Overture sample size

Latest percentage of total web searches?

[Robert Charlton]

I think it's been a while since we've discussed what Overture stats mean in terms of total monthly web searches. At this point, with Overture results appearing in lots of places, the right multiplier to use has gotten very fuzzy for me.

I understand that a lot of Overture numbers are pretty skewed by automated searches, and I've observed that demographics between Google and Overture numbers are very different, so not all ratios will be the same... but what are the latest thoughts about what the percentage (or percentage range) is?

[WebGuerrilla]

You always come up with the tough ones. :)

I haven't really spent any time looking at it since we last discussed it.

I guess it would be a good time to dig through some old notes.

[Robert Charlton]

It was a little bit easier when the results only appeared on GoTo itself. The best I can find in my notes is that the multiplier used to be 20 or 25 (ie, GoTo represented 4% to 5% of all crawler based searches), and that, as the results spread to other engines, the multiplier dropped to about 15.

Before posting last night I did some searches for our old threads (when it was still called GoTo) and got a 404. But I'm sure that whatever the figures were, they're different now.

[Robert Charlton]

PS to the above... I'm seeing someone quoting the 15 multiplier as gospel. No way! That was a long time ago... I have no idea what it is now.

[gmoney]

I’m trying to determine Overture’s sample size by trying to find ways to compare Overture’s data to Wordtracker’s data since Wordtracker allows us to get a fairly accurate impression of their sample size and how it relates to the Web.

Because Overture combines plurals and perhaps other variations in their “Match Driver” campaign, it is difficult to draw a comparison with Wordtracker. I tried various different keywords and got various different results. I feel pretty good about my tests with the keyword “Einstein”. This one isn’t all that susceptible to the whole plural business with Overture. Anyway, the ratio of Overture/Wordtracker(count) for “Einstein” was 13.7. The ratio for “Albert Einstein” was 9.9. I also summed up all frequencies for search terms that contained “Einstein” in both Overture and Wordtracker and then I took the ratio of Overture/Wordtracker and found it to be 8.2.

Anyway, if you take the Overture/Wordtracker ratio to be roughly 10 and use the data that Wordtracker database size is 307 million and their estimate that the total number of searches on all the search engines combined is 319 million per day then you can estimate that Overture’s (and its affiliates) sample size is roughly (10*2)*(307/60)/319=0.32 or 32% of all searches conducted on all search engines combined. This would indicate that the “multiplier” is roughly around 3.

[Slade]

gmoney: Wow...

The best I can find in my notes is that the multiplier used to be 20 or 25 (ie, GoTo represented 4% to 5% of all crawler based searches), and that, as the results spread to other engines, the multiplier dropped to about 15.

Someone help me get a handle around what yall are talking about?

[gmoney]

My thinking is that the Overture search tool gives estimates for the number of searches on Overture and its affiliates (Yahoo, MSN, etc.) so that you can get an idea of the amount of traffic you get from Overture and the amount you might expect pay to Overture. It would seem natural to include the affiliates in the search term tool since, after all, Overture’s affiliates are the primary ingredient in Overture’s service. Overture by itself is maybe 5% but if you include its affiliates you might expect the number to be quite a bit higher.