uncertain by 60, at risk by 20, and add. What comes out is a range rather than a single figure, and the range is what finance receives. The weights come from the last four quarters of actual outcomes; invented percentages produce an invented total.
5. When the quarter closes, check the forecast against what happened. Any account marked likely that left gets a churn reason — a category recording why a customer left — and if that reason wasn't visible in the step 2 items, the items change. Without this loop the forecast stays frozen at the assumptions of its first round.

The account owner fills in the evidence and assigns a first-pass grade. A manager then reads and adjusts those grades, because grades assigned alone drift optimistic. That drift isn't a character flaw, it's a property of the seat: writing "at risk" next to a customer you've supported for a year feels like writing that your own work went badly.

Finance is on the receiving end. The range from step 4 goes into the next quarter's revenue plan, and when the gap between the low and high ends is wide, finance asks which accounts caused it. Grace reviews the largest contracts herself, since when one account is 30 percent of a quarter's forecast, a single grade on that account moves the company's plan.

The cadence is one full round in the first week of the quarter, then a monthly pass over the grades only. Grades change mid-quarter more often than people expect. That a champion moved on, or that a budget was cut, tends to surface abruptly in something like a quarterly business review, and one sentence there drops an account from likely to at risk.

Forecast rounds and the renewal stage run coupled together. The forecast sets the order; the actual work begins in that order from 90 days before expiry. Start without an order and you work from whoever is loudest, which pushes the quiet at-risk accounts to the back.

"Twenty expirations this quarter: twelve likely, five uncertain, three at risk."

Those three counts are the starting point for how the quarter's hours get spent.

"You marked this likely, but the last call with the budget owner was four months ago. Drop it to uncertain."

The evidence items exist so they can argue with the grade.

"The low end of the forecast is under next quarter's plan. Let's work two of the three at-risk accounts this month."

The forecast reaches past one person's calendar into the company's plan.

At Pagedeck, Sofia spends two days on this in the first week of the quarter. Filling in twenty rows takes half a day; the rest of the time goes to the accounts whose grade is genuinely unclear. Likely and at risk are largely settled already, and the uncertain column is where intervention still changes the result.

One rule holds it together: when a grade goes down, the reason goes in a single line beside it. Reading those lines in the next round shows immediately whether the last judgment was sound, and without them a team repeats the same misread for four quarters.

The most common assumption is that the point is accuracy. The word forecast makes it look like a score. But the point is allocation. If you graded an account at risk, worked it, and it renewed, the forecast was wrong and the job was done right. Make accuracy the target and people quietly stop intervening in order to be correct. What deserves the scoreboard is whether the at-risk column shrank because someone worked it.

Using the grades to evaluate the account owner is the other frequent mistake. A low share of at-risk accounts looks like good management, so it's tempting as a metric. Once it is one, writing "at risk" becomes costly. Owners move unclear accounts to uncertain and uncertain ones to likely, and at quarter close the company lands well below even the low end of its own forecast. What's worth evaluating is how close last quarter's forecast came to reality, not the shape of this quarter's grade distribution.