Sign up for weekly recaps of the ever-changing search marketing landscape.
Google Search Share Up After Instant, Yahoo Down After Bing Transition
comScore has released its latest search share figures to analysts. UBS is first out of the gate to report them. In the wake of Google Instant, Google is once again gaining share, after several months of slight decline. Meanwhile after months of gains, Yahoo drops after becoming “powered” by Bing.
To be clear, UBS itself doesn’t attribute the gains or losses to Google Instant or the Yahoo transition to Bing. Instead, it’s the return of university students that UBS finds the most likely culprit for changes. It could be coincidence. Other factors might also be involved. But these were two huge changes that perhaps also contributed.
September 2010 Search Share
Here’s the share of searches that each search engine is estimated to have handled in the United States during September, followed by its share in the previous month of August:
- Google: 66.1% (up 0.7 from 65.4%)
- Yahoo: 16.7% (down 0.7 from 17.4%
- Bing: 11.2% (up 0.1 from 11.1%)
- Ask: 3.7% (up 0.1 from 3.8%)
- AOL: 2.3% (no change)
Back To School Boost For Google?
Analysts Brian Pitz and Brian Fitzgerald find Google’s gain due to university students, who presumably use Google more than other search engines when they are back in class. From their research note:
We attribute the impressive gains in part to university students returning to campus. Google remains a top pick as we approach 3Q earnings, which are set to be released Thursday.
I’m not sure that I’d entirely agree. Google is certainly well known for a summer slump when school is out. However, Google had been showing slight drops even before the summer break. But, I’ve not had a chance to dig into comScore’s historic figures for explicit core search yet, so I can’t say for certain. I’ll try to do that in the near future.
What’s A Search?
It’s been unclear how Google Instant might change search share metrics, since it launched in early September. The new Google Instant feature causes searchers to often see several sets of results while they are typing their query, without the page needing to reload or the the URL in the address bar changing. That’s traditionally how a “search” has been counted, by estimating how many search-specific URLs are shown to users.
comScore is apparently not counting any of the searches that flash by as people type but rather only the final set they see. Says UBS:
To account for Google Instant, comScore has again changed methodology. “Explicit core search” essentially backs out Instant and only records a search when the user clicks a link or presses the enter button. “Core search” also records a search when users pause for 3 seconds to contemplate suggested Instant results.
If all this talk about “Core Search” and “Explicit Core Search” and new methodology isn’t familiar to you, see our Q&A: comScore’s New “Core Search” & “Explicit Core Search” Figures article from August that explains all about this change. In a nutshell, Explicit Core Search excludes some searches that many wouldn’t consider to be actual searches, such as someone clicking through a slideshow.
“Core Search” Over Time
The report, which we’re citing with permission, also included a chart looking at search share over time. This chart is for core search — not explicit core search — so the figures won’t match those above and will include some of those slideshow searches I’ve mentioned above:
What’s interesting is that even with the more inclusive core search, Yahoo still sees a drop. Slideshows and other “contextual” searches, as Yahoo sometimes calls them, had been giving Yahoo a boost in recent months. That came to an end in September. That was also the first full month where Yahoo stopped using its own search technology, shifting to have most of its main results being powered by Microsoft’s Bing at the end of August.
comScore’s figures get released to the public usually a day or two after they go to analysts. Watch for them here.
Yahoo basically calls bull on comScore’s methodology here, claiming that it effectively overcounts searches due to Google Instant. I agree, there’s potential for this there. But in the example given, unless you paused for three entire seconds before typing “Sp…” before continuing on to finish “Sprinkles Cupcakes,” comScore wouldn’t count the other search results that flash in front of you. And three seconds is a really, really long time when it comes to entering a search query.
For related posts, also see Techmeme.