Follows a slightly different framework, as it’s more of a quasi-experiment with bots than users. However, that doesn’t make it less useful to us. Companies like Thumbtack, Pinterest, Etsy, and others proved that when you have enough traffic. And lots of templatized pages — which ecommerce stores have — you can do SEO testing. The image below shows an SEO experiment from Thumbtack, where three different title tag variations were tested. Thumbtack put a few thousand pages into each title tag bucket, then launched the test on October 7.
On October 14 they could see that the alternate variations had underperformed the baseline title tag by 20-30%. So they reverted the titles back to the original title tag, and the traffic returned to a normal level. Graph showing data about ab Telemarketing list testing forms of testing: you take a group of pages with the same template. Change one variable and see how their traffic, CTR, or rankings diverge. Strictly speaking, SEO testing is a quasi-experiment because it lacks the randomization factor. You know exactly which pages get which treatment (changes made), as opposed to actual A/B testing, in which a user randomly
sees the control or variation page. The randomization is important because it doesn’t favor a specific user or group. It’s okay for SEO, though, because the user is a machine and therefore shouldn’t be susceptible to bias. A/B TESTING MISTAKES Many costly mistakes in A/B testing can be avoided when you know what to pay attention to. Starting out isn’t easy but it’s better than not starting with experimentation at all. Some mistakes have to be made, but I want to give you a list of tips that should prevent you from making the ones costing the most money