How to Forecast Revenue: The Three Sources That Produce a Quarter

How Do You Forecast Revenue Before the Quarter Starts?

You forecast revenue by decomposing the quarter into the three sources that will actually produce it, then valuing each one honestly, rather than checking whether your pipeline clears a coverage ratio. The mistake we see most often at ORM is treating the pipeline you can see as the forecast. The real question is not "do we have enough pipeline." It is "do we understand how the quarter is going to happen before the quarter begins."

Those are different questions with different answers. A team can sit on 4x pipeline coverage and still miss badly if that pipeline is aged, concentrated in a few large deals, or built on close dates the reps keep pushing forward. Another team can open the quarter with thin coverage and beat the number because it creates and closes fast inside the period. Coverage is an input. It was never meant to be the conclusion, which is the whole reason the 3x pipeline coverage rule is wrong as a standalone forecast.

So we build the revenue forecast from the motion, not the snapshot. Three sources decide the quarter.

What Are the Three Sources of Revenue in a Quarter?

Every dollar you book in a quarter comes from one of three places: carry-over deals already in pipeline, in-quarter deals that do not exist yet, or pull-forward deals dragged in from a future period. Most teams model the first, ignore the second, and forget to charge themselves for the third. Split them apart and the forecast starts describing the operating mechanics of the quarter instead of a single number. On day one, only about 20% of the pipeline already carrying an in-quarter close date will close in the quarter, which means roughly 80% of that visible value does not land when the CRM says it will. That is not a data-quality problem. It is the reason the in-quarter and pull-forward sources exist, and why a forecast built only on visible carry-over runs short.

Most teams over-trust the visible pipeline and under-model the invisible pipeline. They study the deals already in the CRM and barely forecast how much will be created and closed inside the quarter, which for many businesses is a large share of the eventual number. Leaving the in-quarter source out does not make a forecast conservative. It makes it wrong. Then they understate what it costs to pull a future deal forward to rescue the current one.

Why Should You Forecast on Closed-Won Values, Not CRM Values?

Because deals close for less than the amount sitting on them in the CRM, and forecasting on the CRM number inflates every quarter you run. We routinely see a pipeline whose average deal size is $80,000 while the average closed-won deal is $40,000. Forecast on the $80,000 and you have baked a 2x overstatement into the carry-over source before you have accounted for a single loss.

Amounts entered early in a deal tend to reflect the best case, and best cases erode as scoping, procurement, and discounting do their work. The fix is not optimism adjustments bolted on at the end. It is valuing each open deal at what deals like it have historically closed for, by segment and by stage. You also do not need pristine data to do this. You need consistent data. Garbage in does not have to mean garbage out, as long as the garbage stays consistent. When you build a sales forecast on closed-won history instead of seller-entered amounts, the carry-over number stops flattering you and starts predicting.

How Do You Age Out Stale Pipeline?

Stale pipeline gets aged out with a hard rule: if an opportunity has had no meaningful activity in 12 months, it does not count toward the forecast. We define meaningful activity as a change in stage, close date, or amount. Anything short of that is a deal sitting still, and a deal sitting still is not a deal.

In the pipelines we model, at least 10% of opportunities are stale and have not been touched in a year. They pad coverage and predict nothing. At ORM each opportunity is grouped by a machine learning model, and every group gets its own close-curve running from 1 to 80 weeks, with most of the expected closing happening before week 12 and very few groups carrying expectation past 52 weeks. An opportunity whose group says it should have closed 30 weeks ago, with no stage or amount movement since, is carry-over on paper and noise in reality. Age it out and what remains is a forecast you can defend.

How Does Seasonality Change the Forecast?

Seasonality changes how much of each source you should expect to land, and most people do not weight for it at all. In the data we see across customers, Q2 and Q4 run stronger than Q1 and Q3, and inside any quarter the third month closes more than the first or the second. A forecast that spreads expected revenue evenly across three months will look behind in month one and sound the alarm right before its strongest closing weeks arrive.

Month one is the trap. Even a healthy quarter can read soft in its first four weeks, and a team that answers that softness by discounting or pulling deals forward damages the number it was trying to protect. Weighting for the pattern is not the same as trusting it blindly. When something structural changes in the market, the historical shape breaks, which is how market shifts break sales forecasts. Decomposing revenue into three sources is what lets you see which one a shift is hitting, whether carry-over is stalling, in-quarter creation is drying up, or discounts are climbing as deals get pulled forward.

How Do You Keep the Forecast Accurate All Quarter?

You keep it accurate by making it dynamic, so it re-reads the three sources as the quarter runs instead of freezing the number you set on day one. A hand-built model of new and expansion revenue can reach roughly 90% accuracy, but it takes real time and effort to produce and it does not move as conditions change. At ORM the models we build target about 95% on new and expansion, hold from day 1 to day 90 of the quarter, and update as the quarter progresses without manual re-tuning. A model trained on your own historical sales performance gets there in about 4 to 6 weeks.

The reason a dynamic model earns its keep is timing. Getting the forecast right in the last week of the quarter helps no one, because by then the quarter has already happened. The value is knowing the likely shape on day one, early enough to build in-quarter pipeline, protect carry-over, or decide whether pulling a deal forward is worth the discount. That is also the honest answer to what sales forecast accuracy you should expect: high, but only if the model keeps reading the quarter instead of the snapshot you took before it began.