This is a case study dealing with an SEO experiment about setting up tweets containing test links and letting other Twitter users interact with them in manifold ways to see whether and how Google indexes them.
Recently I publicly speculated whether Twitter optimizes tweets for Google now
so that they get indexed again similarly to the past when Google has direct access to the so called Twitter firehose of data. Last time I tested whether tweets got indexed was in 2009. I’m not even sure whether by then Google officially had access to the data. I think it was later on. In any case back then tweets got indexed, at least some of them. I wanted to find out again, at least to be sure not to spread outdated advice on my upcoming ebook.
I pitched the test to Ahrefs and here we are. We did it. I tested with the help of numerous of my followers. I was astounded by all the help I got almost instantly. I didn’t announce the test in advance in order not to spoil it. I wanted real life spontaneous engagement from my followers rather than a rigid set-up. At first it seemed like nobody was even online but boy I was wrong! My pretty bland and meaningless tweets got interacted with a whole lot more than I expected. Thus we can be pretty sure that the results are valid.
The set up
First I’d like to explain the basic set up I did. You know, I’m not a scientist or statistician. I didn’t use some “big data” vaults like Google does. I just did it the DIY way. To be honest, anybody can do SEO testing and it’s not difficult. I’d really like to encourage you to do so. It’s even fun and intriguing. You never know whether you follow some outdated or wrong advice from day one unless you test the assumptions yourself.
- I planned 6 tweets to test one kind of interaction or lack of it with each one: lack of interaction, retweets, favorites, replies, engagement (in any form), repetition.
- I created unique passwords to include in each tweet.
- I created 6 unlinked pages on onreact.com
- I included a link to each of the unique pages which also included the unique passwords.
- I checked on Google whether the unique password existed on the index. None of them did.
- I tweeted the tweets during a time when both Europeans and Americans (two of my largest audiences) are awake so that they can take part.
- I asked for help with a few introductory tweets and then I tweeted the tweets I had prepared beforehand.
- I added some extra instructions in-between the test tweets but I didn’t interact with the engagement on the test tweets in order not to complicate the procedure. I wanted to test simple and genuine one way engagement.
The support from the community has exceeded my expectations. I expected 3 to 5 retweets where I asked for but instead I gained 11 retweets for
SEO test 2. 9gIm1XpkM49OuPmFgqzk http://onreact.com/test/2.html Please retweet!
9 favorites for test 3 on favorites.
10 replies for test 4 on replies.
17 reactions on test 5 on engagement.
Test 1 on lack of interaction and test 6 on repetition have been correctly ignored with one exception on 6 where I asked my follower to remove the favorite and retweet afterwards. Test 2 has been favorited once by a bot seemingly who didn’t remove the favorite despite me asking for it almost immediately. I think it doesn’t spoil the results too much though.
AJ Ghergich of @seo who inspired this whole case study in the first place joined in the fun approx. an hour after it started and added another level to it. He embeds his tweets on his popular site. Thus we were able to study the impact of getting your tweets shown in a widget on a popular site:
Day of Tweet (Monday)
Within one hour none of the tweets has been indexed. Google doesn’t seem to be able to index tweets in real time or even within an hour despite their popularity.
The only instance where the password could be found is a third-party site collecting popular tweets.
I checked all other search engines I care for including
None of them showed a sign of my tweets either.
This site has only nofollow links so it’s a dead end and effectively won’t impact indexation of the original tweets. That’s why I will ignore it from now.
Day 1 (Tuesday)
One day after the tweets have been published and engaged with the situation being changed significantly for test 2 tweet, the one that got retweeted a lot. Look at what happened:
We witness a lot of pages from AJ Ghergich have been indexed where his Twitter widget shows my retweeted tweet. Below them we see for the first time an actual instance of Twitter being indexed:
Day 2 (Wednesday)
One day later, Google improves a bit by showing only a few pages of AJ’s site and then the actual Twitter page of Gyi instead of that random Twitter list nobody cares for.
I also checked whether the other tweets have been indexed. I didn’t expect it, but guess what happened? Test 5 got indexed, too. That was the tweet with all kinds of engagement, including one meager retweet. That one retweet was enough to get indexed though. Sebastian Schweyen’s profile has been indexed pretty fast so that his retweet became visible even before mine.
Day 3 (Thursday)
On day three, Google decided to show one of the latest retweets instead of one of the early ones. This is good to know. That way you could hijack a popular tweet in Google by retweeting it as the last person and rank for Lady Gaga’s updates. Al Moghadam of Koozai even apologized for being late to the party but his effort wasn’t wasted. That’s not yet the end of the story:
What do we see here? I clicked on the “omitted results” to finally find my original tweet of test 2 being indexed. Why did it take so long? Google simply indexes my profile page every two to three days. It could have been just the profile page but here we see that actual tweet URL. So all the retweets have led Google to believe the deep link to the actual tweet was important enough to follow and index.
What does it all mean?
Apparently retweets increase the likelihood of getting your message indexed. Linking to our Twitter profile is better at it but the best way seems to be embedding the actual tweets throughout your blog. Google treats Twitter like any other websites. The more links lead to your Twitter profile, the better and the higher the likelihood of your tweets being indexed at all or faster than other accounts.
AJ does it right as an old SEO and content pro. He gets fresh content for his site from his Twitter stream so that Google indexes his site more frequently and thus he also gets his tweets indexed faster. That’s also one of the reasons why he outranked mine as we’ve seen in the example from the last post.
Of course, this little test wasn’t scientific enough to be used as proof. I think it’s a good start to perform your own tests. You can even copy and paste the set up. In case you decide to do a test and write about it, I will be glad to add a link to your test results to the original post.
I want to thank everybody for taking part in the experiment and thus making it possible in the first place. So many people helped that I won’t even be able to personally thank each and everyone of you.
Image Credit: The “Twitter bird link” illustration has been made by the cool people over at Freepik.com