Recently Google announced that Google Suggest would be ‘graduating’ from Labs (after 4 years!) and making its way to every Google user. At the moment it appears only in the search box on Google.com and not the search results page (or Google.co.uk yet), but if you’ve not seen or heard about it, Google Suggest gives you at most 10 suggestions for queries as you type.
As I’ve been playing around with Google Suggest I’ve come upon a few oddities and implications that I thought I’d share with the community. Personally I think this has great implications, certainly from a PPC point of view, and perhaps from an SEO point of view, though I’m less qualified to comment.
But first, the oddities:
1. ‘The indexed pages’ number
As you’ll see from the picture, Google displays the number of indexed results for each suggested query. My first question is why? Why is the user interested? Sure, it’s nice from a keyword research point of view. In fact, I discovered while doing some research today that for a particular keyword and its variations, the variations with the most traffic had the fewest pages indexed. But why does the user want to know this information…or is Google using it to subtly tempt the user to visit the search result with the biggest number? I’m not quite sure what effect these numbers will have on search volume.
But oddly, these numbers aren’t correct. I mean, when you click the result and visit the SERP Google usually tells you there are less indexed pages than the Suggest tool said. I tried this earlier with a number of terms and the difference seems to be about 1-2% each time (sometimes more). I presume that this is because Google can’t omit duplicate content at the point of typing in the query since meta descriptions are only considered after the query has been made?
2. The suggestions
Start typing in the search box and see what comes up and ask yourself why it’s there. Google says, “Our algorithms use a wide range of information to predict the queries users are most likely to want to see. For example, Google Suggest uses data about the overall popularity of various searches to help rank the refinements it offers.” In other words, more popular results come up first.
This blog has some good examples of strange suggestions that don’t seem to work like this at all, but I’ll mention one other briefly here.
Typing ‘j’ brings up 10 suggestions, the first being ‘John Lewis’. Second is ‘jobs’. Why wouldn’t ‘jobs’ be a better suggestion than ‘John Lewis’? In fact, do many Americans (since this feature is mainly available on Google.com) even know what John Lewis is? (A UK department store.)
We might assume that Google would present the most valuable (Adwords) result first, yet these suggestions seem to contradict that too – ‘John Lewis’ is surely not as competitive a term as ‘jobs’?
It does seem that recency matters – but it’s unclear how recent something can be to appear. ‘Joe Biden’ appears just below ‘John Edwards’ but a quick look at Google Trends shows that whilst Biden is clearly beating Edwards for volume in the past two days, Edwards has had much more search volume over a sustained period.
After much consideration, I don’t think I really understand Google Suggest. Perhaps it’s because it seems like a bit of a lazy way for Google to not have to make their algorithm cleverer than it already is (no longer do they really need to work out what someone means by their query, they just give them two options and see which they pick), or maybe because it’s probably going to make users more lazy (and get rid of the long tail). Or perhaps I’m getting too worked up about it.
Nevertheless, I still don’t understand why they’ve included those ‘indexed pages’ numbers! Any ideas?
