All Latest 650 A/B Tests

Does a "Bestseller" badge increase sales? This experiment tested exactly this by adding a visible badge on a juice cleansing product detail page.

In this experiment a juice cleanse product (developed by Dr. Ulrike Fischer) was supported with statements and visuals to reinforce her as a doctor and an authority figure. The control only referenced the product itself. Google translations include:

A) Headline: "Reset your body with a juice cleanse"

B) Headline: "The juice cleanse with a scientific basis" 

B) Listed out specific vitamins and supplements.

B) Added Subheadline: "Developed by Dr. Ulrike Fischer, who holds a doctorate in nutritional science" 

In this experiment, the variation remembered and prefilled user's size choices for the duration of the session. The control version always started with a fixed product size value (ex: XS). The variation prefilled them between products or screen refreshes. Impact on sales was measured.

In this experiment, a thanking confirmation message was appended at the top of the checkout screen of a local food delivery service. Impact on completed transactions was measured.

In this experiment, selling points and benefits of a subscription were placed as bullets at the top of a checkout page. The benefits highlighted things such as: unlimited applications, access to vetted jobs and the ability to cancel anytime. Impact on sales was measured.

In this experiment, the choice of adding a product to cart was confirmed with a further positive message in the header of the next step (on the add to cart page). Once users left the product detail page, instead of simply stating the product name, the title was rephrased as "Product [X] Makes a Great Gift. They'll Love It!". I view this as a higher "intensity" experiment, given that the add-to-cart page was in some way already confirming the choice. Impact on sales was measured.

In this experiment, a list of benefits were shown for each of the 3 plans on the last step of a contract cancelation service. Benefit items not included in the lower plans were also shown with grayed out styles (and an "x"). Clearly the higher paid plan had all the benefits listed. Impact on transactions was measured. 

In this experiment, a list of benefits were shown for each of the 3 plans on the last step of a contract cancelation service. The lowest plan only had one benefit, whereas the highest plan had 3. Impact on transactions was measured. 

In this experiment, product color swatches were replaced with real product photos. Whereas the control showed the colors as more abstract circles. Impact on sales was measured.

The GrowthRock team ran an experiment on one of their client's product detail pages. Instead of using a scrollbar (for mulitple images), clickable product photo thumbnails were used instead. Impact on sales was measured.

In this experiment, very short form testimonials (with a reference to over 3 million customers) were replaced with 3 more elaborate ones. These elaborate or benefit testimonials contained: highlighted statements, star reviews, emphasized location, tag summaries and photos of the purchased product. The control also contained a 3 testimonial carousel interaction.

This test appeared at the bottom of a longer homepage with additional product listings above.

In this experiment, a social proof popups were added to the signup / funnel pages for a contract cancellation service. The added social testimonials appeared as other customers used the service, anywhere except on the final checkout page. Additionaly, the interval at which these messaged appeared was 60 seconds, and they were shown for 10s. The messages translated to "[Name] had recently canncelled a contract with [Company], in the last 38 minutes". Impact on sales was measured.

This experiment explored a pricing layout that enabled more feature comparisons. It also conveyed more clearly which features were missing between plans. The test has been inspired by this Netflix experiment. Impact on sales was measured.

In this experiment, standard $ signs and cents were formatted into a smaller superscript. Impact on add-to-cart and sales was measured.

In this product detail page experiment, reassurances under the add-to-cart button were rearranged. The control contained copy about: free shipping and free returns formatted as two gray boxes, with a variety of payment methods and their corresponding logos underneath.

The variation used a more convention bulleted, line-by-line format. It also contained free shipping and returns, but also elaborated with "100 day returns", an "anti-hole guarantee", and "purchase with invoice" (perhaps more popular in Germany?). Impact on sales was measured.

In this experiment, a floating top navigation was shown with a "Join" button. In the control, the navigation was only visible at the top of the page. Also keep in mind that signup starts were also triggered throughout multiple CTAs throughout the page and from particular job detail pages. The a/b test ran on a listing page of Backstage - a casting call job site. Impact on signups  and checkouts was measured.

In this listing page experiment, color sets of the same product were tested against individual products with unique colors (with additional product tiles). Essentially, the A version here contained fewer product items (with color sets), while the B version contained more results and tiles (with grouped products). Impact on total sales was measured.

(The original control and variation was inverted, but was flipped to match the fewer or more results pattern).

In this experiment, a cancelation guarantee was added believing it would make users feel safer while canceling their contracts with Volders (the paid service being offered). The variation appended a Guarantee in the headline as a hyperlink with an explanatory tooltip shown on hover. This variation change was added to multiple screens throughout the checkout flow (a 5 step process).

In this experiment, hamburger navigation with product links, was removed from the shopping / add to cart page. Tertiary links (How It Works, Guarantee details and support pages) were also removed, while a "Secure Checkout" message was brought into the header. This header version was already present on the next checkout page, and was copied over one step earlier. The idea was to keep customers within the checkout flow. Impact on sales was measured.

In this experiment, a more difficult step of a contract cancelation service flow was rearranged toward a later step. In the variation, the easier step (hypothetically) with personal details and address fields was placed as the first step. Whereas the step with contract or account numbers (hypothetically more difficult) were placed as the second step.