Most marketers are doing A/B testing, whilst Multivariate testing and Bandit testing, are less used.
A/B testing involves testing two versions of a page. There is a challenger and a control (usually the original page if you’re testing a different version against an existing page). Visitors are randomly assigned a version. Tests usually continue until a page (or version) is statistically (and significantly) better than the other.
Multivariate testing (MVT) is very similar to A/B testing but involves testing more than two versions
Bandit is similar to MVT except, increased traffic is sent to the most successful option, rather than sending equal numbers of visitors to each version. This allows you to increase conversions in line with the most successful methods during the test, rather than waiting until the end of the test to shift traffic to the most successful page. It also means that if suddenly another page starts performing better, you can suddenly shift traffic to that one. With bandit testing, traffic may also be sent to more than two versions like MVT.
The diagram below by Crazyegg shows the difference between Bandit and AB testing.
It depends on what your testing and for how long.
A/B testing is the most popular method as shown above, it’s also very easy to setup and manage.
Bandit testing is best off used instead of MVT, as it’s similar to MVT but with the added protection that whichever version is winning and most successful, you can direct traffic to this version and hopefully increase conversions quicker.
According to the Conversion Rate Optimization (CRO) report by ConversionXL, most marketers follow a Conversion rate optimization process
However, just under 4 in 10 (39%) say it isn’t structured or documented, but 36% say it is.
Most optimization, according to a CRO report by ConversionXL, also reveals that most CRO is conducted by one person in the company.