Why does google's autocomplete not align with google's keyword tool?

[adriandg]

Is google autocomplete based solely off keyword search volume? Or is there some other factors i am missing here?

Here's an example:

Auto complete suggestions for 'storage toronto':

[storage toronto cost] [storage toronto downtown] [storage toronto rates] [storage toronto leaside] [storage toronto prices] [storage toronto dupont] [storage toronto laird] [storage toronto eastern ave] [storage toronto ontario]

Google adwords keyword tool results for these:

<colgroup><col width="151"> <col width="129"> <col width="169"></colgroup>
| Keyword | Global Monthly Searches | Local Monthly Searches  |
| [storage toronto cost] | 0 | 0 |
| [storage toronto downtown] | 36 | 28 |
| [storage toronto rates] | 0 | 0 |
| [storage toronto leaside] | 0 | 0 |
| [storage toronto prices] | 73 | 73 |
| [storage toronto dupont] | 0 | 0 |
| [storage toronto laird] | 0 | 0 |
| [storage toronto eastern ave] | 0 | 0 |
| [storage toronto ontario] | 0 | 0 |

So here is what i find confusing:

If [storage toronto cost] is the top suggestion for [storage toronto...] then why does google say it has 0 monthly searches?  Why isn't [storage toronto downtown] the first suggestion? or better yet, why isn't [storage toronto prices] the top suggestion?

So either: 1) google adwords keyword tool is wrong.

or 2) google suggest isn't based on just volume??

I've run these same keywords through Bing's Excel keyword information spreadsheet query and it came back saying all keywords had 0 searches ever, except for...drumroll:

[storage toronto prices]  with 7 monthly searches, once, in august, and 0 all other months.

Now i assume that bing/yahoo numbers are significantly smaller, but this does show that that the same keyword is the most popular, so in some way suggests that google's keyword is accurate.  So i guess this brings be back to my confusion, what other factors is google's suggest based on, because it obviously isn't primarily search volume.

And yes, i have made sure to clear caches, and disable personalized search and search history, and tried the query in several browsers, just to double check i wasn't getting a personalized list, so we can rule that out.

Thanks,

Storwell.

[jeffreytrull1]

Interesting analysis, but I doubt the auto-complete is as calculated as you are trying to make it here.

Have you seen the sites and jokes like this?: http://www.autocompletefail.com/

That pretty much says to me that you shouldn't put a lot of stock into Google auto-complete ranking the order of terms in relation to search volume.

[adriandg]

Hmmm...

Well my initial attempt here was to try to expand on possible keywords that I am missing out on.  I feel i have been putting too much stock in google adwords' keyword tool.  I have started to use the Bing keyword suggest excel tool but generally the numbers are so small in Bing/Yahoo that it is hard to get good results out of it.

Do you have any suggestions to help find more long tail keywords?

[jeffreytrull1]

Here's a guide that gives you some more sites that can help: http://www.seomoz.org/article/keyword-research-guide

It's a little dated, but most of the stuff on there should still exist.

Personally, some of the strategies I used are:

1. Look at my analytics. Where are my visitors coming from? Are there any similar terms using synonyms or different word order that I need to target?

2. Look at competitors. What keywords are they targeting? You can find hints by looking at their content, title tags, meta tags (if they're well-optimized, that is).

I think another important question to ask is Am I ranked as highly as I can be for the higher-volume terms? To me, I'd rather effectively target the terms that the most people are searching for rather than attempt to target a lot of low-volume long tail terms.

[RyanKent]

Why does google's autocomplete not align with google's keyword tool?

It's not meant to. It's kind of like asking why aren't search results ordered by PR.

Like overall rankings, Google's autocomplete has an algorithm which offers results based on numerous factors. Popularity is only one factor. Another factor is currently trending topics. Another example is your personal search history. You can read more about autocomplete here: http://www.google.com/support/websearch/bin/answer.py?hl=en&answer=106230

A detailed study analysis of factors which influence autocomplete can be found here: http://www.beattheautocomplete.com/study-results

[adriandg]

Oh thats funny! So it seems that google is judging that a lot on CTR, problem is though with that is that CTR is heavily biased by what the actual rankings are.  Nobody is going to page in 5 pages looking for something, they will likely click one of the first 5 results. So in terms of local optimization, where there is an extremely heavy geographical bias towards the city center, you end up with the google suggest of:

Keyword DOWNTOWN.

this is so silly.  Google is gaming it's own algo!

Lets take a quick check of my theory: Flowers Seattle ...

flowers seattle wa
flowers seattle restaraunt
flowers seattle downtown
flowers seattle capitol hill

Lets try another...

car rental chicago...

car rental chicago airport
car rental chicago downtown
car rental chicago midway
car rental chicago train station

[RyanKent]

Using a legal analogy, there are those who make the laws (i.e. politicians along with lobbyists and interests groups) and then those who work to understand the laws in place and use that information to serve people (i.e. lawyers).

That analogy may not be perfect but it is easily applied to this situation. As an SEO, I work to understand the policies and practices of search engines. I then utilize that information to help service the interests of my clients. If I was to spend any time on how I felt search engines should do things, I would lose a piece of my sanity. I try not to judge.

Search engines have a lot of educated people who spend considerable amounts of time researching details and data which we do not have access to. I try to think positively in that if the results don't seem to make sense I presume there is a logical reason for it. That's my approach anyway.