A fashion retailer in Northern Europe worked with Indivd to examine visitor behaviour alongside orders and revenue. The review distinguished visitors who entered from those who stayed, then compared results across the two locations, trading hours and traffic levels.

The analysis went beyond the difference between stores. During busier hours, revenue per visitor fell at both locations, while the share leaving straight away changed relatively little. This raised a separate question about what happened among visitors who stayed, rather than an increase in people turning around at the entrance.

Together, the findings pointed to two tests: a different entrance layout at one store, and additional staffing cover between 13:00 and 15:00 at both.

The stores

One question, three answers

The locations are referred to here as Store 1 and Store 2.

Store 2 generated 2.6 times the sales of Store 1. Its overall conversion rate was also higher: 14.8%, compared with 12.0%. But orders divided by visitors who stayed produced almost identical Net Conversion.

Store 2’s higher overall conversion coincided with a lower bounce rate, while Net Conversion was nearly identical. That makes early exits a priority for investigation before drawing conclusions about differences in shop-floor performance.

The analysis

Two numbers, two jobs

Conversion combines what happens at the entrance with the results among visitors who stay. Net Conversion separates out the first group, allowing orders to be considered relative to visitors who remain.

Net Conversion must still be read alongside bounce. It can rise mechanically when bounce rises, because fewer visitors remain in its denominator. The aim is to reduce bounce while maintaining Net Conversion, not to maximize Net Conversion on its own.

Visitor and sales data covering the same hours

The analysis covered 99 trading days between June and September 2026.

Six occasions were affected by network outages in the stores. Those hours were excluded from both visitor counts and sales, keeping the datasets aligned to the same measurement windows. Revenue is reported excluding VAT throughout.

See how to assess people counting accuracy before using visitor counts in store-performance comparisons.

The findings

The bounce gap persisted across the trading day and week

Store 1’s higher bounce was not confined to a particular opening hour or day of the week.

Across the hourly comparisons, bounce ranged from 41% to 49% at Store 1 and from 23% to 33% at Store 2. Across weekdays, it ranged from 37.9% to 47.1% at Store 1 and from 27.2% to 31.5% at Store 2.

The analysis found no link between higher visitor traffic and higher bounce at either store. Busier hours did not consistently send a greater share of visitors straight back out of the door.

The persistent difference made the entrance, window and first impression worth investigating. It pointed towards testing a store-level change and evaluating the result over weeks, rather than reacting to each day’s bounce figure.

Busier hours brought less revenue per visitor

Trading hours were sorted into five groups from quietest to busiest, comparing against the same hour of the day.

Revenue per visitor who entered declined across those groups at both stores.

Revenue per visitor who entered, excluding VAT. Trading hours are grouped by traffic level and compared against the same hour of the day.

At Store 1, revenue per visitor was less than half as high in the busiest group as in the quietest. Store 2 showed a smaller decline, from 199 SEK to 148 SEK.

Revenue per visitor reflects both orders per visitor and average order value. A decline can therefore result from fewer orders per visitor, a lower average order value, or both.

Bounce changed much less. Between the quietest and busiest groups, it moved from 45.4% to 43.8% at Store 1 and from 29.6% to 32.0% at Store 2.

The decline in revenue per visitor was therefore not accompanied by a comparable increase in visitors leaving straight away. The next question concerned what happened among visitors who stayed when the stores became busier.

What the number carries

Different findings point to different tests

Reading the metrics together identified three practical next steps:

• Try a different entrance layout at Store 1. The persistent bounce gap made the entrance a priority for investigation. A layout change would test whether a greater share of visitors stayed, while continuing to read bounce and Net Conversion together.

• Test additional staffing cover between 13:00 and 15:00. This period had the most visitors per staffed hour and the lowest Net Conversion at both stores. Additional cover for a few weeks would test whether changing staffing affected the results.

• Review staffing cover relative to visits throughout the day. Falling revenue per visitor during busier hours made it worth examining which hours had the least cover relative to visitor numbers, rather than looking at staffing or traffic in isolation.

These were proposed tests, not measured improvements. The relationship between busy trading conditions and performance was visible in the data, but it did not establish that staffing caused the pattern or that additional cover would reverse it.

Use one store as a comparison for the other

The stores’ weekly results moved together. The correlations were 0.87 for visitors, 0.83 for average order value and 0.75 for Net Conversion.

This supported using one store as a comparison when testing a change at the other. A movement in both stores would suggest a shared influence to investigate. A movement confined to the store making the change would direct attention towards local conditions.

The comparison would help distinguish shared patterns from location-specific changes, rather than judging a test solely by whether one store’s results rose or fell.

What the entrance cannot answer

The pilot measured the entrance, not activity within the shop floor.

It could show how many visitors entered and stayed, and place those figures alongside orders and revenue. It could not identify where inside the store visitors disengaged, because internal areas were not being measured.

Identifying that next part of the visit would require measurement inside the store.

Common questions