Keyword statistics

I was doing keyword research for a client yesterday.

Rather than trust any one keyword source, I used a few.
1. The free Google Keyword suggestion tool, to see how many searches happen in a month, and how tough the competition is.
2. A dummy AdWords campaign. Throw heaps of money at it, and estimate the number of clicks possible.
3. A paid keyword database. One of the most high profile keyword sources, with a relatively costly monthly subscription.

The variations were stunning. The paid database was under-estimated by multiple orders of magnitude. (This was for Australian figures). Apparently the figures come from deals with ISPs, select SEs and panel samples.

Given that the future was unreliable, I thought I’d try the past.

Went to Google Keyword Suggestions, and looked for the phrase “choosing a dog”.
Google suggested 1300 searches last month. (although the average is 1000/month)
I have number 1 position at Google for that phrase, so I went to Google Analytics to see how much traffic they sent me last month, specifically from Google. I got 284 hits, which was pretty nice. It also meant that position 1 at Google got me 21% of their searches last month. Not bad for MyDogSite.

Tried it with another keyword, to see if that was a fluke.
Google Keyword Suggestion tool said that last month, there were 6600 searches for “Halloween Australia” in Australia. I went to Google Analytics, and looked at my stats. Last month I got 521 hits from Google for that one keyword. Again, it was number one for the keyword at Google for most of the month, so that equated to 7% of the traffic. For interest, that was www.halloween-australia.com, so it would look pretty relevant to searchers.

For fun, I tried one last search, on a smaller scale – for “David Jones Christmas windows” (I’m a Christmas junkie). Google said there were 36 searches last month. I got 4 of them. So 11% from position 1. Small scale, but it’s a niche. (See my Christmas Australia site).

Now the paid tool, whose name I no longer want to mention, said that I should have received 7 searches all year, for “Choosing a dog”. That’s a little different to Google’s figures of 1300/month, and my actual 284 hits.

They have a disclaimer that the figures cannot be comprehensive, and are meant for comparing to other keywords, relatively. Fair enough. But those figures are out by more than an order of magnitude. They’re just misleading for my purposes.

Yep. We are no longer subscribing to that service. The free Google tool has won.