Robert_Charlton - 11:43 pm on Jun 11, 2012 (gmt 0)
I'm not sure I'm seeing precisely the same thing that Brett is discussing.
The additional inline results I'm seeing this week appear to be search refinements that resemble what I'd call related inline Sitelinks. It looks almost as if these inline refinements are additional choices that searchers might look for if they backed out of a site and did a new search on Google.
In that regard, they're more query-specific than regular Sitelinks or inline Sitelinks that I've seen before. Different queries, of course, provide different kinds of choices.
I see them as ongoing tests. Google has been doing this for several years now, with the choices either based on user behavior data... or trying other variants to provide user behavior data. During the early days of Panda, eg, I was seeing Google return pages that were algorithmically relevant but basically weak and destined to be culled.
Right now, I'm seeing 2 or 3 additional inline refinement on test searches I've just tried. This current bunch, I'm thinking, may be trying to correlate prior search behavior with user intention... that's what it looks like. Regular sitelinks (up to 6) in some cases as well, may be displayed if it's obviously a navigational query.
While a general search on a brand name will return a lot of pages for a site, as soon as you start getting very specific, Google appears to reduce that number considerably. I've also been watching this since it first happened, and Google, if anything, appears to have reduced the number of pages given to a manufacturer. That said, if I were trying to sell Apple iPads online, I wouldn't be happy about the results Google is returning.
Like many dozens of other refinements that Google has tested, these may not stick. My feeling is that while they're all generally moving towards greater searcher satisfaction, they are also extra targets, providing extra exposure for Google to test whatever it's testing.
These inline Sitelinks are not the only way Google pushes extra clicks to the surface, but I'm sure they're one of the ways.
For more on some aspects of Google testing we've been discussing, I recommend checking out this discussion and the threads it links to...
Zombie Traffic and Traffic Shaping - Analysis