What A/B testing services does an SEO company provide?
Two kinds, and they answer different questions. SEO split testing changes a group of similar pages, leaves a comparable group unchanged, and compares how the two groups perform in organic search. Conversion testing, the familiar kind of A/B test, shows visitors different versions of a page to see which one turns more of them into leads or sales. An SEO company may offer one or both. Either way, a test on a live site has to follow Google’s rules for testing, which cover cloaking, canonical links, redirects, and how long a test runs.
SEO split tests: splitting pages, not people
To have an SEO effect, a change has to be what Google crawls on a page, so a single URL cannot serve as both the test and the control. The test splits pages instead. On a site with many pages built on the same template, such as product, category, or location pages, the company changes one element on a randomly chosen half: the title pattern, the heading structure, internal links, structured data, or the amount of text. The other half stays as it is. Because both groups run through the same period, outside events such as a Google update or a seasonal swing are shared by both, and the difference between them is the measure of the change.
Split testing needs volume: a small set of pages cannot show a modest effect with any confidence, so the service fits sites with many comparable pages. On a site with a handful of key pages, a before-and-after comparison is the only option, and the company should say plainly that it is weaker evidence.
Conversion tests and Google’s rules
Google’s page on how to minimize A/B testing impact in Google Search sets the rules. On cloaking: “Don’t show one set of URLs to Googlebot, and a different set to humans.” Test variants at separate URLs should carry a rel=”canonical” link pointing to the original, which Google recommends over noindex. A test that sends users from the original URL to a variant should use a 302 (temporary) redirect, not a 301 (permanent) one. Tests controlled by cookies need care, because, as the page notes, “Googlebot generally doesn’t support cookies”, so it sees the version shown to browsers that refuse them.
The page also warns against tests that run too long: “If we discover a site running an experiment for an unnecessarily long time, we may interpret this as an attempt to deceive search engines and take action accordingly.” When a test ends, the winning version should be put in place and the test’s alternate URLs, scripts, and markup removed. Google also notes that small changes, such as the size, color, or placement of a button, “often have little or no impact on that page’s search result snippet or ranking.”
When the two kinds collide
A title that earns more clicks may not be the one that converts best, and a layout that converts well may cut text that helped the page in search. The company should assign each test to one owner, avoid running an SEO test and a conversion test on the same pages at the same time, and check search performance during conversion tests as well as conversions during SEO tests.
How a test should end
Each test should end in a decision: keep the change, revert it, or test something else. Ask the company how it splits pages, how it decides a test has run long enough and that the result is not noise, whether your site has enough comparable pages for split testing, and how it removes the test elements afterwards.