In an attempt to reduce the bounce rates of SERP traffic, we built an a/b test. Half the traffic from SERP saw the product recommendations in the 'usual' location below the primary product sku selector. The other half saw the recommendations above, as the first element on the page.
Before the test was begun, we first ran a baseline which split the visitors into two groups while showing the same display to confirm the timeframe needed to reach statistical significance.
Once the baseline was established, the setup of the A/B test itself was straightforward: the same recommendation items would be shown, but they would be high on the page for half the visitors and lower for the other half.
As a shopper landing from an off-site search, I want to also see alternatives early, because I have not yet made up my mind
The 'B' group quickly saw a 5% drop in overall bounce rate when visiting the PDP as the first page in the visit from off-site search (Google, Bing, Yahoo, etc.) and the group maintained that difference until reaching statistical significance. Digging further into the analytics, we found that the entirety of that difference could be attributed to visitors being funneled into the alternate recommendation products.
This is a good example of a simple change having a marked effect: Quick to code, straight-forward analytics, fast results.
