As a technician I have to say I was awed by Google’s technical prowess when I first saw Google Instant in action! The ability to do a predictive query list was not that big of a deal since word processors have been doing that for over a decade. Even the ability to run the multiple queries required is not that big of a deal since Google has been writing scalable distributed processes for a long time. But their ability to get all that content down and across the last mile of copper at that speed – that was impressive. Now the big question:
What impact this will have on Advertisers?
Any time you make this level of change to the user interface there will be changes in behavior and our bet on the first impact will be the decline of the long tail keywords. We are already seeing studies that 60-70% of searchers are using the recommended keyword searches and others are stopping with the search result they see. Both of these outcomes have the potential to shorten the search query from the user.
Google has published a small specification on how these searches will be handled from an impression and CTR basis and this potentially could change how we read the data. Here is their official posting within Adwords:
When someone searches using Google Instant, ad impressions are counted in these situations:
1. The user begins to type a query on Google and clicks anywhere on the page (a search result, an ad, a spell correction, a related search).
2. The user chooses a particular query by clicking the Search button, pressing Enter, or selecting one of the predicted queries.
3. The user stops typing, and the results are displayed for a minimum of three seconds.Reference: https://adwords.google.com/support/aw/bin/answer.py?hl=en&answer=187309
We believe this will increase the total number of impressions since all it takes is a 3 second delay in response from the user to create the impression. In the past item 1 was not possible but it is also unlikely to create more impressions. What is very likely however is that it will shorten the search query.
When I first saw Google Instant I was impressed by the speed of the query execution but I suspect this is a programming illusion created by pre-calculating all or part of the results page. By doing this they have more control over the computing power demand to execute the query. Pre-calculating intensive operations is a well proven method for improving response time in interactive systems.
Not all of Google’s volume comes from Google.com so the impact of Google Instant is only some percentage of their overall volume. Bringing this to the entire market will require changes to browsers and many different pages to get to a high percentage of all the users. What we know at this point is that this is an important change that will change the way people interact with Google results. Search technology is where people, language, and systems collide and it is an incredibly complex process with inconsistent outcomes. It is time to watch carefully and start to rethink some old beliefs.
I would like to close with a broad prediction. My guess is that if you have a long-tail strategy you will find the volume in that tail falling as Google Instant implementation share increases. If your strategy is more of a broad term high volume keyword model you are likely to see more traffic, but with increasing competition based on the visibility of the predictive keywords. It will take the market about two seconds to realize that these predictive words are going to benefit from being on the list. There will be a whole new world of SEO created to chase optimization of the predictive list. The quest will become how do these predictive keywords get set and what is the algorithm doing to select these words.