If I do a search for apigee on StackExchange.com (which passes through to a custom google search):


....it yields "About 59,600 results", and an examination of the results shows the vast majority if not all are from StackOverflow.com.

If I do a tag search on StackOverflow.com:


...it yields only 523 questions.

Even accounting for the Google search perhaps returning a hit for every occurrence of the word, and returning the same SO question once for every incidence of the word in the question, these results are so far apart that it doesn't make sense.

Am I misunderstanding something here?

  • On SO you're searching the tag; if you just search for the word it yields about 1,500. – ben is uǝq backwards Aug 23 '15 at 20:40
  • I'm 100% there is no bug, at least on SE side. 99.999% it's something with the way Google search works, 0.001% it's a bug in the custom search. (on Google side) – Shadow The Princess Wizard Aug 23 '15 at 20:40

Google search is a massively different beast when it comes to searching, and you should in no way expect a site search to return similar results to Google.

Most notably, Google searches the entire page for the keywords you specified. That means it searches the question, the tags, the answers, the comments, and even the "related" column at the right. Google also doesn't always limit which pages it will look at, and will include things like tag wikis, question lists, and user profiles in the results. That potentially means you could end up with results which have absolutely nothing to do with Apigee, but simply have the word on the page somewhere.

Our site search, on the other hand, only searches on a per-post basis. When searching for a tag, it will return all questions with the tag and all answers under a question with the tag. Searching for the term will only returns specific posts which actually contain that term within them.

Past that, I'm actually not sure how it comes up with that 59,600 (mine shows 59,400) - that's just the number that Google Custom Search returns. The search itself only actually returns 100 results, so it's not possible to audit what all is being considered to be "on our network" in that search, and as mentioned by Normal Human Google even returns incorrect numbers when it's less than 100 results. An actual search on google.com for "site:stackoverflow.com apigee" returns only 3,200 results, which is a much more reasonable number to expect.

  • 1
    Not sure if it's technically possible, but I think the world would be a better place if the "related" sidebar was not shown to the Google spider. – tbone Aug 24 '15 at 21:07

One source of discrepancy is the tag vs full-text search. If you search for apigee as a word, you get 1584 results.

But the main source of discrepancy is outside of Stack Exchange: the estimates given by Google CSE can be wildly inaccurate. For example: search for "apigee usergrid" promises 412 results; but once you switch to the 7th page of the results, it turns out there were only 66. (This is not the worst such example, just one I could find quickly.)

With lone "apigee" the bluff doesn't get called because the custom search only ever shows 100 results at most. Anything above 100 should be read simply as "100+", and even that need not be a true prediction.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .