I have a hard time with all the measurements I get from different Market Research companies. Data from Nielson NetRatings appears to be measuring audience while HitWise appears to be looking at total search volume (in the US mainly). I think there’s a confusion with several sets of numbers - which one to use ….that’s the question.
Case in point - Nielsen NetRatings releases, on a weekly basis, the top ten search engines by Unique Visitors and Average Time spent on the search engine per visit. Nielsen NetRatings numbers are often published in IMediaConnection, as they were today.
|
Top 10 Online Search Engines/Portals & Communities Destinations |
||||||||||||||||||||||||||||||||||||||||||||
|
||||||||||||||||||||||||||||||||||||||||||||
| Source: Nielsen//NetRatings |
Meanwhile, a couple of weeks earlier, Bill Tancer released top search numbers for the week ending July 29th (two weeks earlier than the Nielsen numbers)
Below is a table showing the top three engines Google, Yahoo! Search and MSN Search for the four weeks ending July 29, 2006.
It would be great if we could get both sets of numbers (which I bet don’t measure exactly the same thing - but look like they do) to be for the same exact time period. According to Bill Tancer:
"As you can see, the top engines combined account for over 94% of all search volume in the US. Since we last released these numbers in June 2006, Google (www.google.com) has increased its share from 59.3% to 60.2%, Yahoo! Search (search.yahoo.com) has increased from 22.0% to 22.5% and MSN Search (search.msn.com) has decreased slightly from 12.1% to 11.8%."
Meanwhile, Danny Sullivan shows us the top search share numbers for July (similar to what Hitwise is showing) - but based on Nielsen Netratings data:
Share Of Searches: July 2006
The chart below shows the percentage of online searches done by US home and work web surfers in July 2006 that were performed at a particular search engine. Internal site searches, such as those to find material within a particular web site, are not counted in these totals. The activity at more than 60 search sites makes up the total search volume upon which percentages are based — 5.6 billion searches in this month.
When we look at Search Volume the charts give the impression that only the top 3 or 4 search engines matter while if we look at the top destinations by audience (first chart) FaceBook is the 10th destination (quite significant as Microsoft just formed a 3 year advertising deal with Facebook, as I reported earlier last week).
Which set of charts do you think Microsoft looked at when they decided to approach Facebook? One of the nice things about the Nielson data is it gives the average time spent on a search site and Facebook, while it had a lower volume of unique searchers - each spent at least 30 minutes (on average) at the site. If one wanted to do an ad distribution deal it might be good to pick one where visitors tend to hang around longer. Also, Microsoft could not have done a Advertising deal with the first 7 search destinations on the list - most already being in exclusive partnerships with Google (or being Microsoft Live - not going to do a deal with itself).
So which set of data drives decisions? How do they match up to each other? They don’t. You chose which one you want to look depending on what you care about.
I think we need to find a way to compare the numbers from different research firms and come to a consensus on which one’s are better to consult and for what purpose. Recently, Avinash Kaushik tried to do just this when he compared Comscore and HitWise in Competitive Intelligence Analysis: Why, What & How to Choose and then took that data and gave some ideas on how to use it in Competitive Intelligence Analysis: Metrics, Tips & Best Practices. Avinash has been a voice in Web Analytics, in defining new definations of metrics such as Site and Cart Abandoment rates.
Still, no one really has come up with a way to match up different numbers and it’s really up to each person to chose - but they might not know which set of numbers are better - I think more work is needed to explain what each chart actually measures - because it’s often not the same things.
