How to forecast revenue from a lumpy pipeline

Forecast a month where three deals decide everything: named-deal scenarios instead of weighted averages, and how to measure your own accuracy.

Three deals decide the month

You have six live opportunities. Two of them are larger than the other four combined. If one of the big ones signs you have a strong quarter. If both slip by three weeks you are short on salaries. This is the ordinary condition of a small business selling considered purchases and no spreadsheet makes it go away.

What a forecast can do is tell you which version of the month you are in early enough to act. That is the whole purpose. A forecast is a planning instrument for your own decisions about hiring spending and borrowing. It is not a prediction and it is not a target.

What it costs and what it will not do

Forecasting creates no revenue. It is pure overhead and the only return is better decisions made earlier. If you would not change a single decision based on the answer then do not build it.

The cost is about an hour a week and a willingness to be wrong in writing. You do not need software. A spreadsheet with one row per deal is enough until you have more deals than a screen will hold. What you do need is somebody who will record what the buyer actually said rather than what the salesperson hopes. That is the scarce input and no tool supplies it.

Who should skip this

Skip percentage weighted forecasting entirely if a single deal can move your month by a quarter or more. Multiplying a large deal by sixty per cent produces a number that cannot occur. The deal either closes or it does not and the weighted figure is a month you will never have. Weighting only behaves sensibly across many similar sized deals.

Skip pipeline forecasting altogether if you run a retail counter a restaurant or any business with daily transactions. There is no pipeline to read. Forecast from recent trend and seasonality instead. Skip it also in the first months of a new business. You have no conversion history so any figure you produce is a guess dressed as arithmetic. Forecast your activity instead and start forecasting revenue once you have closed enough deals to see a pattern.

The method that survives lumpiness

Forecast named deals rather than totals. Each live opportunity gets a value an owner and two dates. The first date is when the buyer says they will decide. The second is your own honest date which for most buyers is later. Use your date.

Then sort into three buckets by evidence rather than feeling. Committed means the buyer has confirmed in writing and the only remaining steps are procedural. Likely means the decision maker is engaged and the budget exists but something is still open. Possible is everything else. Your forecast is the committed total with the likely total shown separately as a range. Never blend the three into one number.

Roll it forward every week and record what moved. When a deal slips write down the new date and keep the old one visible. A deal that has slipped three times is telling you something that no probability percentage will.

How to tell whether your forecast is any good

Pipeline value is the flattering metric. It grows every time somebody adds an opportunity and it never falls until you force a clean out. A large pipeline containing old untouched deals is a comfort blanket rather than information.

Measure your own accuracy instead. At the start of each month write the committed figure down. At month end compare it with what was collected. Do that for a quarter and you will learn your personal bias which is almost always optimism. Once you know the size of that bias you can correct for it and your forecast becomes useful for the first time.

The common mistake

The expensive error is spending against a forecast rather than against cash. The forecast says the quarter is strong so you hire take a larger office or commit to a media budget. Two deals slip a month each which is entirely normal and the commitments remain.

Treat committed revenue as the only figure you may spend against and treat likely revenue as a reason to prepare rather than to commit. Businesses do not usually fail because the forecast was wrong. They fail because somebody spent as though it was right.

Booked is not banked

One more distinction keeps small businesses solvent. A signed order is revenue on a date that suits your accounting and cash on a date that suits your client. Add your real collection lag to every line of the forecast and look at the cash version separately. A quarter can be excellent on paper and still leave you unable to pay a vendor in the second week of the month.

This guide explains how something works. It is not legal, tax or financial advice, and rules change. Confirm anything regulatory with a qualified professional before you act on it.