It seems that Google may be starting to roll out "Previous Query Refinement" to their natural listings as reported by Danny Sullivan earlier this month after his interview with Google Vice President of search products & user experience, Marissa Mayer. If you've not already heard of this term before it basically refers to how Google changes the ads and listings it displays based on the previous queries of the searcher.

The popular example in this case is if someone does a search for "Spain" and then afterwards, a search for "travel" the listings will be based around "Spanish travel" rather than the individual keyword. In addition Google will be looking at a list of previous searches rather than just the previous query and could therefore potentially highly relevant results far quicker than before. As Danny Sullivan stated "Search history could be Google's long-term memory and the recent queries could build the working memory…".

The screenshot below clearly shows how this is already happening with adwords as the previous query "harrison ford" is affecting the adwords results despite the next query "cornwall" being unrelated. Google combines the two searches to show ads related to "fords in cornwall".

The same can be seen on the following example.

This is of course what Google users, who have been logged in and enabled the "Web History Service", have been seeing already however rolling this out to searchers who haven't logged in, will undoubtedly have a dramatic impact on the rankings of many websites.  This will also mean that the variation in Google results will become so dynamic that no two searches will be the same.

Whilst the jury's still out on how much the Google search results will be improved by "Previous Query Refinement", it may level the playing field for webmasters that don't have the budget to compete on major search terms and give them a better chance of being included in this kind of search query. In terms of how popular this will be with searchers, already some are commenting on how this could potentially produce extra "noise" in Google search results and of course there will always be the "naysayers" who will doubt the intelligence of such systems. However this does have a lot of potential as well as challenges.

No doubt some SEO's will be reeling over this as the reliability of data produced from keyword research tools will be questionable due to the difficulty in reproducing search queries based on a previous history. However technical challenges of this sort are what we thrive on in this industry and solutions are always possible.