How do you forecast a store's revenue?

Forecasting revenue lets you plan purchases, staffing needs, cash flow or certain sales actions.

But a useful forecast isn't about taking last month's figure and arbitrarily adding a few percent. A revenue forecast is an estimate of future sales built from the store's own data: sales history, recent trend, seasonality and known events. It doesn't say exactly what will happen. It estimates what is likely given the information available.

Forecast and target: two different things

The distinction matters. A forecast tries to estimate what should happen based on the available data. A target is what the store wants to achieve.

Say the data points to a forecast of €45,000 for next month, while the sales target is set at €50,000. Both figures can perfectly well coexist. The gap simply shows that extra actions will probably be needed to reach the target.

Mixing the two up can lead to poor decisions. Ordering inventory based on an ambitious target rather than likely demand can, for example, increase the risk of overstock.

What data should you use to forecast revenue?

Several types of information can be used.

Sales history

Past sales are naturally the starting point. It helps to look at:

• the last few weeks to gauge recent momentum;
• the last few months to identify a trend;
• the same period last year to account for seasonality;
• several years when the available history allows it.

Relying only on the last month can be misleading if it was exceptionally good or bad.

Define the right scope

Before comparing figures, you need to check that they represent the same thing. A forecast can cover:

• revenue excluding or including VAT;
• one store or several locations;
• all channels or only in-store sales;
• a day, a week or a month;
• the whole business or a particular category.

Comparing a physical store's revenue with a figure that also includes click & collect and online sales would skew the analysis. A consistent scope is a basic requirement for a good forecast.

Account for seasonality

Many stores have periods that are naturally stronger or weaker. Holidays, vacations, sales, back-to-school, tourism or local events can change sales significantly.

You therefore need to compare periods that are truly comparable. Ideally, December should be compared with other Decembers rather than with November. A sale week should be compared with a similar period rather than an ordinary week.

Calculate a simple seasonal coefficient

One possible approach is to calculate:

Seasonal coefficient = period revenue ÷ average revenue

If a December has historically been 1.35 times the average monthly revenue, its seasonal coefficient is 1.35. That coefficient can then be used to adjust an estimate. Treat it as a historical benchmark, though, not a guarantee that next month will behave exactly the same way.

Identify the recent trend

Seasonality isn't enough. A store may have made €40,000 last June but grown steadily since. Copying last year's figure would then probably underestimate activity. Conversely, a decline that has set in over several months must be taken into account.

You therefore need to look at:

• steady growth;
• a lasting decline;
• stagnation;
• price changes;
• changes in product range;
• changes in opening hours or location;
• events that have lastingly changed activity.

The goal is to tell apart what comes from the season and what reflects a real change in the store.

Three simple methods to forecast revenue

There is no single method that suits every store.

Method 1: start from the same period last year

This is often a good base when activity is seasonal. You take the revenue of the comparable period, then apply the known adjustments. For example:

• revenue for the same month last year: €42,000;
• current trend: +6%;
• planned sales campaign: estimated impact of +4%.

The forecast becomes:

42,000 × 1.06 × 1.04 ≈ €46,300

This method is easy to explain and to adjust.

Method 2: use a moving average

For a fairly stable business, an average of the last few weeks can provide a good reference. For example, you can calculate average revenue over the last eight weeks, then adjust it for the season or known events. This method reduces the influence of an exceptionally strong or weak week.

Method 3: break revenue down

When the data is available, revenue can be broken down into several factors:

Revenue = traffic × conversion rate × average basket

This approach is particularly useful for understanding why revenue might change. An expected rise in traffic, an improvement in conversion rate or a change in the average basket can each be factored in separately. It does require sufficiently reliable data on traffic and conversion, though.

Factor in known events

A forecast shouldn't rely on history alone. Some future events are already known:

• promotion;
• sales;
• public holiday;
• school vacations;
• festival or local event;
• change in opening hours;
• product launch;
• construction work;
• exceptional closure.

When a similar event has already happened, its history can help estimate its effect. On the other hand, arbitrarily assigning “+10%” to a promotion with no reference data remains an assumption. You should therefore clearly distinguish effects measured in the past from the assumptions used for the forecast.

Build several scenarios

A single forecast can give a false impression of precision. It is often more useful to build several scenarios:

These scenarios let you prepare different decisions without pretending to know exactly what the future holds.

Measure the forecast error

A forecast only becomes truly useful when you go back afterwards and check what happened. Say:

• forecast: €46,300;
• actual revenue: €44,800.

forecast: €46,300 / actual revenue: €44,800

The gap is €1,500. You can then look into why:

• did the promotion work less well than expected?
• was traffic lower?
• did a stockout limit some sales?
• was the trend overestimated?

This step lets you gradually improve your next forecasts. Forecasting, measuring the gap, then adjusting the method is more effective than looking for a supposedly perfect formula from the start.

The limits of a forecast

A forecast remains an estimate. It cannot precisely anticipate an unpredictable event: a local incident, exceptional weather, an unexpected closure, a supplier stockout or a sudden market change.

Its reliability also depends on the quality of the data used. A very short history, atypical periods or incomplete data increase uncertainty. The further out the horizon, the more that uncertainty generally grows. A forecast should therefore be updated regularly as new information becomes available.

Which metrics should you track?

Several metrics help you interpret and improve forecasts:

• revenue;
• number of receipts;
• average basket;
• quantities sold;
• sales by product family;
• stockout rate;
• traffic and conversion rate when available.

Margin can also be tracked alongside. It isn't used directly to calculate future revenue, but it keeps you from managing the store purely on sales volume. A rise in revenue isn't necessarily an economic improvement if it relies on very costly promotions.

Forecast by location

For a multi-site business, an overall forecast can hide very different trends. One store may grow while another declines. It is therefore often better to build a forecast for each location before consolidating the results. See Multi-store management.

This approach also lets you adapt more precisely:

• inventory;
• orders;
• staffing needs;
• certain sales actions.

What AI can bring to forecasting

The value of AI isn't to “predict the future”. It can mainly make it easier to use the available data, relying for example on sales analysis and anomaly detection. A merchant can ask, for example:

“Based on my recent sales, what trend is emerging for this month?”

Then dig deeper:

“What explains this trend?”

Or:

“Which product families contribute most to the current change?”

AI can help compare several periods, identify trends or highlight items worth examining. But the quality of the analysis remains directly tied to the data it has access to and the assumptions used.

How can Gillia help?

Gillia lets you query the store's available data to analyze how activity is changing. A request can start with:

“Compare my sales for the last three months with the same period last year.”

Then continue:

“What trend emerges?”

or:

“Which product families mainly explain this change?”

When the necessary data is available, this analysis can serve as a basis for estimating upcoming periods and be updated as new sales are recorded. The point is above all to understand the factors that influence the projection, rather than receive an isolated figure with no explanation.

Forecast to decide better

A forecast is only useful if it improves a decision. It can help prepare:

• restocking ahead of a busy period (see in-store inventory management);
• supplier orders (see supplier orders);
• staffing needs;
• cash flow;
• a sales campaign;
• production, for the businesses concerned.

The goal isn't to guess exactly next month's revenue. It is to reduce uncertainty enough to make better decisions today.

With Gillia, forecasting is part of a broader approach: understand trends, anticipate changes and adapt the store's actions, tracked day to day in the store dashboard. To go further, explore Gillia's use cases: anticipate upcoming revenue.

To go further, discover the Gillia POS: every sale updates inventory, customers and the online store.