Today, comScore is releasing an additional set of search market share metrics designed to better reflect “core” search activity, something that’s become an issue as companies such as Yahoo and Microsoft have inflated traditional metrics through the use of “slideshows” and other “contextual search” activities. Below, a Q&A with Cameron Meierhoefer, comScore’s executive vice president of analytics, about the changes.
Before the Q&A, some further background. Each month, comScore releases figures that show the number of searches that happened on the major search engines in the United States and the percentage of share each service has of the overall total. The figures first go out to financial analysts and other clients, usually mid-month. Then a public press release with the figures follows the next day.
Since March, Yahoo in particular has been gaining notable share by generating searches that many probably wouldn’t consider an an actual search. For example, someone might be viewing a photo slideshow. Clicking to advance the slide causes a page reload in a way that is counted as a “search” under comScore’s traditional metrics.
To understand this more, I recommend reading the first article below. The other two articles give important context about how even though a search engine might see its share rise, that doesn’t mean competitors have less searches, when you look at the actual number of searches happening:
- Time To End The Bull Search Engine Share Figures?
- When Losers Are Winners: How Google Can “Lose” Search Share & Yet Still Stomp Yahoo
- Bing & Yahoo Search Share Up, But Google Has Little To Fear
What are you reporting?
We will be reporting two distinct views of the Core Search market. ” Core Search,” which includes all qualified searches, is consistent with previous months that we’ve reported. Alongside this, we will also now report “Explicit Core Search,” which excludes contextually-driven searches [such as slideshows].
Why make this change?
We are changing how we report Core Search to include a breakout that captures Explicit Core Search and does not include the influence of contextually-driven searches. We’re providing this breakout to provide transparency into how the search market is changing.
We’ve provided a general definition about what qualifies as a search. A search result page must present results, provide the opportunity to refine the query, and it must be clear that the user is executing a search. This has been the basis of qSearch reporting for many years. Over time, some search engines with content networks have developed services that qualify for those rules but have a different level of search intent than is the case with a traditional search.
A few months ago, we saw a significant spike in the volume of these non-traditional searches. We called this out directly in the release notes we provide to our clients and mentioned it our press releases and blog posts.
The increasing volume resulting from the engines leveraging content experiences to provide search results led to quite a bit of confusion over what the “true” share positions were in search. Debate continued regarding what should and should not be counted, and so we’re modifying the way we report to provide additional transparency.
So now you’re reporting the true searches?
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. Moving forward, we’ll be publicly reporting market share with and without contextually-driven search.
Wall Street analysts use our data to help measure the search industry. They need a benchmark that is consistent with respect to their understanding of monetization [searches that are more likely to result in a click on a paid ad or contribute to a search engine's bottom line, in some way].
While contextually-driven searches have proven to drive volume, it’s still not clear how they contribute to the bottom line. Are these “true” searches, or not? We don’t feel it’s our decision on whether these matter [contextually-based searches versus traditional ones]. We believe it’s our obligation to measure the market, and do so in a way that lets all stakeholders make their own decisions about how to value different parts of the market.
Shouldn’t “Explicit Core Searches” really be “Core Searches?” And the core searches plus the contextually-driven searches called something else? Why not just issue one number with the contextually-based searches removed?
We’ve reviewed this with all the engines and with a number of scenarios, and in the end, transparency and granularity won out. Our experience is that the market is savvy enough to figure out the differences. While it is simpler to have a single figure, it’s better to have transparency and to be able to look at it in multiple ways rather than one ruling on what is a true search and what is not a true search.
How exactly are you defining “Explicit Core Search?”
We are defining Explicit Core Search based on the legacy search definition with an additional qualification around user intent. Explicit Core Search requires user engagement with a search service with the intent to retrieve search results.
We are applying this qualification to any new implementations, to make sure that Explicit Core Search continues to reflect what’s historically been reported in Core Search. These are searches where users are interacting with results. Where there’s an explicit, direct intention to interact with a search engine to receive results.
There are some “real” searches that core search, and even the new explicit core search don’t include, right? Such as maps?
Yes, Core Search does not include map-based searches across any of the engines.
So a search on maps.google.com or maps.yahoo.com wouldn’t be counted?
These searches are counted as a part of our expanded search reporting, but do not contribute to Core Search.
How about a case where you search on regular Google, get a map “blended” in the results, then click on the map to get even more results. You’ll count the first as a core search but not the second, right? Isn’t that odd?
Yes, we would count the first search as a search but not the map-based one that comes as a result of the blending. There’s a lot of discussion about this type of situation. Clearly both are part of search marketing. But IYP [internet yellow pages] monetization is the fundamental difference between the two. The map data is available as part of our expanded search reporting.
Yes, but that makes it hard for the public to measure. For example, you’ll break out Mapquest in your total search figures, but Google Maps gets consolidated into “Google Sites,” so you can’t tell how important the mapping site is.
Our clients have the ability to drill down more deeply. We must always balance what we provide to the marketplace for free and keeping value for our paid clients.
Beyond maps, searches on other vertical properties like news would be counted as part of core search?
Yes. Yahoo News will return results in the Yahoo News channel and get counted, despite the fact it doesn’t happen from the Yahoo home page or web SERP [search engine results page]. The same is true for Google, though most searches at Google are largely good old Google search.
YouTube isn’t counted as part of Google’s core search traffic, but there’s a huge amount of “true” searches for information that happen there. Why not fold those into the explicit core search figures?
Yes, there are a lot of searches on YouTube. But this would be a modification of how we already report Core Search, and YouTube has previously been excluded from those figures.
What about suggested searches and new navigation options. Both potentially generate new searches off an original query, raising volume but not necessarily indicating more answers are being sought or given?
While these do produce some additional searches, that’s offset by the refinement people would normally do. In other words, people traditionally tried to refine their queries on their own. These features make it easier to do what they were already doing It appears to be a bit of a wash, and the engines themselves are doing this to try to direct people to the information they want and improve the user experience.