How to forecast revenue when deals are lumpy

If two or three orders decide your quarter, a single forecast number will mislead you, and you may plan salaries and stock on a figure that never happens. The remedy is to stop averaging. Separate the big deals from the everyday ones, weight the everyday ones from your own records, handle each big deal by name, and report a low, a middle and a high figure. Plan commitments you cannot reverse on the low.

The weighted number that may never occur

Suppose everyday orders in your pipeline weigh to ₹28 lakh. Two large deals of ₹24 lakh each are each a coin toss. Weighted, they add ₹24 lakh, and the forecast reads ₹52 lakh. But look at what can actually happen.

Both big deals lost :  28 lakh
One won, one lost   :  52 lakh
Both won            :  76 lakh
Forecast, weighted  :  52 lakh  (one-in-two chance of landing exactly here)

Here the weighted figure happens to match the middle outcome, but change the odds on one deal and it lands on none of them. The everyday deals behave like a crowd, so their average holds. The big ones are single events. Treat them differently.

Where to draw the line between crowd and event

Pick a size rule you can state. A workable one: any deal worth more than a sixth of your usual quarter is a big deal and goes on its own list. Everything else is everyday. Revisit the rule if your order sizes change. If you have only a handful of orders a year, every deal is big, so skip weighting and keep a list of named deals with an honest note on each.

Weights from your own closed deals

For the everyday deals, take your closed deals from the last year or two. For each stage, such as quotation sent or negotiation, count how many deals reached it and how many ended as orders. That fraction is your weight. It is yours, not an industry figure, because your buyers behave their own way. With fewer than about twenty deals per stage, say that the weight is rough and widen the range.

Add an age rule. A quote sent last week and one sent four months ago are not equal. Any deal sitting in a stage longer than your usual cycle drops to a lower weight, or out of the forecast, until the buyer gives a date. The stages must describe the buyer's actions, as in pipeline stages for a service business. A stage called "good feeling" cannot be weighted.

One line for each big deal

Write the same five facts beside every big deal, and update them whenever something moves.

Buyer / deal        : Seafood processor, 2 cold rooms, ₹24 lakh
Decision date given : 12th, in writing
Who decides         : Two partners; one has not seen the quote
What could stop it  : Bank loan sanction
Level               : possible

Define the three levels before you use them. Firm means the buyer has paid an advance or issued a written order. Probable means the price is agreed and one named approval is pending. Possible means the buyer is still comparing. Do not attach a percentage to a single deal and pretend it is science.

Build the range

For the low figure, take the everyday total, trimmed if recent months fell below history, plus only the firm big deals. For the middle, add the probable ones. For the high, include the possible ones and the everyday total as your best recent month suggests. Report all three, every time.

Use the low for commitments you cannot reverse, such as a lease or a salary. Use the middle for ordinary spending. Use the high only to see what you would need to deliver.

A case: a cold-room builder in Nellore

Here is an invented case. A firm in Nellore builds insulated cold rooms for fish and fruit traders. Most months bring a few small jobs for traders, and now and then a processor orders a large installation. The owner's earlier forecast showed the small jobs plus one big order at half value, and the quarter missed it in one direction or the other.

Once the lists were split, the mistake appeared on one line. A large deal was marked firm because the trader said yes on the phone. No advance had come and nothing was in writing, and a bank sanction was still awaited. After the owner defined firm as an advance or a written order, that deal moved to probable and the low figure fell. The owner delayed a planned hire by a quarter. When the sanction came through three weeks later, the quarter closed close to the middle figure, though one probable deal slipped into the next quarter and the hire went ahead a month late.

The rule did not make the forecast prettier. It made the low figure honest, which is what the lease depends on.

Look back at every quarter

Keep each version of the three numbers with its date. When the quarter ends, write the actual result beside them. If actuals keep landing near the low, your middle is too generous. If they beat the high, your levels are too cautious. That comparison is how the forecast improves. For each big deal, ask one question when you review: what has changed since we last looked? If the answer is nothing and the date has passed, drop it one level. A deal does not stay probable on its own.

When this is more than you need

With hundreds of small orders, the weighted method alone is fine and the big-deal list is overhead. The method also fails when stage movement is not dated, because without dates you cannot compute ageing and the history is guesswork. Start with the basic method in a simple forecast for SMEs, then add the layer for lumpy orders.

Draw up the first range with real deals

List every open deal above your size line and fill the five facts beside each. Count last year's results by stage for the rest, and date the sheet. If the records need cleaning first, do that before building weights on them. If you want help setting the sheet up on your own records, tell us what you need; how forecasting fits the wider job of running a team is covered under sales.