I’ve never had much faith in the estimated number of search results returned by Google (for good reason). Every so often I run queries that I know will return the largest results set possible and I note the number of pages found. For example: searching for “the” and “http”. And boy do the supposed number of results keep getting bigger and bigger! Like the length of the preverbial “fish that got away” with each telling of the tale…
Today when I searched for “http”, Google returned 15,470,000,000 results. That doesn’t seem too out of whack until you consider that only 8 months ago (September 2005) that same query on Google returned 2,380,000,000 pages. (Back in November 2004 that query returned 588 million pages.)
We could chalk that monumental increase up to changes in Google’s algorithm in how they handle that unusual search term. So let’s look at another large-result-set query…
Today a Google search for “the” returned 23,850,000,000 pages. In September 2005 that query returned 3,420,000,000 pages.
And today a search on Google for “www” returned 25,270,000,000 results. In September 2005 that number was 3,500,000,000.
Then, a quick dive over to Yahoo revealed that the estimated results numbers actually DROPPED over the past 8 months. Yahoo today returned 1,400,000,000 for “http” compared to 2,130,000,000 in September 2005. Today Yahoo returned 7,370,000,000 for “the” compared to 11,200,000,000 in September. And today 8,810,000,000 for “www” compared to 14,300,000,000 last September.
So Google’s estimated numbers of results appear to be ballooning while Yahoo’s are shrinking. Hmmm.
Moral of this story: Don’t bet your life on the numbers of results returned by Google. They’re most certainly overrepresented.
Of course this is based on anecdotal evidence. I’d love to hear if anyone has done a more thorough study of estimated numbers of search results over time.
SIGN UP FOR EXCLUSIVE WEEKLY CONTENT 
So, what would be a more reliable way of providing competition for calculating KEI then if we can’t rely on these results numbers?
Hi Dan,
I’d say it’s much more reliable to manually review the top 10 results and see how SEO-savvy those top 10 pages are. If they aren’t very clued in, then it’s usually pretty easy to break into the top 10, regardless of what the KEI score tells you.
This is in fact interesting. I have comics appear sometimes on the first result of 76 million, sometimes it’s 35 million (“career day” for example.) I have many results that say I’m in the top ten of only a few hundred thousand results and that tells me it’s not a common item that people want to search. If you type “most searched” in image search for example, my cartoon is one of the first ones due to naming, not the fact it was googles most searched image.
Really interesting stuff, Google can be quite the complex beast at times.