Does a social proof view count on the product page lift conversion?
We added a count of how many people viewed each product to the product page, tested by week and by month. Conversion fell on mobile and desktop for both versions over 14 days, by as much as 35.0% on mobile.
Lost Responsive 14 days Called at 90% Reviewed
The hypothesis
By showing how many people viewed a product, we will build trust in the product and the purchase, resulting in a higher conversion rate.
A view count is a simple form of social proof: it shows other people are interested in the same product. We tested 2 time frames. A weekly count shows recent interest and could add urgency. A monthly count shows a larger number and could add more trust.
Both counts came from real traffic data and were shown as exact figures, not rounded ones. In past tests, exact numbers have read as more believable than round ones, which can look estimated or made up.
What we changed
We added a count of how many people viewed the product, by week or month.
Both variations added 1 line of social proof to the product page, showing how many people had viewed that product. Variation A counted views over the past week. Variation B counted views over the past month. Each count came from real traffic data and was shown as an exact number. The test ran on mobile and desktop, and we measured each device separately.productpag
- Added a count of people who viewed this product
- Count covered the past week
- Exact number from traffic data, not rounded
- Added a count of people who viewed this product
- Count covered the past month
- Exact number from traffic data, not rounded
The result
Conversion fell on every device and variation, by 11.5% to 35.0%.
Mobile took the bigger hit. Conversion fell 28.7% with the weekly count and 35.0% with the monthly count. Revenue per visitor fell 34.2% with the weekly count. The monthly count's revenue change was too small to read.
On desktop, conversion fell 11.5% with the weekly count and 13.6% with the monthly count. Revenue per visitor fell 25.6% and 24.5%. Average order value moved in both directions across the 4 results, so we did not read into it.
Shane’s note
One detail I'd keep even though this test lost: we showed exact view counts, not rounded ones. A number like 19,536 reads as real, while 20,000 or 20k looks like someone estimated it or made it up. The view count hurt conversion here, but if you test social proof numbers, use the real ones.
What we took from it
The view count hurt most on mobile, where space was tightest.
On mobile, the count was 1 more thing competing for attention and it put more space between visitors and the call to action. Visitors had to stop and work out what the number meant, and a weekly count may have looked low enough to hurt confidence instead of building it. Desktop has more room, so the line was easier to take in, and the losses there were smaller.