Internet
How to
Newspapers
Twitter
SEO advice
Categories: How to, Internet, Twitter. Tags: , ,

Is Google using signals from Twitter for its rankings?

October 21, 2009 12 Comments

Warning: this post is mostly wild speculation, although there is some evidence in here.

Bing has launched a Twitter search engine (there's coverage here on Search Engine Land) - it lets you search tweets. (Update: Google has too.)

What I've been wondering since last week is whether Google is already using signals from either Twitter or URL shorteners as part of its ranking decisions. You could see why it want to - seeing which pages people are retweeting and passing around on Twitter would help it work out which pages are relevant for 'newsy' search terms (those where there is a big surge in searches for particular keywords).

Would this explain why one of my blog posts with few tweets was ignored - but a similar, slightly later one with 00s of tweets jumped into the top 10 results?

The evidence

Testing such a thing would be a nightmare - how could you set up 2 different, but similar, pages and get loads of people to tweet one and not the other?

But I accidentally did something pretty close to that when the Jan Moir affair erupted last Friday.

On that Friday, I wrote two posts:

Jan Moir: hot (as a search term)

Jan Moir: hot (as a search term)

As the Jan Moir affair grew, 'Jan Moir' hit the top of the Twitter trend list, and more and more people searched, tweeted and wrote news stories and blog posts about her. At 2.45pm I noticed that searching for 'Jan Moir' triggered the hotness graph in the Google results, revealing it was the 42nd most popular search in the previous hour.

During all this, my first post was largely ignored by Google - it wasn't in the top 50 results for a search on 'Jan Moir'. I was a bit surprised by this - my blog usually does fairly well straightaway for posts obviously related to newsy keywords. (And non newsy ones - 19 minutes after publishing this post, for instance, I rank 5th for a search on 'Google signals Twitter', not that there's much competition for that search term!)

The second post, however, was retweeted straightaway many, many times. And by 4pm, just an hour later, it was the 7th result for a search on 'Jan Moir'.

Both posts were about Jan Moir. Both had 'Jan Moir' at the front of the title. Yet one was ignored by Google, and the other appeared in the top 10 results. Over the weekend, the 2nd advert-related one appeared as high as 5, and as a number of people linked to it, it settled down at position 11, which is where it is today.

So did Google decide the first post wasn't worth a high ranking amongst all the pages being written about Jan Moir - but decided the second was - well before there were any web links to it - because of the volume of tweets about it?

The caveats

There are all sorts of other reasons this might be complete and utter rubbish.

  • Maybe Google was using some other data (EG Google Analytics) to decide what was relevant (but visitor numbers are a bit circular / self-fulfilling as a relevance metric - and I've not seen anyone suggest that visits to a page are an important ranking factor).
  • It could just be coincidence.
  • Perhaps Google only decided the search term 'Jan Moir' was 'newsy' quite late in the day, and decided to promote blog posts in the rankings only from that point. So the first post missed any QDF boost. Jan Moir tweets, according to Trendistic, really went hot at about 9-10am, peaked at noon, and only started to decline at 5pm. But who knows when search volumes picked up ...

What do you think? (If you've read this far and you're a Sphinn member, why not vote it up at Sphinn?)

You might also like
  1. Is the new retweet the end of angry Twitter mobs?
  2. Does Google edit swear words from the results of popular searches – or is it doing Jan Moir a favour?
  3. Twitter’s latest SEO improvement: meta descriptions
  4. New Wikio rankings for Online Marketing section
  5. Twitter’s appearance in Google: sort it out …

Share this post

Follow me on Facebook or Twitter

12 Comments »

  • Shark says:

    Really interesting post - you're right, it'd be incredibly difficult to test. It could just be a coincidence, but it's definitely something to keep in mind - if Google has decided that a query deserves fresh results then it'll need to work out how to find popular & recent articles on the topic, and Twitter could be a good way for them to find that. Hard to prove, but certainly plausible.

  • Just after I wrote this, Google announced that: "we have reached an agreement with Twitter to include their updates in our search results. We believe that our search results and user experience will greatly benefit from the inclusion of this up-to-the-minute data ..."

    So it seems they certainly have access to the data they would need.

    • dan barker says:

      hi, Malcolm, you may have noticed the twitter web interface uses Google Analytics. So from a purely technical point of view... (ie. ignoring any agreements not to look)... they've had the actual data for a long time to be able to do this.

      dan

  • Couldn't it just be link related? That second post generated 37 links vs 6 links for the first post (according to Yahoo). Say QDF triggers a part of the algo that requires on the fly link analysis to determine which articles on a "hot topic" are currently the most popular - that may make sense?

    • Sure, could be. -But within an hour (possibly much quicker, I only checked after an hour) the 2nd one was in the top 10. Can people have written sufficient posts that quickly (many of those links were in posts written over the weekend) and had them indexed by Google - and could Google have crunched the numbers that quickly?

      • I reckon they could do. Maybe you just had 1 or 2 links at that time, but that was enough to establish you as an authority, and hence why you ranked. I'm seeing evidence of Google analysing this kind of stuff much quicker than ever before, so I don't see why they couldn't do some on the fly link analysis for hot topics.

  • Mark Pack says:

    Circumstantial evidence looks striking, but ... I've had posts appear surprisingly high up in Google rankings (including a blog post about the death of Baby P topping Google search results for a long time - despite all the big media attention) and in those cases tweets wouldn't have been the explanation. So this may just be another of those odd cases?

    • Mark - yes, it's common for new blogs posts to get mixed in the top 10 for big news stories. And it could have just been an oddity.

      The interesting thing for me was that for a search on [jan muir], the posts were treated so differently despite their titles beginning jan muir - which is normally a pretty good way to do well in google for a given search term. Who knows though!

  • jaamit says:

    This is a really interesting question Malcolm - it came up at the #ProSEO seminar this week - Rand Fishkin reckons from anecdotal evidence there is definitely a rankings boost for links that have been retweeted a lot. Also Will Critchlow wrote this excellent post about Google using editorial nofollow links a while back which I would strongly recommend reading.

    Yes, it's really difficult to test with any degree of certainty because a natural 'hot tweet' will also generate other indicators such as fresh links, traffic (if you believe Google takes usage data as a ranking factor) and a sudden spike in search volume.

    But in these cases IMO you have to 'think like a search engine' and speculate a little. Google's link algo is all about detecting editorial votes for a page. Nofollow or not, lots of retweets with the same link is one of the strongest editorial vote signals you can get on the web. Frankly Google would be crazy NOT to regard this. I think their news today of the deal with Twitter only confirms that they want to use tweets more in ranking fresh content. So even if they arent using it now, there's no doubt they will be soon.

    • jaamit says:

      incidentally have you looked at Bing Twitter search? The fact that it shows you "Top Links Shared In Tweets About [XYZ]" shows they ARE following, indexing and ranking links within tweets. Would be pretty dumb not to pull these into main results for QDF terms.

  • [...] clear starting point for real time social metrics to be part of web ranking (although there is some anecdotal evidence to show Twitter might already have some influence)? Will this change how Google works with all the [...]

  • Hobo says:

    I've been surmising something similar of late. If folks are using twitter to spread links, surely Google would want to be looking at these links now or in the future.

Leave a comment!

Add your comment below, or trackback from your own site. You can also subscribe to these comments via RSS.

Be nice. Keep it clean. Stay on topic. No spam.

You can use these tags:
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

This is a Gravatar-enabled weblog. To get your own globally-recognized-avatar, please register at Gravatar.