Eying Search Engine Market Share in the era of Bing

Share

At the end of May Microsoft announced its new search engine, Bing. As data from Bing’s first full month becomes available, I thought it would be interesting to take a quick look at the current market share enjoyed by the major search engines in the US and a “typical” European market, Italy. The real test of Bing’s success will to be to check back in a few months to see if Bing has picked up traction with users or not. As the folks from Cuil can attest, a burst of publicity doesn’t necessary translate into loyal search users.

Search Engine statistics, USA vs. Italy

Most web intelligence services are currently US centric with very little worldwide reach. Unless stated otherwise, the data which follows is for the US market. Where available, I’ve also provided data for the Italian market, which for search engine usage is rather typical of most west European markets.

The first Bing statistics: Google’s monopoly in jeopardy?

There are several companies offering competitive benchmarking data for Internet marketing research. Due to differing data collection and sampling methodologies, it is difficult to directly compare data from one service to that from another. What you can do, however, is see if, taken together, any distinct patterns emerge. The data that follows will show that regardless of who is doing the measuring and how they do it, Google has an enviable lead in the US market and what appears to be an untouchable 90% or so in our representative European market, Italy.

Demand scientific rigor if you buy competitive intelligence data

Most of the web statistics suppliers cited here are in the business of selling competitive intelligence data. Note that some may lack scientific rigor behind their methodology. Others may have rigor, but we cannot independently evaluate this as they don’t disclose their methodology in any degree of detail. Do explore this issue further before committing to any one supplier. If your budget permits, you might want to consider using multiple suppliers. See our background article for further information on web benchmarking services. The availability of good actionable Web Analytics data is what makes web marketing much more accountable, and thus profitable if done right, than traditional marketing. Like any tool, web analytics data needs to be selected and used appropriately.

Multiple dimensions to measure search market share

In the case of search engines, there are multiple dimensions to potentially measure: unique visitorsto a search engine, visits (presumably a visitor will return during the reporting time period), queries (a user may make multiple queries during a visit), search referrals (successful queries which result in a user choosing a query result to visit another site) and search session time.

Issues which may skew results

Measuring any search dimension is easier said than done. It is probably safe to assume that the search data provided by professional analysts takes the issues that follow into account – one of the advantages companies enjoy when paying professional business intelligence data firms to provide custom data sets.

Search and non-search activity on the same domain

Both search and non-search activity often occurs on the same domain, such as google.com, which hosts non search related services. In the case of Google, we have Google Analytics and Google Reader, to cite two. Both are sources of traffic on Google.com and both can provide referral traffic to a website.

Measurement services should filter for true search queries, those containing a keyword parameter, otherwise search data may be inflated. Many self-service tools don’t allow for proper filtering.

Country specific vs. worldwide data

The most accurate, albeit imperfect, way to track the origin of web traffic is based on the geo-location of a user’s IP. In the case of Bing, it is the only way. Google uses specific country level domains for its local variants, e.g. Google.it. Yahoo differentiates country specific search using subdomains of search.yahoo.com, e.g. it.search.yahoo.com, except for the USA where it offers both us.search.yahoo.com and search.yahoo.com as options.

The Ask-Google arbitrage game

Some search engines like Ask.com advertise on other search engines, particularly Google, diverting users from one engine to another. This rather questionable tactic by Ask is called AdWords arbitrage. Such activity may artificially inflate Ask’s real market share.

Users don’t use one engine exclusively

Many of the users who use Google, are also users of Bing and Yahoo Search. Some of them may be just typing “Google” in the Yahoo! or Bing search box in order to navigate from a Yahoo or Microsoft service, like e-mail, to Google, but these users are using Yahoo! or Bing search non-the-less. You need to understand the context of any given statistics in order to properly interpret them.

More is not always better

One common Web Analytics trap to be aware of is that more is not always better. A greater query volume per user may be a better indicator of poor search results and persistent users rather than of search engine success. Again, high level statistics will only tell part of a more complex story.

Alexa

Probably one of the best known sources of competitive web intelligence data among layman, Alexa is often cited despite an unfortunate lack of scientific rigor behind Alexa’s methodology. To compensate, Alexa does have a very pretty site comparison widget, although this functionality seems to be broken at the moment if the widget isn’t hosted on Alexa’s domain. To the best of my knowledge, Alexa does not publish search specific data nor can it be retrieved from their self-service tools.

Illustration 1: Visits to the domains google.com, yahoo.com, bing.com and ask.com from world-wide sources in 2009.

Keep in mind that Alexa tracking will include any non-search traffic to a domain, from any world-wide source, thus the above is NOT an accurate representation of search engine usage.

Compete

Compete offers a free self-service dashboard which allows you to compare websites by domain, or if logged in to a free account, by subdomain.

Illustration 2: Visits to the domains google.com, search.yahoo.com, bing.com and ask.com

In the specific example of search, we have over-reporting for Google as we cannot filter out non-search services. Compete did provide search specific data for June 2009 in their blog:

Compete credits Google with 74% of US search query volume as divided by the top 5 players, Yahoo at a distant second with 17% and Bing at 7%. Note that Compete also provide custom reporting on any URL element as one of their paid services.

comScore

comScore cites Google with 65% of the US search market as divided by the top 5 players, about 10% less than Compete’s reported value. Little consultation to Yahoo! at 20% and Bing at 8%, according to comScore data.

Google Trends

Google Trends is another worthwhile component of a competitive web intelligence toolkit. In addition to keyword trends, Google offers some insight into website traffic volumes. Unfortunately, Google isn’t clear as to where this data comes from, but they have the Google Toolbar, search referrals from the Google Search engine and/or Google Analytics direct measurement and referral traffic at their disposal. Given Google’s web reach, their data has the potential to be as accurate as anyone else’s, at least for the markets where Google is dominant. Google trends only reports on primary domain names, so it isn’t possible to discriminate between search and non-search traffic.

For Bing, Google is currently showing just shy of 1.4 million unique visitors per day (you only see numbers if you are logged in). Google Trends doesn’t yet provide visit nor monthly data.

Unfortunately, it seems that what is good for the gander isn’t good for the goose. Google displays the message “google.com has no data available for ranking” should you try to include Google in a bench marking list, needlessly opening Google to charges to hypocrisy. Still, Google Trends is a remarkably good tool.

Hitwise

*Includes executed searches on Bing.com, Live.com and MSN Search but does not include searches on Club.Live.com.

The above data is courtesy of a Hitwise press release; Hitwise provides rolling 4week reports in its online data center¹. Hitwise also offers data for many of the other English speaking countries where they operate but, alas, Hitwise is not yet present in Italy.

Nielsen Online

In the above chart, I’ve calculated the number of visits to each of the major US search engines based on data supplied by Nielsen regarding unique visitors and number of visits per visitor. While Google may only have about 66% of all search visits, according to Nielsen Google attracted 83% of all visitors who search (Yahoo! counts 40%, Bing 25%, Ask 15% and AOL 13%). As is probably clear, many search visitors will use more than one search engine over the course of the month. Nielsen reports its US panel size at 28,000 although this will soon increase to more than 200,000²

*Historically Google has also provided search results to the number two search site, Telecom Italia’s Virgilio portal³

Unlike many suppliers of web statistics for competitive benchmarking, Nielsen Online has a truly worldwide presence. In Italy, their panel size is about 20,000⁴. Similar to other markets in Western Europe, Google has a clear monopoly on internet search, hovering around 90% of search visits.

Quantcast

Quantcast is an interesting source of data particularly for sites which allow Quantcast to directly monitor their traffic. Quantcast’s current methodology description does not clarify the source of Quantcast data for other sites. In the case of search engines, it could be browser centric, e.g. toolbars, and/or network centric, e.g. from ISPs:

These data sets are sourced from multiple parties including market research companies, ISPs (broadband and dial-up) and toolbar vendors and cover in excess of 2 million individuals (1.5 million in the US). Quantcast also provides support for Quantified Publishers to share non-PII data via the direct measurement solution and this provides cookie-level data on many tens of millions of individuals.

Keep in mind the above chart includes both search and non-search activity.

StatCounter

StatCounter is a browser centric Web Analytics service which currently tracks more than 3 million websites across the world, for a total of about 4 billion page views per month. StatCounter provides some aggregate data to the public through their Global Stats dashboard as a way to publicize their services. Due to the nature of the StatCounter service, StatCounter measures actual click-trough traffic from search engines to websites, i.e. just successful search queries.

StatCounter USA data is based on 336 million search engine referring clicks during a 3 month time frame⁵.

Global reach is one thing that differentiates StatCounter from many other sources of competitive web intelligence data. Do beware the Italian data sample size is not disclosed, an unfortunate limitation.

One more statistics source to come

I’ve identified one more source of world-wide data and hope to include their statistics in this space within the next few days. Check back for an update.

And the winner is… Google

For many readers, the end result is probably not too much of a surprise. Google has a very comfortable 65-79% of the US market and somewhere between 88-96% of the Italian market. In a related article, we take a look at Bing and what it means for SEO. For now, most European web marketers should keep an eye on Bing developments while US Web Marketers should consider incorporating Bing nuances into their SEO strategy.

Similar Posts:

• Are we still lying with statistics in the internet age?
• Yahoo Web Analytics (ex IndexTools) soon in no man’s land?
• Google Zeitgeist has a Sibling, Google Trends
• And then there were two: Microsoft – Yahoo! Search Deal at SMX West 2010
• In-house SEO: SEO Reporting & Scorecarding for Management

If you haven't already, you might subscribe to my feed by Email, RSS feed and/or follow me on Twitter, which is updated on a more frequent – and more meaningless – basis. Finally, if you're a Sphinn user, Sphinn love is welcome :-). Thanks for visiting!

Share