Case study
Comparison table case study: a top 3 picks table lifted desktop sales per visitor 35.4%
This comparison table case study covers a desktop test on an Amazon affiliate review site. A short text list of the top five picks was replaced by a visual table of the top three. Clicks did not rise, but sales did: sales per visitor went up 35.4% over an 11-week test at a 97% confidence level.
Comparison table case study at a glance
| Client | An Amazon affiliate site in the adult niche that reviews products sold on Amazon (name kept private) |
|---|---|
| Vertical | Affiliate |
| Control | "At a Glance": the top five picks as a text list |
| Variation | "Our Top 3 Picks": a three-column table with images and a button per product |
| Devices | Desktop only |
| Test length | 11 weeks, to a 97% confidence level |
| Headline metric | Sales per visitor (clicks did not rise) |
+35.4%
Sales per visitor, desktop, 11-week test
Source: Amazon affiliate review site, desktop-only split test, 97% confidence level
Which site was tested?
The site in this comparison table case study is an Amazon affiliate site in the adult niche. It reviews products sold on Amazon and earns a commission when a reader it sends there buys. Its name is kept private. The original case study was published in January 2022.
What was the problem with the original section?
The original "At a Glance" section listed the page's top five picks as lines of text with labels such as "Our top pick", "Runner-up" and "Best bulk option". It was quick to read, but it showed no products and gave a reader little to decide with, so the decision was left to Amazon's product page.
What was the hypothesis?
The hypothesis was that a visual table of fewer products would raise visitors' buying intent by helping them decide on the page. Three columns, each with a label, an image, three short points and its own button, make the choice easier than five lines of text, and a visitor who has chosen before clicking is more likely to buy on Amazon.
What changed?
The variation replaced the five-line list with an "Our Top 3 Picks" table:
- Three products instead of five. Fewer options to weigh at the moment of choosing.
- A label on each column. "Runner up", "Our top pick" and "Best bulk option", with the top pick in the middle column, framed.
- A product image for each. So visitors could see what they were choosing.
- Three short points per product. Enough to compare at a glance.
- A "Check on Amazon" button under each product. One clear action per column.
How was it tested?
The top 3 picks table was split tested against the "At a Glance" section on desktop only. It ran for 11 weeks, until it reached a 97% confidence level, and was judged on sales data, with heatmaps to show where visitors clicked.
It was a follow-up. An earlier test of the same two designs on mobile and desktop had found that "At a Glance" did better on mobile and "Our Top 3 Picks" did better on desktop. Rather than pick one design for every device, the team retested the table where it had won, on desktop, where three columns have room.
What were the results?
Sales per visitor on desktop rose 35.4%. Overall clicks did not rise. The heatmaps showed more engagement with the top 3 table than with the "At a Glance" list.
Why did sales rise when clicks did not?
Sales rose without more clicks because the comparison table did its work before the click. Visitors saw the products, read three points on each and picked one on the page, so the people who reached Amazon were more often ready to buy. The original case study puts it the same way: visitors were better educated about the product before they clicked.
That is why an affiliate test judged on clicks alone can mislead. Had this test been called on click-through rate, the winning table would have looked like a draw. Your affiliate conversion rate has two parts, clicks and the orders that follow, and only the second pays.
What can you apply to your own affiliate site?
You can apply this comparison table case study to any affiliate page that lists recommended products: show fewer products with more help to choose, give each its own button, judge the test on sales per visitor, and test mobile and desktop separately.
- Show fewer products, better. Three products with images and a few key points can beat five lines of text.
- Label each product for a kind of buyer. Top pick, runner-up, best for bulk: the label does part of the choosing.
- Measure sales or revenue per visitor. Use a separate Amazon tracking ID for each version, so the sales data tells you which one won.
- Split mobile and desktop. The same design lost on mobile and won on desktop here. A layout that needs width may need a different design on a phone.
- Plan for the traffic you have. A desktop-only test sees part of your traffic and takes longer, here 11 weeks.
One check for AI search: an AI assistant answering a "best" question can only use your picks if it can read them. Keep the product names, labels and key points as text in the page, not baked into an image. AI agent readiness is one of the eight checks in Convertica's free CRO audit.
For a full campaign built around comparison tables on a large Amazon affiliate site, read the 10Beasts case study. For another test on a review site in the same niche, read the exit popup case study. To estimate how long a test of your own will take, use the A/B test sample size calculator.
Comparison table case study FAQ
What did this comparison table case study test?
It tested two ways of presenting an affiliate site's recommended products on desktop. The control was an At a Glance section listing the top five picks as text. The variation was an Our Top 3 Picks table with three columns, each with a label, a product image, three short points and a Check on Amazon button.
How much did sales per visitor increase?
Sales per visitor rose 35.4% on desktop, in a test that ran for 11 weeks and reached a 97% confidence level. Overall clicks did not increase. The original case study puts the gain down to visitors being better informed about the product before they clicked through to Amazon.
Why did sales rise when clicks did not?
Because the table changed who clicked, not how many. With images and a few key points for each of three products, visitors could choose before leaving the page, so more of the people who reached Amazon were ready to buy. Clicks are a weak measure of an affiliate test; judge it on sales or revenue per visitor.
Why was the comparison table tested on desktop only?
An earlier test of the same two designs on mobile and desktop found that the At a Glance list did better on mobile and the Top 3 Picks table did better on desktop. Three columns need width. So this follow-up test ran on desktop only, where the table had room.
Why did the test take 11 weeks?
A desktop-only test sees only part of a site's traffic, so it takes longer to collect enough sales to reach a reliable result. This one ran for 11 weeks to reach a 97% confidence level. You can estimate how long your own test needs with an A/B test sample size calculator.
Should every affiliate site use a top 3 picks table?
No. In this comparison table case study the same design lost on mobile and won on desktop, on the same site. Use it as a hypothesis: test it against your current layout, and test mobile and desktop separately, because the space you have changes what works.