\n Stay Forward <\/h2>\n
\n Get exclusive insights into digital
media's top-trending topics delivered
directly to your inbox. <\/p>\n <\/div>\n\n
\n
\n
I’m a self-admitted testing geek.<\/p>\n
In a\u00a0straight A\/B test, you pick one variable to test, and keep everything else the same. Lots of people make two very common mistakes with this: Either they try to test two variations that are so different that it’s impossible to know what caused the change in performance, or they test something so minor that it’s unlikely to have much influence on performance at all.<\/p>\n
Want to test more than one thing? That’s where multi-variate testing comes in. But be careful: Multi-variate testing has a lot of drawbacks. It’s more expensive, for one thing, and it’s harder to see significant results with such small segments. Most businesses are best served by keeping the number of test segments as small as possible.<\/p>\n
To determine what to test, it helps to start with figuring out what’s the “most broken” — or, to put it in other words, where the biggest area of opportunity lies. For example, many e-commerce companies find that the biggest pain point for them is the shopping cart checkout process. Testing a variation designed to improve cart checkout rates might be a great use of a testing budget for such a company. Testing costs dollars; the trick is to put the dollars where they’re likely to make the most difference.<\/p>\n
It may sound obvious, but the first thing to check is whether you’re able to measure the results of your test. Ensuring that all your tracking tools are in place and properly set up to measure conversion. If you’re checking to see whether changing the colours on your page will result in more phone calls, it first makes sense to check whether you can tell which calls come from which page. If you can’t measure it, you can’t test it.<\/p>\n
Statistics<\/a> aren’t just for math nerds. Before you run that test, you need to figure out if you will get a statistically significant result out of it.\u00a0 In other words, will you see a big enough difference between your test and control page to be able to say that you’ve learned something?<\/p>\n There are two main variables to consider here:<\/p>\n There are a number of free calculators (like this one<\/a>) that can help you do the math on this one.<\/p>\n \n Get exclusive insights into digital \n
\n \n \n Stay Forward <\/h2>\n
media's top-trending topics delivered
directly to your inbox. <\/p>\n <\/div>\n\n \n \n