Shop.org on Facebook
Shop.org on LinkedIn
Shop.org Twitter feedSEO implications when employing multivariate testing
A client recently asked me if there were any SEO concerns or repurcussions when using multi-variate testing systems such as Touch Clarity, Optimost, Vertster or Offermatica. That was such a good question I figure I should blog my answer.
First off, I should mention that I’m a fan of multivariate testing. It is a wonderful tool to help optimize conversion rates.
With that said, it is not a tool to optimize search rankings. In fact, quite the opposite could happen. If you look at conversion rate in isolation, the system may recommend a version that actually causes the search rankings for that page to tank! For example, if important keywords are removed because they didn’t convert well in the test.
I also like to think like a Google engineer whenever I implement something. I ask myself “What would Matt Cutts think of this?”. According to the engineers on Google’s Webspam team who I’ve spoken to, they want Googlebot to be part of the test set. In other words, they don’t want Googlebot to be excluded from ever seeing the content that real human users in the test set see — which makes sense. But given how multivariate testing is typically implemented, this is not the case. That’s because Javascript/AJAX/DHTML is used to modify the page content. Spiders don’t execute that, so that content is, in effect, hidden. The concern from Google lies in the fact that scuch an approach *could* be used for gaming and spamming.
So, I’m not wanting to discourage you from conducting tests, or from using multivariate testing vendors. Just bear in mind these things when you apply these tests beyond your PPC landing pages and onto your public website. It’s best to have all the information!
Stephan,
the one multivariate testing product that complies with the Googlebot rule is SiteSpect (www.sitespect.com). This is because it doesn’t use JavaScript … the “fully-rendered” test page is served directly to the browser. Thus, at the web site operator’s option, tests can be configured to include robot traffic.
Hope this helps,
Dave Morgan at SiteSpect, Inc.
As a multivariate testing vendor, we at Memetrics are also often asked about the repercussions of MVT on SEO. We published an article on it in our January newsletter, lift! that may shed some light. The title is “Multivariate Testing and SEO: A Little Process Integration Goes a Long Way”:
http://www.memetrics.com/marketing_resources/mvt_and_seo.html
Enjoy!
Bonnie Thomas
VP of Marketing
Memetrics
http://www.memetrics.com
It’s true any time you make a change to your website’s content it can affect your site’s ranking. But the general rule of thumb is that if your website is more relevant and improves the user experience, that tends to help your ranking in search engines. Besides would you rather be number 1 with no sales or number 5 with thousands.
Being a beta tester and partner of Google’s Website
Optimizer, I can plainly say that it’s purpose is to help site owners create content that resonates with their visitors and improves their experience. That at the end of the day is the only thing that matters. Simply having the something like Google’s Website Optimizer code itself on your website should not affect your
ranking.
Conversionally yours,
Bryan Eisenberg, Future Now, Inc.
I enjoyed reading that article, Bonnie. I agree with the article’s main premise that SEO and multivariate testing can co-exist and the search engine friendly state of the pages can be maintained. However, I see search engine FRIENDLY and search engine OPTIMIZED as two very different things. With multivariate testing, you can minimize the damage and keep things search engine FRIENDLY. But you lose the search engine OPTIMAL state. That’s because multivariate platforms do not make API calls to keyworddiscovery.com to figure out which titles, headlines, body copy are best to use from an SEO standpoint, nor are things such as keyword prominence (or its cousin of much lesser importance: keyword density) taken into account.
Thanks for the comment, Bryan! I don’t believe it has to be one or the other. I think you can have that ranking AND the conversion. It just requires that you don’t optimize for conversion in isolation. For example, you could ONLY load in headline candidates for testing that have gone through keyword research and SEO first. Or you could tweak the winning headline for SEO after the fact. Either way, it needs to figure in to the equation.
Simply going for what is more relevant and improves the user experience won’t necessarily equate to higher rankings. Case in point: Freshpair.com, who is completely dominating in the SERPs by cleverly weilding anchor text, link buying, internal linking, and so forth. From what I can tell, the tactics they employ are not being driven by relevance and user experience. But boy are they winning! I elaborate more on this in my column in the current issue of Practical Ecommerce: SEO Report Card: Freshpair.com.
Regarding Google’s Website Optimizer, its purpose is consistent with Google’s company line of “Build websites for users, not for search engines”. And I agree with that — up to a point. In Google’s ideal world, SEO would not exist and manipulating their ranking algo would be taboo. But you can’t show up at a gunfight with just a sword either. Google’s Website Optimizer helps with one piece of the equation but neglects another piece which IMHO is just as important. I want the highest conversions possible AND the highest rankings. (while staying within Google guidelines all the while, of course!)
Stephan, thank you for a great entry to your blog.
As a disclosure, I work for Kefta, a leading personalization and targeting service. We utilize sophisticated MVT and logistic regression algorithms, but typically only within relatively homogeneous, targeted, visitor groups.
Having said that, we published findings on a similar subject last June in conjunction with Brad Fallon and his team at SmartMarketing / StomperNet. For his MyWeddingFavors site, he was able to produce a 65% improvement in conversion for organic search traffic. If anyone finds this interesting, the results can be found at… http://www.kefta.com/press/releases/20060620_mwf.html.
Another consideration regarding SEO impact is to determine what aspects of a page you’re going to test. For example, changing body text of your page brings up greater concern than changing a key banner image. These types of decisions, in conjunction with objectives of the experiment, should be evaluated prior to launching any campaign on a page that has significant SEO value.
Thank you,
Mark Ogne
VP Marketing Services
Kefta, Inc.
http://www.kefta.com
Hello all,
I am the CMO for Touch Clarity and we are mentioned in the original post as a MVT vendor. While this confusion is understandable, we are actually in a different category, one we call On-site Behavioral Targeting, which has some important differences that I think need to be clarified.
While it’s true that our targeting platform can be used to function in a similar manner to MVT approaches to determine the best page layout or top creative from n number of versions in a highly automated and very cost effective manner, On-site Behavioral Targeting is really about maximizing the engagement, conversion and revenue by predicting and targeting the optimal content to each individual visitor based on analyzing that individual’s anonymous behavior on the site during the current visit and in previous visits.
Now to the question at hand: we carefully analyze the customers SEO rankings before and after we are implemented to prove to our clients that there isn’t any negative impact. Our customers place default content into each slot that we target and because the spiders will read the alt text for that content slot, there isn’t a difference in ranking before or after Touch Clarity Targeting is in place. And we constantly monitor this ongoing for clients.
One very positive impact we see with our customer implementations is significant uplift on their SEO-driven conversions, since we use the highly predictive variables passed in the referring data from the search engines, such as keywords entered, and that information is leveraged, among other variables, to target the optimal content on the entry pages for organic searches, which as we know is harder to control. This in effect, makes it more aligned with the best practice of PPC dedicated landing pages but in an automated manner.
If anyone has specific questions, you can reach me at brent.hieggelke@touchclarity .com.
Brent Hieggelke
Chief Marketing Officer
Touch Clarity
Thanks for clarifying Touch Clarity’s place in the MVT world, Brent. Your company was actually one of those mentioned by name by our client (a Fortune 100 company) when they asked the question of me regarding potential SEO impact of MVT platforms.
Good stuff guys. This has been a lively discussion and I’ve really enjoyed reading everyone’s comments, as I’m sure other Shop.org members have too.