Revenue dropped 8% this week. That's an observation, not yet an explanation. Analyzing a store's sales means examining sales data — revenue, number of tickets, quantities, products, periods or customers — to understand what is changing and why.
A good analysis therefore goes beyond tracking a few metrics. It looks for the factors that actually explain their variation.
Along which axes should you analyze sales?
The same data can be viewed from different angles depending on the question being asked.
By day or by hour
This view lets you spot when activity is strongest or weakest and compare time slots that are truly comparable.
By week or by month
It lets you track change over time and tell a one-off variation from a longer-lasting trend.
By item or product family
This analysis shows which items or categories contribute most to revenue and which are rising or falling.
By customer
When sales are linked to identified customers, you can study their purchase frequency, average basket or contribution to revenue.
By location
In a network of stores, comparison helps you spot performance gaps and look for their causes.
By payment method
The split between card, cash or other payment methods can also be tracked when it matters for the business.
These axes can be combined. For example, you can analyze a product family in a specific location over a given period. The goal isn't to pile up filters, but to choose the ones that answer a specific question.
Which metrics should you look at?
A few metrics make a good foundation.
Revenue
It measures the total value of sales made over a period. But on its own, its evolution doesn't explain what happened.
Number of tickets
It tells you how many transactions were made. A drop in revenue can therefore come from fewer tickets, even if the amount spent per customer stays stable.
Average basket
It is calculated as follows:
Average basket = revenue ÷ number of tickets
It measures the average value of a transaction. A change in revenue can therefore be broken down into a change in the number of tickets and a change in the average basket — see increasing the average basket.
Quantities sold
They let you distinguish a change driven by volume from a change driven by price or by the composition of sales.
Sales by product and by family
This view identifies the items or categories that truly contribute to an increase or a decline.
Comparing the right periods
An isolated figure lacks context. Revenue of €20,000 can be excellent or disappointing depending on the store, the period and usual results. So you need to compare.
Several references are possible:
- the previous period;
- the same period last year;
- a target;
- a historical average;
- another comparable location.
But not every comparison is relevant. A Saturday shouldn't necessarily be compared with a Monday. A week with a public holiday isn't directly comparable to a normal week. A promotional period can also skew the reading. The quality of the comparison matters as much as the metric itself. This look at the past also serves as the basis for revenue forecasting.
Telling a one-off variation from a trend
A drop on a single day may not matter at all. The same drop repeated over several weeks deserves more attention. It's therefore useful to look at the evolution over a long enough period to distinguish:
- a one-off incident;
- a seasonal variation;
- an anomaly;
- a lasting trend.
A series of small drops, for example, can be more important to analyze than a spectacular fall on a single day. This distinction is what keeps you from reacting to every normal variation in activity — see also detecting sales anomalies to spot these signals automatically.
Looking for what explains the variation
Once a gap has been identified, the analysis must look for its cause.
The number of transactions changed
If revenue drops while the average basket stays stable, the problem may simply come from fewer tickets.
The average basket changed
With comparable footfall, a change in the average basket can strongly affect revenue. You then need to look at what customers are buying: fewer items, cheaper products or a different basket composition.
A family or a few items explain the gap
An overall variation can sometimes be concentrated on a very small number of products. Identifying those items is often more useful than analyzing the whole catalog.
A promotion skews the comparison
If the previous period included a major commercial campaign, directly comparing the two periods can give the impression of a decline when it's simply a return to a normal level.
A stockout limited sales
An item may sell less not because demand is falling, but because it was no longer available. This is an essential distinction: observed sales don't always represent the full demand.
Looking at the product mix
Two periods can show the same revenue while being very different. Imagine customers buy more low-margin products and fewer high-margin items. Revenue can stay stable while profitability changes.
Sales analysis therefore benefits from looking not only at how much the store sells, but also at what it sells. The product mix lets you see the share that different families or ranges take in total sales.
Don't confuse revenue with profitability
This is an important limit of sales analysis. An increase in revenue doesn't necessarily mean an improvement in the economic result.
A promotion can generate a lot of sales while sharply reducing the margin. Conversely, a slight drop in revenue can come with an improved product mix. When the necessary data is available, the analysis should therefore be completed with margin metrics. Revenue measures sales. On its own, it doesn't measure their profitability.
How to analyze a drop in revenue?
A simple method is to work from the general to the detail.
- Has revenue really dropped?Compare against a relevant period.
- Has the number of tickets changed?This helps identify a possible footfall or transaction effect.
- Has the average basket changed?If so, look at the composition of purchases.
- Which families contribute most to the gap?No need to analyze every item if two categories explain most of the variation.
- Which products explain the variation in those families?You then drill down to the items actually concerned.
- Is there a known cause?Promotion, stockout, price change, seasonality, a particular event or another change in activity.
This method gradually turns an overall metric into a usable explanation.
How can AI help analyze sales?
A dashboard is particularly effective for tracking metrics defined in advance. AI offers a complementary approach: querying the data directly when a question comes up.
For example:
“Compare my September sales with August.”
Then:
“Which families explain the drop in revenue?”
And next:
“In this family, which items declined the most?”
The value lies in this progression. Instead of going through several reports and filters to find the origin of a variation, the merchant can deepen the analysis question after question. The dashboard and AI therefore don't meet exactly the same need: one makes recurring tracking easier, the other supports exploration and explanation — see also store dashboard.
From analysis to action
Understanding a variation is only worthwhile if that information then helps you make a decision.
An item whose sales are accelerating sharply may call for a check on stock and restocking. A family that has been declining for several weeks may justify a more detailed analysis of the items concerned. A product that hardly sells anymore while its stock remains high can become dead stock. A decline concentrated in one customer segment can lead you to study their behavior or prepare a loyalty action.
The action depends on the cause identified. That's why analysis should come before the decision, not the other way around.
How does Gillia help with sales analysis?
Gillia lets you query the business's available data in natural language. The merchant can start with a general question:
“How have my sales changed this week?”
Then dig deeper:
“Which families explain the drop?”
and continue:
“Which products are mainly responsible for this gap?”
This approach lets you move gradually from an overall metric to the data that explains it. When an analysis needs to be repeated regularly, it can also become a routine so you don't have to redo the same check manually — the very principle of an AI agent for retail. The goal isn't to multiply statistics, but to bring out the information useful for a decision. More sales-related use cases can be found in our full catalog of Gillia use cases: read revenue by family or by item or get a chart on demand.
To go further, discover the Gillia POS: every sale updates inventory, customers and the online store.