Website Testing: How to Get the Most Mileage from Your Existing Traffic
With the digital landscape becoming more and more competitive, businesses must go above and beyond to create the most personalized and streamlined online user experience for their customers. This is why website experimentation has become a necessity for companies to innovate and stay ahead of the competition. Digital marketers, CMOs, and product managers alike are now tasked with an onerous challenge: increase online conversions and drive revenue with their existing traffic. While difficult, this task is not impossible. With CRO tools becoming increasingly more available, marketers need not look far to find the right solution to their problems. After testing begins, however, the real struggle becomes getting the most testing mileage for your current traffic volumes.
Why Traffic is so Important
In order to achieve the statistical significance necessary to determine whether a new website design is better than your current design, you’ll need a lot of traffic and conversions. This issue is further exacerbated by the effect size you want to detect — for instance, if you’re testing a small change, the effect on your conversions will probably be small, perhaps a 5% increase. In order to detect an effect of this size, you’ll need more traffic to be sure that the data you achieved isn’t simply noise masquerading as a real conversion lift. Unfortunately, traffic is expensive, and the increasing cost of PPC advertising hasn’t helped. This leaves many marketers in a tricky spot, as Eric Nalbone points out in our Optimization Podcast, “eBay can hit for contacts, but I have to hit for home runs.”
How Different Testing Tools Vary in Terms of Traffic Usage
A/B testing Tools
The most widely used CRO method is A/B testing. While you’re only splitting your traffic between two different website designs, you still need to be selective about which tests you should prioritize. If your hypothesis about what will drive conversions up is incorrect, then you won’t see an increase in conversions, and it’s back to the drawing board. In fact, only 1 out of every 7 A/B tests give significant results. Combined with the fact that you can only test one website design at a time with an A/B test, this will slow down the rate at which you can increase your conversion rate.
Thankfully, there are multivariate tests that compensate for this deficiency in A/B testing. Multivariate tests allow you to test all of your ideas at once, splitting traffic among each website design, observing the conversion rate of each, and ultimately determining which website design elicits the most conversions. There is an important detail here that complicates things. When you test every single one of your ideas, you need to split your traffic among each website. The more ideas you’re testing, the less visitors per website variant, which means that the sample size of each website design can be an order of magnitude smaller than the sample size of an A/B test. In order to achieve reliable results, you need to achieve statistical significance, and that will take a while since your traffic is split hundreds or thousands of ways.
While you are increasing your testing velocity by testing all of your ideas at once, you’re really trading off the speed at which you can test. This results in a sort of website testing catch-22: it seems like you either trade off testing speed or testing velocity. But advances in AI have shown us that it doesn’t need to be this way.
How AI-Powered Website Testing Maximizes Insights Per Visitor
Sentient Ascend is an AI-powered website testing solution that solves each of these problems. It uses principles from biological evolution to find the best website designs. You input all of your ideas into Ascend at once and Ascend will test, combine, and mutate each website variant until it finds the design that increases conversions the most. For more information about how this process works, read our white paper on Website Natural Selection.
How Does Ascend Increase Testing Velocity?
When performing A/B tests, you’ll end up with a backlog of ideas you want to test, but won’t be able to get to because you’re prioritizing other tests. With Ascend, you can input all of your ideas at once, and Ascend will use its evolutionary algorithms to combine and mutate them into winning website designs. This means no more waiting for your tests to reach statistical significance before testing — start testing them all at once.
How is Ascend Faster Than a Multivariate Test?
Ascend is faster than a multivariate test precisely because it does not need to test every single variation, rather it tests website elements in combination and then uses evolutionary algorithms to breed together and evolve the best performing candidates until they’ve reached their winning form, optimized for peak conversion rates. Better yet, during experiments, Ascend routes traffic accordingly to the better performing variants, while pruning low-performing variants automatically in real time. This ensures that you maintain as many conversions — and as much revenue — as possible while testing.
So, even with demanding constraints placed on you, there is a website testing solution that allows you to get the most mileage from your existing traffic. With this in mind, you can stop your multivariate tests, unpack your A/B testing backlog, and load up all of those ideas waiting to be tested into Ascend and watch your conversions rise.
Ready to save your marketing spend and get the most testing value out of your current traffic? Click here to get in touch!