How Do I Forecast When Every Month Is Different? A CFO's Answer for Lumpy, Seasonal Revenue

Founders with seasonal or project-based businesses ask me this almost every week. A tour operator that books most of its year in two busy seasons. A construction company waiting on one large project to close. A beverage brand whose sales swing hard with retail orders. They all want a forecast, and they all believe their own revenue is too unpredictable to forecast at all.

It is not. The lumpiness is exactly why you need one. Here is how I build a forecast that holds up when no two months look alike, and why the founders who wait for things to settle down are making the most expensive mistake of all.

"My revenue is all over the place, so what is the point of forecasting?"

The point is not prediction. The point is steering.

A forecast is not a promise to the universe that you will hit a number. It is a written-down version of what you currently expect, so that when reality comes in different, you can see the gap and react early. A founder with no forecast finds out about a slow quarter when the bank balance gets scary. A founder with a forecast sees it coming a couple of months out, while there is still time to do something about it.

Lumpy revenue does not make a forecast useless. It makes a forecast valuable. The whole reason you feel out of control is that the swings are surprising you. A forecast turns surprises into things you already half-expected.

"How do I actually build a forecast when I cannot predict the lumps?"

You do not predict each lump perfectly. You build from what you can reasonably estimate, and you keep the model simple enough that you will actually update it.

Start with what you already know. Most businesses, even the lumpy ones, have a backbone of recurring or repeatable revenue. A software startup has its subscriptions. A healthcare clinic has its standing appointments. A marketing agency has its retainers. Put that baseline in first, because it is the part you can forecast with real confidence.

Then layer the lumpy stuff on top, line by line, based on what you can see in front of you. For a project-based construction company, that means listing the specific jobs you expect to bill and roughly when. For a beverage brand, it means estimating orders by the channels you sell through. For a tour operator, it means spreading bookings across the seasons the way they have historically landed. You are not guessing in the dark. You are translating your real pipeline and your real history into months.

Build it monthly, not annually. An annual number hides every cash crunch inside the average. Twelve monthly numbers show you the actual shape of the year, including the months where money goes out but very little comes in. Those are the months that hurt, and they are exactly the ones an annual view erases.

Keep the first version rough on purpose. A forecast you can update in twenty minutes gets updated. A forecast that takes a full day gets abandoned by the second month.

"Looking back at my revenue over the years, how do I use the trend?"

History is your best forecasting input, and most founders never look at it properly. Pull two or three years of monthly revenue and lay it side by side. You are hunting for two things: the trend and the pattern.

The trend is the overall direction once you ignore the noise. Is the business growing, flat, or sliding, year over year? The pattern is the seasonality. A tour operator might do the bulk of its revenue in two stretches of the year, with quiet months in between, and that shape repeats. A retail-driven beverage brand might spike around specific buying seasons every year. When you stack the years, the pattern jumps off the page.

Once you can see the pattern, you stop treating a slow month as a crisis. A quiet winter for a seasonal business is not a problem. It is Tuesday. The trend tells you whether to forecast next year higher or flatter than this one. The pattern tells you how to shape it across the months. Together they turn raw history into a starting point you can defend.

"How are we tracking toward the forecast, and how often should I review it?"

This is the question that matters most, and it is the one founders ask least. Building the forecast is the easy part. The value is entirely in the comparison.

Once a month, sit down and put your forecast next to what actually happened. Forecast versus actual, line by line. Where you were close, leave it. Where you were off, ask the only question that counts: was this timing or was this real? Timing means the revenue is still coming, just later than you expected, so you slide it to a future month. Real means the revenue is not coming at all, so you cut it and you find out why.

Then you update the forecast with what you now know and roll forward. The third version of your forecast is far better than the first, because each month of actuals teaches you something about how your business really behaves. This monthly rhythm is the whole game. A forecast you build once and never revisit is a wish. A forecast you compare and update every month is a steering wheel.

Monthly is the right cadence for almost everyone. Weekly is too noisy for revenue and turns into busywork. Quarterly is too slow, because by the time you spot a problem you have lost a quarter to fix it. Once a month, every month, no exceptions.

"Should different channels or segments get their own forecast lines?"

Often, yes, and this is where lumpy businesses get a lot of their accuracy back. If a beverage brand sells through several retail channels, lumping them into one revenue line hides what is actually driving the swings. Break the forecast out by channel and the lumpiness starts to make sense, because one channel orders on a totally different rhythm than another.

The same logic applies to a clinic with several service lines, or an agency with project work alongside retainers. When you forecast each stream on its own terms, the messy combined number resolves into a few cleaner, more predictable ones. You also learn faster, because your monthly forecast-versus-actual review now tells you which specific channel missed, not just that the total came up short.

The caution: do not over-split. If breaking out a line does not change a decision you would make, it is just extra maintenance. Split where the streams genuinely behave differently. Keep everything else together.

"Can the forecast connect to my live sales data and update itself?"

Eventually, and it is worth aiming for, but do not let it become an excuse to delay. Plenty of founders tell me they will start forecasting once the model can pull live numbers automatically. That is the trap.

Start manual. A simple model you update by hand each month teaches you how your business actually moves, and that understanding is what makes any automation useful later. Once the rhythm is a habit and you know which numbers matter, then connecting it to your live sales data to refresh on its own is a genuine upgrade. It cuts the busywork and keeps the comparison current. But automation on top of a forecasting habit is leverage. Automation instead of a forecasting habit is just a fancier way to avoid looking.

The real lesson

Waiting for your revenue to stabilize before you forecast is backwards. It will not stabilize, and even if it did, you would have spent all that time flying blind. The lumpiness is permanent, so the skill you actually need is forecasting through it.

Build the simple version this month. Compare it to reality next month. Update it and roll forward the month after. By the third pass you will have something most founders never get: a clear, current view of where the money is heading, and enough warning to do something about it.

Peter Xia is a CPA and fractional CFO. He shares finance breakdowns for founders on @CanadianCFO.

Next step: see it in your free Instant CFO Snapshot.