Test prioritization: pages
July 2019
Running the wrong tests can cost you months of effort and make you question your very existence. To decide which tests to run first, I suggest thinking about where they’ll run before considering what they’ll change.
So before tackling your list of ideas and hypotheses, audit your site pages for opportunities.
How many conversions come through this page?
Go where your (converting) visitors are. By testing pages that are crucial to the funnel, or consistently visited on the way to conversion, you’ll have maximum impact. (And maximum chances of getting a conclusive test result.)
This means that your checkout flow or signup page will always be a good candidate for testing. As for your home page, product pages, pricing page … it depends on the numbers.
Once you’ve noted how many monthly visitors come to each page, and how many of them go on to convert, you’re nearly there. We need just one imaginary number to complete the picture.
How much can we improve the page?
You know the existing conversion rate of each page you’re evaluating. How high can that conversion rate actually go?
I don’t know of any useful industry benchmarks for this question. (If you do, please hit Reply and share - plenty of people would like to hear about it.)
As I see it, even if you could compare your shoe store’s data with other online shoe stores, it’s questionable whether you’d obtain a valid target conversion rate. They’re selling different products to different visitors, with a different mix of traffic.
So instead, acknowledge that this is made-up math - but pick a number. If currently 80% of visitors to your Billing Info page end up converting, shoot for 85%.
Next, how much additional monthly revenue will that 5 point increase bring?
Once you’ve answered these questions for each page, you can rank them in terms of (estimated) revenue opportunity. Starting with the greatest opportunity is probably a solid plan.
How many variations can we run in a month?
One final consideration, since we already have all the data we need, is test duration. Given our page’s baseline conversion rate and traffic, as well as the improvement we hope to detect, can we even run a test? Can we run several variations?
If you’re now looking at a couple of pages that can support a test with multiple variations, where there’s a chance to increase revenue substantially, you’re off to a great start. Tomorrow we’ll look at prioritizing elements on a page.