Blonde model in first product image: +3.17% revenue per visitor
Test run byDave DiederenSyntra Labs
Commentary byLiftArchiveOur analysis, not theirsSwapping the model in the first product image to a blonde model lifted revenue per visitor by 3.17% and conversion rate by 2.40%. Dave Diederen ran this three-way test, and a brunette model version did worse than the original photo. The test ran for one week, so treat the result as a promising lead, not a settled answer.
- Revenue per visitor (primary metric): +3.17% for the blonde model, $5.54 to $5.72. The dashboard estimated about $18,828 extra per month.
- Conversion rate: 9.32% on control vs 9.55% with the blonde model, a 2.40% relative lift.
- The brunette model converted at 9.01% and earned $5.43 per visitor, both below control. Picking the wrong model can cost you.
- Average order value showed no change ($59.44 control, $59.89 blonde, $60.29 brunette).
- The confidence ranges were wide and crossed zero (-7% to +12% on conversion, -10% to +18% on revenue per visitor). That makes the lift directional.
The change
Only the model in the first gallery image changed.
Simplified mockup drawn by LiftArchive from the shared screenshots.
Control
The product page opened with the store's existing first gallery image.
Variant
The first gallery image was swapped for a shot of the same product on a different model. One variant used a blonde model and a second used a brunette model.
Why it likely won
Our reading of the result, not the practitioner's claim.
- the first image is the first thing a shopper judges, so who appears in it shapes whether they picture themselves with the product.
- the blonde model may have matched this store's audience or the product's colour better. The data does not say which.
- the brunette result shows this is about fit with the audience, not simply 'a new photo beats an old one'.
Caveats: The test ran for 7 days (March 5 to 12, 2026) with 25,040 visitors across three groups, and the confidence intervals crossed zero. The product category and the photos themselves were not shared.
Who ran it
Dave Diederen (@DaveDiederen) shared this test on X. The numbers and screenshots are theirs.
View the original post on X
How to run this test on your store
- Hypothesis
- If we change the model in the first product image to one that better matches our core buyers, then revenue per visitor will rise because shoppers relate to the product faster.
- Primary metric
- Revenue per visitor
- Guardrails
- Average order value
- Add-to-cart rate
- Return rate
- Who should test it
- Fashion, beauty and accessory stores where a person wears or uses the product in the hero shot. It matters less for product-only photography or stores with little traffic, where a 2-3% lift takes a long time to confirm.
Questions
Does the model in a product photo affect sales?
In this test it did. The blonde model lifted revenue per visitor 3.17%, while a brunette model fell below the original photo, so the choice of model can push results either way.
How long should a product image test run?
This one ran for one week and the confidence ranges still crossed zero. Plan for at least two full weeks, and longer if you have less traffic.
Should I test more than one new image at once?
Testing two variants against control, as here, shows you the direction faster. Each extra variant splits traffic further, so you need more visitors to get a clear result.
Similar tests
@oliverkenyon@oliverkenyon@oliverkenyon@oliverkenyon@oliverkenyon@oliverkenyon