Why comScore’s Search Engine Market Share Is So Different from Your Analytics [Data]

Why comScore’s Search Engine Market Share Is So Different from Your Analytics [Data]

UPDATE: Since this post was published, comScore has reached out to clarify their methodologies. You can read about their response and my reasons for changing this post’s title (originally: Why You Shouldn’t Trust comScore’s Numbers for Search Engine Market Share) at the bottom of this post.

It’s important for digital marketers to have accurate numbers and the real story on search engine market share. Search engine market share impacts our marketing spend, campaign objectives, and bottom line. So, why is there a discrepancy? This article will shed some important light on that.

comScore is an industry standard for big data. The company regularly releases search engine market share numbers that are often cited as the authoritative word in the search industry. Currently, comScore puts Google’s share as 67%. However, a Conductor study of more that 100 million organic search visits found that 85% of traffic came from Google versus Yahoo and Bing. In fact, many researchers in search find that their numbers of search engine share don’t line up with comScore’s.

Update: Since some have asked, both Comscore’s market share numbers and our analysis is US search traffic.

Not long ago we posted an article on the Conductor blog musing about the future of Microsoft’s Bing. In it, we cited comScore’s search engine market share numbers as proof that Microsoft was struggling to capture market share from the incumbent, Google.

comscore search engine market share

Several readers, mostly SEO practitioners (as opposed to executives), reached out to question the use of comScore’s search engine market share numbers, saying their experiences with their own sites’ analytics were that comScore’s attribution of Google’s share is low – some said *very* low. One reader said one of his sites had 82% of traffic come from Google, and another had 92%.

The Real Answer Matters to Practitioners and Executives Alike

Where the truth lies when it comes to search engine market share matters more than just knowing how to talk intelligently about it at the next industry conference mixer. It matters because the size of non-Google share determines whether resources are allocated to optimizing, tracking, and generally giving mind share to any search engine that doesn’t start with ‘G’ and end in ‘oogle’.

comScore’s numbers also create a split between executives and practitioners. Executives point to the established authority’s lower Google share and practitioners point to the considerably higher share they are seeing in their web analytics.

In the words of one seasoned SEO Manager:

“Where that [comScore’s market share numbers] causes trouble is when comScore/ qSearch numbers are known to executives, and analytics professionals come in and show their statistics and the discrepancy is observed. I’ve had multiple executives comment to me that our numbers don’t reflect “reality” – “reality” of course being the comScore/qSearch numbers.”

The reader closed by urging us to do research on search engine market share by analyzing a large set of natural search visits from a diverse collection of sites and publish the findings so the digital marketing community – both executives and practitioners alike – can benefit from the analysis and be better informed about how best to approach search market share.

The Search Engine Market Share Debate: Just a Big Misunderstanding?

Before we dive into the findings, we should mention that this post is not intended to discredit comScore. The issue is more a misapprehension than anything else. As I understand it, comScore’s methodology for calculating search engine market share involves including all entity sites in traffic calculations. That means that searches at all of the web properties owned by Microsoft or Yahoo, not just their search engines, get bundled into the results.

Marketers may expect that comScore numbers reflect only search engine market share, basically an aggregate of sites’ web analytics. They don’t realize that this is only part of what comScore considers in its calculations. This could explain the discrepancies. In that case, we’re not actually disagreeing about market share. We’re talking about two different number sets.

I am hopeful that one way or another this post validates the data – that is often cited as authoritative in the industry. Hopefully digital marketers and executives alike will find the analysis valuable, and it will settle the search market share argument once and for all.

And, if anyone at comScore wants to reach out to correct my explanation of their methodology or to clarify the discrepancies between comScore’s numbers and what many are seeing in their own analytics, you can reach me at nsafran [at] conductor [dot] com.

85% of Organic Search Visits Comes from Google

We analyzed 6 months worth of data from 63 websites – a mix of B2B and B2C. We looked at more than 100 million organic search visits, segmenting traffic into ‘Google’, ‘Bing’, ‘Yahoo’, and ‘Other’ buckets. 85% of overall organic search traffic came from Google, while 6% came from Bing and 5% from Yahoo. We saw no statistically meaningful difference in search traffic distribution for B2B vs. B2C sites. When compared to comScore’s 67% of Google share, we can see the cause for concern.

comscore search engine distribution

Nearly 3⁄4 of Websites Receive 80% or More of Search Traffic from Google

Looking at a distribution of the sites’ Google search traffic by ‘traffic buckets’ (e.g. the percentage of sites receiving 75% or above of traffic from Google vs 80% or above etc), we see that the sweet spot seems to be between 75-80%, with nearly 3⁄4’s (73%) of websites analyzed receiving 80% or more of their search traffic from Google. If you bucket all of the sites that received 85% or more traffic from Google, still, more than half (54%) of all websites received their search traffic from Google as opposed to other search engines.

Google organic search distribution

We Might Be Underestimating Google’s Search Engine Market Share

Actually, those sites with high percentages of traffic from Google could be even higher. A scatter plot of the data shows that a handful of outliers with less than 50% of Google traffic are bringing down the percentage of sites in ‘high percentage buckets’ (above). These outliers receive less than half of their traffic from Google, with the remaining half coming roughly evenly from equal parts Bing and Yahoo. These may be sites that have been penalized by Google or those who have taken a focused approach to specifically optimizing for Yahoo and Bing, but we left them in our analysis so as to accurately reflect a random sampling of web sites.

Google search traffic distribution chart

Believe in the Power of Your Own Analytics

Our analysis of more than 100 million search visits showed that 85% of total visits come from Google. Almost 3⁄4’s of sites we analyzed receive 80% or more of their search traffic from Google. When placed side by side with comScore’s published 67%, it is enough to give marketers pause.

If you’re an SEO practitioner, use this analysis to present an alternative view on search traffic distribution to executives who champion comScore’s numbers as gospel. If you’re an executive, this can be a valuable alternative data set to help you make informed decisions about resource allocation and search strategy. With this analysis, both executives and practitioners should now feel more comfortable basing their tactics and strategy decisions on their own traffic data even if those numbers reflect a Google share that is vastly greater than the industry standard comScore.


Update: comScore’s Response to This Post

The folks at comScore reached out, and we talked about their search engine market share methodologies. We discussed why their published numbers differ from our findings and from what SEOs tell us they see in their own analytics.  There are several reasons for the differences, some of which include:

  • comScore measures share, not clicks: comScore measures searches, not clicks.  They are measuring the share of searches that take place prior to the user clicking on a search result.  Our analysis was of traffic—what occurs after the user clicks on a search result.
  • comScore does not currently measure mobile:  comScore currently measures PC and laptop searches, they do not measure mobile.  Given growing mobile activity and factors such as default search engines on mobile devices, this can result in significant discrepancies.
  • Multiple clicks: The folks at comScore pointed out that a searcher could search and then click through on a result, click ‘back’, click on another result, click ‘back’ etc.  This would account for one ‘search’ when measuring by ‘search market share’ and another when measuring by ‘click market share.’
  • Searching, but not clicking: A searcher could perform multiple searches and never click-thru to a result.

All in all, I appreciate that the comScore guys reached out and pointed out that there are credible reasons why there are significant differences between comScore’s ‘share of search’ assessment and Conductor’s analysis of search share via traffic analysis.  Through no fault of their own, comScore suffers from making a nuanced argument; marketers see their numbers without hearing the differences outlined above between their methodologies and what their analytics tell them and that creates the tension between two very different sets of data I describe in this post.

I stand by the research as serving to call out why marketers need to look beyond comScore’s search engine market share numbers when making search resource and strategy decisions, but based on comScore’s clarification I am toning down the original headline to more accurately reflect the need for marketers to look beyond comScore, not that comScore should not be trusted entirely.

-Nathan Safran, Director of Research at Conductor

About Nathan Safran

Nathan is the Director of Research at Conductor and leads Conductor’s research and content team. Nathan is a monthly columnist at Search Engine Land and Search Engine Watch. Nathan’s research on digital marketing has been widely covered in both industry publications and mainstream media such as Techcrunch, Venture Beat and the Washington Post. Prior to joining Conductor, Nathan was an analyst at Forrester Research.

Related Posts

  • Pingback: The Recirculation Gap: Why Google Sends More Traffic Than Its Search Market Share Suggests()

  • http://www.seoskeptic.com/ Aaron Bradley

    It’s not uncommon to see Google’s share of search engine traffic being reported by non-comScore data aggregators at around 90%, so I don’t see what the big deal is.

    Oh wait, are you talking about *US* search engine market share? Silly me, I was talking about search engine share in the UK!

    So …. *are* you talking about the US? And, if so, did those 100 million visits you looked at come from the US?

    • Nathan Safran

      The post has been updated to reflect that we are referring to US.

      • http://www.seoskeptic.com/ Aaron Bradley

        Thanks Nathan!

  • John Brave

    On my own site (big in our domain), Google’s share of traffic is 69% and that is only because we seem to be punished somehow.

    Google lists us for the two most common keywords that refer to our site on page 11 while Bing, yahoo and DuckDuckGo list us on page 1 or 2.

    So I would say, that in normal circumstances, Google’s share is way higher than comscore’s report.

  • John Fernandez

    Obviously it’s important to note that this is US data as Aaron Bradley points out (comScore’s April 2014 numbers are clear that it is US data: http://www.comscore.com/Insights/Press_Releases/2014/5/comScore_Releases_April_2014_US_Search_Engine_Rankings).

    We’re currently seeing 93.89% of US Organic Searches coming from Google (Jan 1 – May 29th). If even “punished” folks (like John Brave) see higher than the 67.6% comScore reports.

    Thanks, Nathan, for this great post! Very helpful!

    • Nathan Safran

      Thanks John, and thanks for the inspiration.

  • http://kevinstirtz.com Kevin Stirtz

    As I understand, the comScore data is based on estimates of where people are performing their searches. Since your study is based on post-click data, is it more accurate? In other words, is your data from actual referrer log data or is it also estimated?

    Thanks!

    • Nathan Safran

      Hey Kevin,

      Our data is pulled from the analytics of the sites we studied, so it’s actual visit data, whereas, Comscore is based on searches–pre-click.

      • http://kevinstirtz.com Kevin Stirtz

        Nathan- thanks for the information. That helps paint a clearer picture of your study.