Thoughts from the SearchPilot team

Three interesting SEO split tests 1. Do H1 changes still work? 2. Will altering internal links give you a big payoff? 3. How good is Google at crawling and indexing JavaScript? Check out the results from real-world experiments.

There has long been a tension between SEO and CRO based on trying to figure out whether what is best for users is also best for search engines. With the launch of what we call full funnel testing, we are now able to test the combined effect of a change on both conversion rate and organic search traffic as well as break the effect down into its effect on each.

Without a structured testing program, our experience shows that it’s very likely that most SEO efforts are at best taking two steps forward and one step back by routinely deploying changes that make things worse. Unfortunately, this applies even to some of the most common recommendations SEO professionals have been making for years. Is this the reason your SEO performance is lagging?

We primarily use traffic as the measure of the impact of SEO changes in order to capture changes to clickthrough rate, capture effects across the full long tail of search queries, and use the most complete data we can get access to. So organic search traffic is the best proxy for business results - and because sensible tests get more of the same traffic (by ranking better within the same keyword space), we typically see conversions increase with the increase in the organic traffic we are targeting. Essentially Google is doing its job by sending more relevant traffic.

In enterprise environments, there are some changes that everyone knows need to get done, but that are nonetheless difficult to achieve because they cut across divisions and teams. Moving from HTTP to HTTPS was such a change for many sites, requiring not only changes to infrastructure but also to image hosting, embedded JavaScript and more. This post tells the story of getting a migration done using the Meta-CMS capabilities of SearchPilot.

When Chrome 68 rolled out and highlighted any site not yet on HTTPS as insecure, it made it incredibly urgent for sites that had been putting off their migrations to get them done. One of our customers needed help to achieve this, and we used SearchPilot to remove the final blockers preventing their team getting this critical change live.

We can now 1. Move pages individually or en masse (including redirecting the old pages) 2. Move sub-domain content into sub-folders and 3. Enable easy SEO A/B testing of full page redesigns.

There is a time and place for hunches and best practice, but they are not at the top of the hierarchy of evidence in digital marketing. In this post, Sam Nemzer builds on his prior experience in clinical science to explore the different ways we can gather persuasive evidence and which methodologies are at the top of that hierarchy.

Despite the claims from Google, we regularly find that there are still downsides to relying on their JavaScript crawling and processing capabilities. This post tells the story of one of our earliest split tests comparing pages requiring JavaScript for the full experience against the performance of those pages when we removed the reliance on JavaScript.

SearchPilot enables SEO A/B testing on large and enterprise websites. For people just starting out experimenting with SEO testing, you might want to be able to play with more basic mathematical approaches. Before we moved to a neural network model we used an approach based on a modified version of causal impact. This post introduces a free version of a DIY tool that uses the causal impact approach that you can play with yourself using your own data.