Thoughts from the SearchPilot team

When a test is based on a strong hypothesis, and yet our analysis is failing to reach statistical significance, we should make a consulting-style call on whether to recommend that the customer deploys the change. In many cases, where we believe that we are improving the site, or making it more discoverable from information retrieval first principles, and where the cost is not prohibitive, we should default to deploy. The primary purpose of testing in many cases is to avoid deploying changes that have a negative impact.

SearchPilot’s mission is to prove the value of SEO for the world’s biggest websites by empowering them to make agile changes and test their impact.

Running A/B experiments for SEO, as opposed to for users, is a complicated subject. There’s a lot of confusion over things like test design, controlling for external factors and some of the tools that can help you run experiments. This is a guide that covers all of that and more along with case studies and example tests.

In my opinion, the latest machine learning models such as GPT-3 can now create useful content that is worth testing for SEO and user benefits.

Download our SEO A/B testing whitepaper to get our detailed thoughts on what exactly SEO A/B testing is, why it’s needed and how it helps as well as a year of lessons from running tests, case studies and a detailed article on how SEO testing can help avoid marginal losses.

In this detailed post, you get an annotated walk through (and access to the full video of) Dominic Woodman’s SearchLove talk that explains all the many things he learned running SearchPilot tests over the course of a year despite coming into that year as an experienced SEO professional.

So many websites have low quality copy on pages such as category pages that is entirely there for SEO purposes. We wanted to test to see what would happen if we removed this text.

When FAQ schema was released, it was highly controversial with a lot of speculation among SEO professionals about whether it was going to be a net benefit to websites or whether it would end up cannibalising clicks that they felt they should have received because searchers got the answers directly in the search results. In this post, we explore data-driven answers to these questions based on our own testing.

One of the important features of SearchPilot is taking care of the statistics and measurement automatically and we put a lot of investment into ensuring our mathematical approach is as sensitive as possible to detect the smallest possible uplifts with confidence. This post drills into an enhancement we rolled out that doubled the sensitivity, enabling us to detect uplifts half the size we could before with statistical confidence.

One of the features of SearchPilot that is most valuable to our largest customers is what we call its meta-CMS capabilities which enable us to make changes to any of the HTML or HTTP headers on any page on their website. This post explores ten common enhancements that we can roll out quickly using our platform. Does your enterprise site need any of these improvements?