Multivariate testing platforms and SEO compatibility

A client recently asked me if there were any SEO concerns around websites managing their content via automated content “enhancement” or “testing” systems such as Touch Clarity, Optimost, Vertster or Offermatica.

I am not aware of any issues with search engines misconstruing this as spam. I had a short conversation with the CEO of Offermatica when he presented at Search Engine Strategies last year and he said that Google is okay with what they are doing. There are too many major AdWords customers using multivariate testing for Google not to be careful not to penalize sites for this.

Offermatica use JavaScript to overlay and in effect replace the existing content for browsers with JavaScript enabled, so the spiders (who don’t execute JavaScript) don’t execute the Offermatica scripts and the underlying default HTML is what is “viewed” by the spiders.

As one blogger describes it, Offermatica’s solution produces a “control” copy for the spiders and through JavaScript (mboxes) display different pages to end-users. The blogger raises the concern that It potentially violates Google (see 1st bullet in “Quality Guidelines – Basic principles” section: and Yahoo!’s (see 2nd bullet in “What Yahoo! Considers Unwanted” section: search quality guidelines.

Nonetheless, as I stated above, I think the engines are careful not to penalize sites running Offermatica, Optimost etc.

The problem I have with conversion optimization services like Offermatica and Optimost is less that they could be misconstrued as spam, but rather the fact that they don’t take the potential SEO impact into account. So, for example, a conversion optimization test of the home page might show a clear winner as far as the the best converting variation. However, when that variation is then implemented as the new permanent home page, the rankings and search traffic may tank. The likelihood for unforseen consequences is great because Offermatica and Optimost do not understand SEO deeply at all.

Scott Miller, CEO of multivariate testing vendor Vertster whom I met at SES explained to me that they use an AJAX/DHTML approach to modifying the page content, done in a way that they claim is imperceptible to the search spiders. They identify areas on the page with specific id attributes, and then modify the content shown to users. This would only cause problems if it was used for “gaming” the search engines, and you get caught. They do not allow this usage of the tool, and will delete any users who may use it for this.

Sounds like a similar approach to Offermatica.

Anyways, a few considerations to take into account when implementing a multivariate testing platform…

About Stephan Spencer

One Comment

  • cell jacela says:

    Very interesting point indeed. Still amiss on how google would treat the different pages served by A/B or MVT tools. If one runs a 5 different elements in 2 recipes which one would google consider as “valid” during the duration of the test run, say you have a sampling size of 30 days. Between those period Offermatica will be serving somewhere between (5x5x5x5x5) versions of a page using one recipe.

Leave a Reply

Your email address will not be published.