Putting Google Instant to the test

With the arrival of Google Instant some pessimists have gone as far as predicting the death of SEO. The trouble is no one has put this to the test so we did.

We asked two of our clients – Photobox and Jobsite – if they’d act as guinea pigs in the quest to finally throw some measureables into the debate. The results make for an interesting read and suggest that, despite all the hype, Google Instant is not an instant game changer.

Since the 6th of September 2010 Google Instant has been available to Google Chrome users (7% penetration in the browser market) and to users that are logged into their Google user accounts, which according to The Latitude Group are around 650,000. According to the same source, the overall affected user population is 2.1%.

To make things clear, Google Instant is a new user interface implementation that dynamically changes the search results (both paid and natural) as the query is being typed by the user, and at the same time provides key phrase suggestions that appear as an extended drop menu below the query.

Paid search managers worry that the dynamic change of paid search results would cause impression figures to increase, and as a result decrease click through rate negatively affecting cost per clicks.

This, according to Google, will not be the case, as an impression will only count if there is no UI activity for 3 seconds or a query has been selected.

Another potential impact that Google Instant could have is on long tail keywords, with one half of the online articles on Google Instant insisting that it will reinforce it and the other half that the long tail will be a “thing of the past”.

The reasoning on both sides is that as the user is typing the query, he or she will be distracted midway through typing by a query suggestion and choosing to use it instead of the query they originally had in mind. The long tail argument comes into play as the user might choose a longer tail suggestion or cut his query string shorter by using a shorter suggestion.

To test if this is a plausible scenario we need to see if there is any change in the traffic towards our long tail, which in paid search is hard to do, as bids and ads can change, negatives can be added muddying our results.

However, there is one set of keywords that is very constant, both in position, and (hopefully) performance. Brand is the only campaign in any account that does not vary in performance and has the least amount of changes to both traffic and bids.

For a brand term to be affected by Google Instant the suggestions provided should be distracting, but relevant. This could happen if the brand name begins with a term that is a generic representative of the product category, where the engine would have a variety of suggestions.

So Amazon would not be a good example in our case as suggestions actually reinforce their brand terms:

A brand such as Photobox,  that does incorporate a generic term, could potentially be affected as the first suggestion is a competitor:

To see if this does affect the brand campaign we created a traffic report for these terms over a period of the last five weeks. The results reveal that traffic has been fairly stable, with only minor fluctuations which are to be expected.

In addition to the above, we did not see any spikes in CPC’s or drop in Conversion Rates, suggesting that the users did not change their search queries.

Natural Search

When it comes to the natural search side of things, all sorts of doomsday thoughts were buzzing through the SEO Community. “Death to SEO” was being bandied around and there seemed to be a genuine panic that natural search will no longer exist after this latest move from Google.

One vital element that appeared to be forgotten was how Google ranks pages in the first place. The way in which it chooses results for the suggestions is still pretty much the same as it has been for a while.

After the initial storm had settled, there was then the discussion about long tail terms dying off, long tail terms becoming more common and also a drop in the number of keywords referring traffic to your site.

All three of those scenarios have a chance of being correct. As to which it is, it is too early to say.

A look at a super generic term such as “Job” was not suggesting much in the way of long tail search suggestions. A Government service (high authority) in the form of Jobcentre Plus appears first, followed by Jobs and then Jobsite.

Over time, Jobsite could start to notice a difference in referral traffic for the term “Job”, as the instant results are dominated by Jobcentre Plus pages, whereas in the non instant version of Google, the top slots are dominated by recruitment websites such as Monster, Total Jobs and Jobsite.

However, Jobsite is the only brand suggested, so searchers may elect to choose that as a suggestion, as well as be exposed to a brand they may not have thought of. This means that traffic could increase.

Jobsite also benefits if a user continues with the term “Jobs”. Not only are they currently ranked first in the Natural Search but they also appear within the suggestions.

This in itself will increase brand exposure. Let’s take a look at the natural search results from before and after Google Instant for the term “Jobs”.

The graph above shows natural search traffic from two weeks before Google Instant (red) to two weeks of Google Instant being live (blue). An increase of over 5% was seen, but is this just normal traffic growth?

A closer look at the before, launch and after of Google Instant reveals an initial boost on 6th September, followed by what would be considered normal traffic patterns.

Currently as it stands, there does not appear to be too much in the way of change for results. However, this is quite possibly down to there only being two weeks of data to take from, plus the fact that only Google account and/or Chrome users see this.

To recap, we know that the number of people that actually get the functionality is small, the changes are not ground breaking, as auto complete has been around for some time now.

It is of course early days. The functionality will become available to more users over time so we will be able to gauge the impact more accurately, although the initial result, both for paid and natural, do not suggest that Google Instant is an instant game changer.