Split Testing your website can be a very powerful tool, helping you understand your visitors, what makes them carry out key actions (adding an item to the shopping cart, signing-up, etc.) or whether there are any particularly weak points on your website.
There’s a lot of conflicting advice on the Internet concerning split testing (or A/B testing and multivariate testing). The key to successfully implementing split testing is in understanding what to test, which depends on the specific website and its audience.
Here are just a few examples:
test a major re-design
It is sometimes good to start with a whole re-design of a page or a new layout of a whole section, comparing the previous version against the newer version and measuring which performs better.
test smaller details
You might test different headlines, images or colour schemes. For example, a more persuasive headline will likely have a positive impact. Split testing proves [by collecting live test data] which element performs better, taking away potentially expensive guesswork.
Pricing might be seen as a seperate subject, but testing the same service or product at different price points might provide very surprising results. If you sell something for £200, what effect on sales would changing the price to £199 or £195 have? What about at £219? No guesswork, just test data.