I’m going to detail an interesting turn of events for one of my personal sites. I’m hoping it sheds some light on how Yahoo! influences their search traffic numbers.
While looking through Google Analytics I noticed a nice little peak in search traffic.
Hmm, nice... but why so temporary?
Also, I’d love to say that the general “up and to the right-edness” of the graph is a direct result of my super-awesome web marketing ability, but seasonality is a big part of it as well.
Outside of the organic traffic report this sort of thing is commonplace. That could easily be a single page that got hot on Twitter or Stumbleupon, referral traffic from a popular blog, or the day of an email campaign. Within the organic traffic report, this definitely requires some investigation.
After segmenting down a bit I uncovered that this traffic was coming from Yahoo!, and was all from one keyword which my site has had the first position on lock for longer than I can recall. This graph tells the story:
Knowing that my site had always ranked first for that keyword, I had to wonder. What causes the search volume for one random keyword to spike dramatically in only one search engine? As you may have been able to guess, there was certainly something going on on one of the Yahoo! web properties that was influencing users to search for this keyword. A quick Google search for ‘site:yahoo.com “keyword”’ and I had my answer (this is not THE answer, just an example of the answer):
From the center slider of the Yahoo.com homepage. Culprit highlighted.
Also, don’t you think it’s a little rude to be calling the Final Four women ugly, Yahoo!? ;)
At this very moment there are 32 slides, 5 of which include a link directly to a search engine result page. The image above leads to a Yahoo! search for “bald eagles”.
Warning: Some Wild Math
By my estimation this slider sent roughly 1,700 visits to my site during the 48 hours it was on the homepage. Considering users aren’t actually searching for this keyword, let’s assume 20% of users clicked the first result. That means around 8,500 “searches” were done for this keyword as a result of the slider. Multiplied times the 4 other click-searches from the slider, then dividing by 2 for one day’s worth of traffic, Yahoo! is creating search volume of 21,250 from the slider per day. Couple this with the tons of other click-searches throughout Yahoo!’s properties and it might become more clear how they’ve been able to maintain 16% of the market share.
Highlighted are click-searches from the homepage. If that 1 slider link, which is displayed for a total of 10 seconds before it has to wait another 310 seconds to show up again, leads to almost 10,000 searches per day, you can bet the static trending topics list sends an exponentially greater number.
But Does It Count?
Back in August 2010, ComScore began reporting “Explicit Core Search” numbers alongside their traditional search engine market share numbers. This was done to counteract the attempts to game the numbers by Yahoo! and Bing. “Explicit Core Search” is defined as “user engagement with a search service with the intent to retrieve search results”. Danny Sullivan did a Q&A on Search Engine Land with Cameron Meierhoefer, comScore’s executive vice president of analytics, about the changes:
Danny: So now you’re reporting the true searches?
Cameron: We are not redefining what counts in Core Search. Instead, we’re classifying contextually-driven search as a different type of search and allowing the end users of our data to decide if or when to include them in their analysis.
The best example of Yahoo!’s contextually-driven searches are their slideshows (example). Notice how the block of “Related Search Results” refreshes on each new photo. In the past, each photo swap was a new “search”, but is no longer included in the “Explicit Core Search” figures. Good for ComScore, this is clearly not something that should be figured into search market share numbers.
What we’re seeing in my site’s case is, I believe, a different story. I’m not entirely sure that ComScore would be capable of filtering these out considering the destination URL for these click-searches isn’t too different from the traditional “core” SERP URL. Of course, the click-search URL contains tracking parameters that ComScore could detect and weed out, but how hard would it be for Yahoo! to get around (if they’re not already)?. I’ll let my favorite book title + cover combo of all time tell you how I feel about my chances of being right:
I don’t think it’s a stretch to say that ComScore’s numbers are still pretty tainted if those 5 little temporal slider links can create almost 1 million searches per month. Considering the breadth of Yahoo!’s web properties, I think we can assume Yahoo!’s “Explicit Core Search” market share is even lower than the 16% reported by ComScore. For what it’s worth, here’s a breakdown of organic search traffic in 2011 to the site we’ve been talking about:
Oh, I suppose there is one SEO action item we can attach to this post: Make friends with an editor at Yahoo! :)