Why Google Just Can’t Suggest

Part 3 – How Google Suggest Affects SEOs Everybody

Before I get on to the meaty issue with merged suggestions, I wanted to just cover how Google Suggest affects virtually all search terms entered by users and throw in some interesting stats to show this.

Case 1 – eBay

I’m going to assume you know who eBay are. In terms of retail sites, they’re the largest in the UK by far in terms of traffic volume (way ahead of Amazon) based on Hitwise stats. Now, like all top sites, virtually all their traffic comes via brand search (I’ll be talking about brand search a lot more in future posts). Their biggest search term is unsurprisingly ebay. So, when you type into the Google search box, this is what you’d expect to see first right?

That makes sense. But let’s think back to before March 2009 – were people really searching for all those other terms?  Well, the good thing about ebay is that it’s a very high traffic term so we can use Google Insights for Search to check this out, without worrying too much about the crappy sampling rate. The resulting graph is quite interesting:

You can clearly see that in April 2009 (after the March 2009 UK launch of Google Suggest), traffic to the term ebay uk increased hugely. The reason is simple, people who would normally just search for ebay saw ebay uk, thought “yep, that’s the one I want” and clicked the suggestion. As a result, Google altered organic search volume to the largest retail site in the UK, literally overnight.

Case 2 – Polo Shirts For Women

It’s not just huge terms and brands that are affected; consider the rather more modest term polo shirts for women. Now, below is data from a site that has ranked 1st organically for this for a very long time, but the fact is that pre March 2009 virtually no one searched for this term. It just wasn’t on the radar and it certainly wasn’t optimised for. So let’s look at Google’s most reliable search volume reporting tool (Analytics FTW!) and see how organic traffic changed, again literally overnight:

You can clearly see on the graph that in April 2009 there was a proportionally huge increase in traffic, purely due to the implementation of Google Suggest in the UK. And the reason? When you type polo shirts into Google, you see the following suggestions:

Now it becomes quite obvious why this term receives far more traffic than it ever did previously. Users who would normally expect to refine by gender once they land on a site that sells polo shirts are now given the option by Google to do this one stage earlier.

Are these suggestions a bad thing?

Google would certainly argue these suggestions in particular benefit users. It’s kind of difficult to say they don’t; what bugs me more is the implementation. The principle problem is that Google will always try and populate all 10 suggestion slots, regardless of the difference in traffic volume between search terms. You then often have a situation where terms that received virtually no traffic before March 2009, now receive much more as I’ve demonstrated above. Should Google really be changing search patterns like this?

Regardless, it’s something all SEOs should be aware of. Google Suggest launched over 1 year ago in the UK, but I’m certain there are many people out there who haven’t re-optimised their SEO structure to take account of keywords which now receive more traffic. Don’t forget to consider your Adwords PPC campaigns as well as the same applies.

Merged terms

As you’ve probably already gathered from part 2, this is the big problem with Google Suggest. Consequently, I’m going to demonstrate the impact of this through a couple of examples.

Case 3 – Search Engine Optimisation

OK, so we know that Google Suggest merging terms is a very, very bad thing. We also know that it came about due to very poor half-assed implementation and they merged based on global search volume, instead of local. Just as a quick reminder, here is what happens when you try and search for search engine optimisation in the UK:

But how did this affect search traffic volume for this term? Well the answer is “significantly”. As this is a low volume term, rather than relying on some Google tool, I pulled traffic stats since March 2009 for search engine optimisation vs search engine optimization from the ever useful Hitwise:

As it’s a low traffic term, even with Hitwise the data isn’t great but you can clearly see that in March 2009 search engine optimisation was the more popular variant. With the advent of Google Suggest, more and more people have been clicking search engine optimization instead (hey, why type when you can click) and thus the “Americanised” version of the term now receives more British search volume.

This has a huge knock on effect; I have seen many people speculate that there is just some natural trend in UK English towards “Americanised” variants of search terms. They look at the changing levels of keyword search volume and understandably reach the conclusion that the language is just evolving. This is simply not the case; the biggest reason by far that “iz” word variants have risen in popularity vs “is” is Google Suggest.

You then get a horrible keyword-circle-of-death syndrome. Google Suggest influences search volume and more people then decide they have to optimise for the “Americanised” version of words. You then have a situation of falling search volume for “English” word variants and falling levels of pages optimised for them. Rinse and repeat for a few months and eventually the original “English” version completely dies out. All thanks to the terrible implementation of Google Suggest. Sorry, but this is just not right.

Case 4 – Personalised Gifts

This example really demonstrates how badly Google Suggest is implemented. The term personalised gifts is quite a high volume term that has received the suggestion merging treatment. Have your salt ready, as here’s a comparison from the Adwords keyword tool:

This was taken at the end of April 2010, so it’s pretty clear which variant gets more UK search volume right? Just to be super sure, let’s check which version is optimised for most in the UK:

So, all round we know which variant people in the UK want to see and indeed which variant is likely to provide them the best user experience. Hence, what does Google magically suggest:

You could not make it up. Thanks to bad implementation and suggestion merging based on global volume, Google STILL suggests the “American” version of the term. You can probably guess by now what’s happened to search volume for personalised gifts since March 2009:

From nowhere, the “American” version of the word now receives more search volume than the “English” version. It’s completely insane and it makes my blood boil a bit if I’m honest. Think how f***ed off you would be if you had spent time optimising and getting to number 1 in the SERPs for personalised gifts, only for the diabolically poor implementation of Google Suggest to take away your traffic in one fell swoop. It’s a complete disgrace and I have a lot of sympathy for SEOs affected by this.

Bing and Yahoo

You may be wondering if Bing and Yahoo adopted the same suggestion merging process. Quite simply, no. They both happily suggest “English” and “American” variants based on what you search for. At least, they do for now (I did notice certain people from Google Haifa have moved across to Yahoo)….

Bing does show the “iz” version first in the UK, which suggests an inferior lack of localisation versus Yahoo (which correctly shows “is” first), but at least they haven’t just merged the terms together!

The bigger picture

I don’t want to get too philosophical but if your children are searching on Google, do you really want them to be forced down the path of “American” spelling, without even being given a choice? It’s only going to get worse in the future….

And what is the future?

What Google really should do is shut down Google Suggest, fire a whole load of incompetent people and start again. It’s a complete shambles of a system and without doubt the poorest piece of implementation Google have put their name to (and yeah, I have seen real time search). Yet instead they continue this policy of building more and more crap on top of the already crumbling foundations. The next step? Well, this picture tells a thousand words:

Yep, the next step (already running in the US) is paid ads at the top of the suggestions. Currently it’s just for brand names, but how much would someone pay for that slot for terms like insurance, poker, loans and so on. Quite a lot I would guess and then your SEO efforts won’t matter at all because users won’t even bother clicking through to the first page of organic results! You better believe Google is considering this very seriously so take this as a warning. Personally I think showing Ads like that is an abomination, but Google needs the cash right?

Had enough of Google Suggest? Me too.

That’s it for my lengthy look at the heinous crime that is Google Suggest. In the final part of this post, I’ll be covering the “Did you mean” suggestions, the reason being they are very much tied in with Google Suggest. So, click below for part 4 when you’re ready

1. Why Super Brands Don’t Care About SEO

Be Sociable, Share!

•