Keyword tracker lists one important keyword at #4 for Yahoo when it's actually #7. Small difference I agree, but is it the case that the tracker is pulling data from a different data centre than Yahoo.com or is #4 a likely spot in the not too distant future - ie a predicted SERP?
I is most likely a data centre problem, but I have noticed a lot of variations between the data tracker reports and what you find with an actual search on Google lately.
The KW tracker uses the api of google and yahoo rather than a specific datacenter. The api does not always seem to mirror google.com and also appears to be less geotargeted than the actual serps.
You're absolutely correct on both points. I think those of us searching from countries other than the USA are definitely seeing search results skewed to our country. I have never read an explanation of what the Google API is returning. Does it use a specific datacenter, is it random, or does it use some other source of information?
I like to think of the api as giving a standardised result unaffected by geotargeting like all of the other results. Whether this is correct I have no idea
I hate looking at the KWt some days (like today), shows -9999 on all my terms, but they all still rank. Nothing like that surge of fear to brighten up the morning LOL
The keyword tracker uses the API (as you know). The API does in fact use the data centers, just like the normal Google search results. Google also geotargets results based on the country you reside in (if you don't use the advanced usage script, Google sees the requests as coming from San Diego, California... if you do use it, the requests come from where ever the server is physically located that you put the script on). So the results it gets *are* what is showing on www.google.com, it just depends on who/where you are.
If you are using the advanced usage script and the server that hosts it is down, what shows up in the tracker?
But that doesn't explain the differences I see between tracker results and my real searches. I used the advanced script so the API is running from the same Internet connection as my search query.
Don't know... never tried it. Google can route to a certain data center on any criteria they feel like, and I'm sure they have more criteria than physical location (IP address). For example, the the load of the data centers and the type of requests are probably prioritized. I'm guessing Google puts a higher priority on "real" web searches (www.google.com), so it could be if the "best" data center is moderately loaded, it could be www.google.com searches go there, and api.google.com searches fall back to a different one. It's all speculation of course, but it's certainly how *I* would design such a system. I've even seen two different browsers routed to two different data centers at the same time on the same computer... so looks like they are even doing different routing based on user agent (not sure why that would be, but I've witnessed it firsthand).
Hehe okay. I just did by accident today and I think it accounts for my -9999 drops. First time that's ever happened.