It’s now official, Google Caffeine is live universally. Google posted about this on their blog last evening.  Matt Cutts also sat down with Danny Sullivan at SMX in Seattle in this video as well.  (With all that caffeine I wonder if they were sleepless?)

What becomes most apparent at first look is that this update is about speed of indexing new content as well as the really staggering amount of storage it takes to accomplish this task. “Freshness Matters”

I’ve not seen as many people notice another feature of Caffeine that may be the most “actionable” form an optimization perspective. Google now has more ability to associate data about any particular piece of content they index.  They are explicitly telling us that they are building capacity  into their algorithm to reference more indication of the quality or importance of a document. Also note that a document might not refer to just a web page, it could be a video or other content.

What does that tell us? Yes, your content will get retrieved and indexed more rapidly than ever before. But you also need to make sure that Google gets as many signals as possible that it is worthy of attention. Links to it, reviews of it, “Likes”, tweets and any bit of information that Google might be able to pick up are more important than ever. This has been fundamentally true for some time, but the number and nature of these signals of quality are only going to increase.

A couple days ago, I came across a post on the official Google blog that detailed a change in the way Google lets its users personalize their search experience. That change is called Google Stars.

In short, Google Stars, the replacement for SearchWiki, are what you see (if you’re signed in to your Google account) to the right of the search results. If you find a site that is highly relevant to your search, you can “star” it. After that, you you will see the starred result at the very top of the search results. Starred sites will also populate your Google bookmarks.

From a personal standpoint, I think this new tool is great. It’s quicker and easier to use than SearchWiki, and the auto-population of your Google bookmarks page makes finding the starred pages extremely easy. Another cool feature is that starred results also show up for relevant searches. For example, if I type “seo firm” and star Fathom’s website, it will also show up for the related search “search engine optimization.”

As cool as all that is from a personal perspective, it’s potentially problematic from a professional perspective. In essence, Google Stars means that search engine marketers could have yet another page one element to contend with – and one that ranks ABOVE THE TOP ORGANIC RESULT – in the SERPs.

The SERPs page after a site has been starred.

So what exactly, you may ask, is the ultimate point of this post? Simply put, as much as things continue to change in the realm of personalized search, the more they stay the same for online marketers. Despite all the changes we’ve seen for personalized search, the fact remains that:

1. Measuring success solely on rankings in today’s online marketing world simply won’t cut it. It is other elements, such as return on advertising spend (ROA), that are ultimately the more important to clients.
2. Good ol’ fashioned, tried-and-true SEO elements such as frequently updated and quality content, useful tools and resources, quality inbound links, etc are still key in ensuring your site is useful and relevant to end users. If you can ultimately achieve that, not only is it going to be more likely your site will show up higher in the organic results, it’s going to be more likely that more users will see your great site – and possibly slap a Google Star on it.

There has been some attention lately to Google’s use of synonyms and related searches. Google posted about it on their official blog.  One change is they are bolding not just plurals in the results, but also terms that mean the same thing.

In short, it is wise to recognize that there are many ways to express the same concept. Google is attempting to be less literal and yet continue to deliver relevant results.  While one searcher may use “sofa” and another uses “couch”, both might be interested in a shopping results from the same furniture store.

This may be good on a couple of levels. First, it could get closer to the searcher’s intent without regard to the size of their vocabulary.

Secondly, it frees up SEO copywriters to write like humans. That remark is, of course, hyperbole. Good writers long ago (in Internet years) abandoned the concept that if using a target phrase once is good, using it ten times on the same page  must be better. However, anything that helps us resist that temptation is an excellent innovation.

While thinking about this, I did a few searches to see how far Google is going to make associations. While you can uncover hints from seeing the terms that are bolded on a SERP, I wanted to cast the net a little wider and looked at “related searches”. There is a caveat here. Related searches are by nature broader conceptually than synonyms. but I’m still amused by what was returned when I searched for “uniform”.

I can understand the terms that modify (school, military, sports). I’m impressed that they recognize brands (Dickies, Cintas).  But considering  “unicorn” and “unicycle” as related is a stretch. And I’m disappointed that “unibrow” didn’t make the cut.

Interestingly, when searching for the plural “uniforms”, you see no unicycles. And I’m certain you’ll never see unicorns!

So, what criteria does Google use to pick the results that appear in the auto-suggest dropdown when you type a query?

While most SEO insiders I’ve talked to agree that auto-suggest results have much to do with current search term popularity, others think that additional factors carry more weight in the algorithm.  When a client’s name shows up low in auto-suggest, for example, they suspect that optimizing that company’s website and beefing up the link authority might bump its name up a few spots on the list.

Personally, I’m in the camp that thinks auto-suggest results are almost exclusively influenced by current search popularity.  But a little test I did recently left me surprised at just how quickly Google updates its auto-suggest results to reflect current search trends.

On Tuesday night, much of the nation (including yours truly) was following a special election in Massachusetts to fill the senate seat left vacant when Ted Kennedy passed away.  I suspected that a large number of people across the country were querying Google to find out what times the polls closed in that state, and decided to see if the auto-suggest results would reflect that trend.

At 7:30 PM EST, just a half hour before the polls closed in the Bay State, I started typing the query “what time do the polls close in Massachusetts” in the Google search box.  If figured the auto-suggest would present my intended query eventually, but was surprised to see it pop up in the second spot after only typing the six letters “what ti.”  To see what I mean, check out the screenshot that I took below.

The small number of characters that I typed to produce the correct auto-suggest, and the fact that it appeared so high on the list, suggests to me that Google was indeed handing back a list of results based on what people were searching for at that time.  (Apparently more searchers were interested in what time it was; do that many people really use Google to check the clock?)

The second part of my experiment really convinced me that close-to-real-time search term trends play a big role in auto-suggest results.  This part was simple: a few hours later, when it became clear that candidate Scott Brown had defeated Martha Coakley and won the election, I went back to Google and started typing the same query again.

The result?  Suddenly, “what time do the polls close in Massachusetts” was nowhere to be found in the auto-suggest list, even when I had typed out most of my query.  As soon as people across the country stopped asking Google that question, the algorithm yanked it from the list.

Of course, there are a few other variables that I could be overlooking: a higher click-through rate for specific auto-suggest results could end up bumping it to a higher spot on the list, I suppose.  But the results of my Massachusetts election test make me even more certain that search engine optimization does absolutely nothing to affect auto-suggest results.

So, what do you think?  Leave a comment below and let us know any observations you’ve made about the Google auto-suggest feature that might shed further light on the variables they use.