Store benchmark: expected sales from area and traffic
From a table of stores it fits the model “sales = a + b₁ × area + b₂ × traffic” by least squares, counts the expected sales of each store and the deviation of the actual from the expected. Stores below the deviation threshold are marked as underperforming; for a new place the expected sales can be obtained from its area and traffic. It shows the quality of the model (R²) and warns about a small chain and a strong link of area with traffic.
At a glance
- What question does it answer?
- How much a store should sell at its area and traffic and who underperforms.
- What data is needed?
- Area, traffic and sales per year by stores of one format.
- How to read the result?
- Look at R², then at deviations: below the threshold is a reason to look for the cause in conversion and check. Example on the default values: Intercept a: −RUB 5.92 million; Sales per m² of area b₁: RUB 13,528 per m² per year; Sales per visitor b₂: RUB 428.7 per visitor; Model quality R²: 0.979. The figures are illustrative: enter your own to see your result.
Why and how to reason
Comparing stores by revenue per meter is unfair: one has traffic, another has area. The two-driver model in the example gives RUB 13,528 per m² and RUB 428.7 per visitor at R² 0.979; the deviations show who sells less than should: store 5 by −12.5%.
When to use it
When reviewing a chain before revising plans, closing or reorganizing stores.
How to read the result
R² and the number of stores, then the deviations below the threshold; look for the cause in conversion and check.
Common mistakes
Incomparable formats, a chain of fewer than 8 stores, reading coefficients as causes.
What data is needed?
- Area, traffic and sales per year by stores of one format.
- The underperformance threshold.
Inputs
| Field | Unit | Value in the example | Meaning |
|---|---|---|---|
| Underperformance threshold | % | 10 | A store is considered underperforming if the actual is below the expected by more than this many percent. The example value is illustrative. |
| New place: area | m² | 150 | The sales area of the candidate. 0 or empty: no estimate of a new place is needed. The example value is illustrative. |
| New place: traffic per year | visitors | 200,000 | The expected number of visitors per year from a counter or a traffic estimate of the place. The example value is illustrative. Optional. Empty: the field is ignored (counted as zero in sums). |
How it is calculated
| Step | Formula |
|---|---|
| Model | sales = a + b₁ × area + b₂ × traffic; coefficients by least squares |
| Expected | expected = a + b₁ × the store area + b₂ × the store traffic |
| Deviation | actual ÷ expected − 1; underperformance if below −threshold |
| Quality | R² = 1 − the sum of squared residuals ÷ the sum of squared deviations of sales from the mean |
Worked numbers
Default values from the app; in the interactive version you enter your own.
- Intercept a
- −5.92 m ₽Sales at zero area and traffic: a service value of the model, it may be negative.
- Sales per m² of area b₁
- 13,528 ₽ per m² per year
- Sales per visitor b₂
- 428.7 ₽ per visitor
- Model quality R²
- 0.979The share of the sales spread explained by area and traffic.
- Stores below the threshold
- 1
- Sales not received by these stores
- 7.74 m ₽The sum of the “expected − actual” gaps of the stores below the threshold.
- Expected sales of the new place
- 81.8 m ₽
Data as a table
| Measure | Deviation |
|---|---|
| Store 1 | 4 |
| Store 2 | 1.9 |
| Store 3 | 13.7 |
| Store 4 | 0.9 |
| Store 5 | −12.5 |
| Store 6 | −5.2 |
Stores: actual against expected
| Store | Area | Traffic | Actual | Expected | Deviation | Verdict |
|---|---|---|---|---|---|---|
| Store 1 | 130 | 133,810 | 55.3 m ₽ | 53.2 m ₽ | 4 % | in range |
| Store 2 | 163 | 311,790 | 132.4 m ₽ | 129.9 m ₽ | 1.9 % | in range |
| Store 3 | 131 | 128,170 | 57.8 m ₽ | 50.8 m ₽ | 13.7 % | above expected |
| Store 4 | 75 | 69,720 | 25.2 m ₽ | 25 m ₽ | 0.9 % | in range |
| Store 5 | 142 | 154,000 | 54.3 m ₽ | 62 m ₽ | −12.5 % | underperforms |
| Store 6 | 126 | 191,090 | 73.7 m ₽ | 77.7 m ₽ | −5.2 % | in range |
6 stores: fewer than 8, the coefficients are unstable and one store shifts the model a lot. Use the result as a reason to ask a question, not as a verdict.
What to do with the result: for each underperforming store split the cause into conversion and average check with the “Sales: traffic, conversion, check” calculation and only then decide on staff, display or closing.
A typical mistake: comparing stores of incomparable format and age and fitting the model on a small chain. Take the region, the format (mall or street) and the store age into account by splitting the table.
Nuances and limits
- Comparing different formats.
- Fitting a chain of fewer than 8 stores.
- Reading coefficients as causes.