Every website changes. Whole industries learn better ways to message their products, color schemes fall out of favor, and entire design schemas are replaced by smarter ones. After all, we’re only two decades removed from a time when this was the hottest thing online.
Take a quick trip to the Wayback Machine and you’ll see how some of your favorite sites have changed over the years. But a lot of those changes are due to wholesale site redesigns, new products, new marketers, slogans, etc. In other words, they’re big and they’re noticeable.
Now, most of us know that sites are actually changing all the time, often in subtle, barely noticeable ways. New products are added, new features surface, new navigation menus pop up to lead us to new site sections, and more. But sites change far often because they’re being testing and optimized. Marketers use A/B testing tools to try new messages, new images, new experiences, offers, button colors, you name it. This practice has been around for quite a while and, well, at this point, most of us are pretty used to it.
If an idea improves conversion rates, it’s generally considered successful and then implemented. But for most sites, testing means slow gains. It means waiting for weeks–if not months–and generally only finding a handful of real winners every year.
Now, if you’ll forgive us a little here: what if instead of intelligently designing your site you could, well, evolve it?
Before you keep groaning, hear us out for a second. A new approach to conversion optimization called Sentient Ascend uses something known as genetic algorithms to evolve winning site designs. It works by taking your “genes”–in this case, the ideas you want to try on your site–and seeing which ones have the more “fitness”–in this case, produce more conversions than your control. Once these algorithms find the promising genes, they start combining them into parents, breeding winners into successive generations, and even mutating, adding ideas that weren’t great in the first generation to later generations, just to make sure nothing gets missed.
One thing you’ll notice about this approach is that it involves not just a single idea–like A/B testing–but multiple ideas (or genes). That’s because testing with AI is a whole different approach to the problem. Instead of trying one or two things, marketers can try dozens or hundreds, massively expanding their search space.
In other words, when you test with AI, you’re not testing a static site. Rather, you’re evolving a site, constantly. It’s a different approach, one that’s allowed our clients to find massive gains in the amount of time a single test would normally take. Evolving sites adapt, just like in nature, to changing circumstances, like audience changes or seasonality. And, while in nature evolution can sometimes get a little weird, with Ascend, you only test what you want to test. The AI doesn’t write copy or design homepages. It does have the ability to try as many copy treatments and homepage designs as you want.