How Analysts Forecast Revenue Before Earnings

Analysts forecast revenue before a company reports by taking the topline apart into its drivers, price and volume or their equivalents, forecasting each driver separately, then putting them back together into a range. Each driver is anchored to two things: what management has guided, and what the analyst can observe from the outside, such as capacity, pricing, and demand signals. The output is not one confident number. It is a set of assumptions you can defend one at a time.

This is the step that follows revenue mapping. Mapping tells you where a company’s revenue and profit actually sit. Forecasting asks the next question: given how this business earns money, what is the topline likely to be next quarter, and what would have to be true for it to land higher or lower. If you have not mapped the business first, you are forecasting a blur.

Start by decomposing, not guessing

The wrong way to forecast revenue is to look at last year’s growth, add a few percent, and call it a number. That treats the topline as a single thing that grows on its own. It does not. Revenue is always the product of quantities, and those quantities move for different reasons.

The cleanest way to see this is a company with genuinely different businesses under one roof. Take Reliance Industries in the quarter ended September 2025 (Q2 FY26). Four of its reported segments looked like this.

Segment	Revenue (approx)	Segment EBIT margin (approx)
Oil to Chemicals	₹1,60,600 crore	~9%
Reliance Retail	₹90,000 crore	~6%
Jio (digital services)	₹42,700 crore	~52%
Oil and Gas (upstream)	₹6,100 crore	~83%

You cannot forecast that company with one growth rate. Oil to Chemicals moves with refining spreads and throughput, which are commodity and capacity questions. Retail moves with store count and sales per store. Jio moves with subscribers and average revenue per user. Each of those has its own driver, its own leading indicator, and its own logic. A single blended assumption would be an average of things that have nothing to do with each other.

So the first job is to write revenue as an equation, per segment where segments exist:

Revenue = quantity times price, for each part of the business that behaves differently.

For a telecom line it is subscribers times revenue per user. For a retailer it is stores times sales per store. For a commodity refiner it is volume times realised spread. Once revenue is an equation, you are no longer forecasting a topline. You are forecasting a handful of quantities, and each one can be argued about on its own evidence.

Anchor each driver to guidance and to what you can see

A driver forecast needs two anchors. One is what management says. The other is what you can observe without them.

Management guidance is the more visible anchor, and the words matter more than most people notice. Consider Asian Paints and its forward guidance, given on a public earnings call. On volumes, management guided to “high single-digit volume growth in the band of about 8 to 10 percent.” On profitability, it pointed to an 18 to 20 percent EBITDA margin band and spoke of “maintaining our margin guidance.”

Read what that actually gives you. It is a number and a hedge. The band is 8 to 10 percent, not a point. “High single-digit” is a range with a soft floor. “Maintaining” is a signal of confidence in the current level rather than a claim of improvement. A serious analyst records the exact phrasing, because the phrasing is what you grade later. Next quarter you check whether volumes are tracking toward that band or drifting below it, and you watch whether the language stays confident or turns cautious. Guidance is a forward claim you can hold management to, which is exactly why it is useful as an anchor. For the mechanics of reading a band and its hedge, see management guidance explained.

The second anchor is everything you can observe from outside the company. Capacity that has been announced and is coming online. Price changes you can see in the market. Input costs that feed into pricing. Demand signals from the sector. For a cyclical business, the position in the cycle often matters more than any single guidance sentence, which is why valuing a cyclical company leans so heavily on where you are in the cycle rather than on the last reported number. The point of the second anchor is independence. If your driver forecast only repeats what management told you, you have not done any work. The value is in the places where your read of the observable evidence agrees with guidance, and especially in the places where it does not.

Build a range, not a point

Once each driver has a forecast, you do not pick the single most likely value and multiply. You build at least three cases: a base case, a high case, and a low case, each with a stated assumption for every driver.

Take a paints business as a worked sketch, using the Asian Paints band as the volume anchor. Volume growth might be 10 percent in the high case, 9 percent in the base, and 8 percent in the low, straight off the guided band. Realised price per unit carries its own small range depending on how much of the raw-material move gets passed through. Multiply volume by price in each case and you get three revenue numbers, not one. The spread between them tells you how much of your answer depends on the pieces you are least sure about.

This is the honest part of forecasting. A single number hides its own fragility. A range shows it. If the high and low cases are far apart, the forecast is telling you that the driver work is sensitive to one or two assumptions, and those are the assumptions to keep watching. If the range is tight, you have more genuine conviction. Either way you have learned something a point estimate would have concealed.

Building the forecast from the reported statements upward, rather than from a data vendor’s pre-chewed sheet, is what makes each assumption traceable. That discipline is the subject of building a financial model from primary sources: when every driver ties back to a line you pulled yourself, you can defend the forecast and you can find the adjustment when reality disagrees.

State what would move it

A forecast is not finished when you have a range. It is finished when you have written down what would move it, and by how much. This is the sentence most amateur forecasts never write.

For each driver, name the one or two things that would push it out of your range. For volume, it might be a demand slowdown in a key end-market or a pricing action by a competitor. For price realisation, it might be a raw-material move you assumed would be passed through but was not. For a segment business, it might be a capacity delay that pushes volume into the next year. Writing these down turns the forecast into a monitoring plan. When the print lands, you already know which numbers confirm your base case, which would tip you to the high case, and which would break the forecast entirely.

This is also where guidance earns its keep a second time. You recorded the exact band before the quarter. After the quarter you compare the realised number to that band, and you note whether management’s language about the future has shifted. Over several quarters that comparison builds a picture of how much this management’s guidance is worth, which feeds straight back into how much weight you give the next number.

The point of the exercise

Forecasting revenue before earnings is not fortune-telling and it is not a single clever number. It is decomposition into drivers, honest anchoring of each driver to guidance and to what you can independently see, a range that admits its own uncertainty, and a written list of what would move it. Done this way, the forecast is useful even when it is wrong, because a wrong assumption is visible and can be corrected. A wrong number with no assumptions behind it teaches you nothing.

The quarter, when it arrives, is not a verdict on your intelligence. It is a test of specific assumptions you wrote down in advance. That is the whole reason to do the work before the print rather than react to it after.