In the TR article leading Google engineer Amit Singhal explains ranking factors for tweets.
Some of the ranking criteria are quite clear and obvious while others remain fuzzy behind Google’s veil of secrecy. There is one big surprise: Google hates #hashtags, or at least some (probably you using too many) of them.
It seems Google treats #hashtags the way it dealt with the meta keyword tag in the past: Neither do they really count nor do they count in a positive way in case they do.
In other words: Using hashtags may hurt your tweet rankings. That’s not all of course.
The perhaps most important news is that it’s not the sheer amount of followers that makes you important in Google’s eyes.
Also Google doesn’t include all or the latest tweets in into real time search results once those appear. The tweets that get shown have been filtered in a manifold way before they reach your screen. Last but not least: There is specific semantic filtering involved. Selected tweets will appear when there is a significant amount of recurring collocations. A collocation is a combination of words. In the article the example is the search for Obama on Google.com
Sadly I couldn’t discover any real time results on that query yet to check it out.
Tiger Woods seems by far more popular or trending than the US president, not for his sports career only but due to his notoriety as a womanizer.
I’ve tested some of my theories and official rankings factors using his name.
The Google spokesperson explains that a cluster of tweets combining for instance Obama and Harry Reid may trigger the appearance of tweets actually relevant to a popular Twitter news item or meme.
Based on the Technology Review article and our knowledge of current (link based) ranking factors for Google before real time search I’d like to extrapolate some most probably useful advice on how to optimize tweets to rank better in Google real time search.
As hashtags seem to get treated similarly to the good old meta keyword tags you have either drop them altogether or use them in a manner that seems natural. Just think of regular users using hashtags not you and your fellow SEO practicioners. There might be a few common sense rules of thumb not to mess up hashtags based on that premise.
In a way Google replicated the PageRank algo for Twitter. This one is not based on links it seems but depends on the number of followers and their “reputation”. Simply put: Even in case you get followed only by a few people but they include Matt Cutts, Danny Sullivan and Robert Scoble you’ll get lots of Google Juice for your tweets.
As in the Obama example explained filtering will mean relevant clusters. Such memes that get hot will determine whether real time results show up and which ones. So in theory it doesn’t suffice to mention Obama all day over and over just to get accidentally shown in real time results once Obama is a hot topic again. You have to be part of the currently relevant cluster to show up. Also you have to
Google is also keen on localizing tweets by finding out where you are and how far away the people who tweeted something are from you. So you’ll rather get to see the tweets of neighbors than those living at the other end of the world.
To prove or disprove the suggestions above I tested some them on the Google.com results for Tiger Woods. I was quite surprised to find out that some of the above ranking criteria do not work very well yet. I was literally able to hijack the results. I added 5 tweets in the course of a few minutes. Most of the were rants, others were completely irrelevant. It was intriguing to see which keyword combinations got included in real time results and which did not:
Showed up in Google real time results:
Didn’t show up in real time results:
So as you see Google is not as good as it promises and I’d like to see it. Who am i to complain though? Nonetheless I guess Amit Singhal and Matt Cutts have some homework to do over the weekend. They haven’t even filtered the “buy viagra” tweet…