How to Build a 90-Day Revenue Forecast

The value of a quarterly forecast is knowing the shape of the quarter on day one, early enough to do something about it. Getting the number right in the last week costs the same effort and buys nothing, because the quarter has already happened by then. This build is designed to produce a usable read before the period starts.

What question should a 90-day forecast answer?

How the quarter is going to happen, rather than whether you have enough pipeline.

Most teams open the quarter by checking a coverage ratio. Coverage is an input, and it should never be the conclusion. A company can hold 4x coverage and miss badly if that pipeline is concentrated in the wrong stage, dependent on a few large deals, inflated by stale opportunities, or resting on close dates that sellers keep pushing.

The better build decomposes the quarter into the actual sources of revenue and forecasts each one separately. That decomposition is what turns a forecast into an operating plan, because each source has a different intervention attached to it.

What are the three sources of revenue in a quarter?

Carry-over pipeline, in-quarter creation, and pull-forward from future periods. Most teams model the first source, glance at coverage, and ignore the other two. Ignoring in-quarter creation understates the quarter for any team with a cycle shorter than 90 days. Ignoring pull-forward hides the cost of saving this quarter by dragging next quarter's deals into it, usually with discounting attached.

Build each source as its own line with its own rate. In-quarter creation comes from measuring how much revenue was created and closed inside the same quarter historically. Pull-forward comes from measuring how often deals with next-quarter close dates actually landed early, and at what realized value.

How much of day-one pipeline should you count?

A minority of it. Across ORM's customer base, about 20 percent of pipeline with in-quarter close dates on day one closes in that quarter.

That number is the reason a coverage ratio feels reassuring and then disappoints. The other 80 percent of the value carrying an in-quarter close date does not get realized in the period. It slips, it shrinks, or it dies.

Two adjustments bring the carry-over line closer to reality. First, remove stale opportunities. Any deal with no change to stage, close date, or amount in twelve months should come out of the number regardless of the stage it occupies. More than 10 percent of open pipeline across ORM customers meets that test.

Second, apply a realization ratio. Pipeline amounts run above closed-won amounts. An $80,000 average deal size in pipeline against $40,000 in closed-won is the shape of the gap. Forecasting the pipeline amount at face value overstates every quarter by construction.

How should the number move between day 1 and day 90?

It should tighten, not swing. A forecast that jumps in week ten was wrong in week two.

Set an expected trajectory before the quarter starts. Weeks one through four are dominated by in-quarter creation, so the pipeline generation number matters more than the close number. Weeks five through nine are where slippage shows up. The final weeks are execution on deals already decided.

Watch close date changes as the primary signal throughout. A rep moving a close date is the strongest indication a deal is slipping, and a deal that slips from one quarter to the next is less likely to close even when it sits in commit. The earliest signal is quieter than that. It is the absence of any signal at all, meaning no activity, no data changes, and no notes on the record. The mechanics of that pattern are covered in the deal slippage glossary entry.

What accuracy should you expect?

Manual quarterly forecasting on new and expansion business generally reaches around 90 percent, at high effort and with limited responsiveness.

That accuracy comes at a cost. It takes significant time to produce, and it goes stale as conditions change, because the assumptions behind it were fixed when the file was built. A model that updates through the quarter can hold higher accuracy across the full 90 days without manual adjustment, and ORM targets 95 percent on that basis.

Grade the forecast by source rather than in total. Carry-over accuracy, creation accuracy, and pull-forward accuracy each point at a different team. A total that lands on plan while two of the three lines miss in opposite directions is a coincidence, and it will not repeat. The measures to track are defined in the forecast accuracy glossary entry.

What do you do when the day-30 read is short?

Identify which source is behind, because the response window differs for each.

A carry-over shortfall calls for deal-level work. Which deals slipped, why, and what would move them back. That work can continue until the last week of the quarter.

An in-quarter creation shortfall has a much tighter window. If your cycle runs six weeks, a deal created in week seven cannot close in the quarter, so the response has to happen inside the first month or it does not happen at all. This is the source that punishes late detection, and it is the reason a day-one build beats a week-six one. Background on entering the quarter with a realistic coverage view sits in why the 3x pipeline coverage rule is wrong.