There’s been much discussion about Twitter’s updated partnership with Google — which has enabled the indexation of tweets in the SERPs. The new ”firehose” has been in place for a few months now, but recently, Twitter results packs began appearing on mobile SERPs. By August 21st, Twitter feeds began to appear on desktop searches; Mozcast has been reporting that this SERP feature is displaying across just under 10% of the queries they track:
I don’t have the tracking horsepower that Moz has, but I’ve been hunting these results because I got some questions! How can I use this feature as a tactic? Is the Twitter pack simply a raw feed, or can you get ads to appear? Can I get into a competitor’s feed? And how does traffic get reported in Analytics?!? As I dug around to see what was in the wild, the questions just kept coming.
Twitter SERP Packs, the Basics
The Twitter Packs display all 140 characters plus some additional rich content – pictures and GIFs as of right now. Pictures come in the form of a thumbnail square, while gifs clearly display the play symbol:
The GIFs don’t play in the SERP (how seriously cool would that be?!?), but clicking on it takes the user to the original Tweet on twitter.
In all the brands that I searched, Google displayed the last five tweets in chronological order from newest to oldest. Trending topics appear to be a unique opportunity that carry some additional weight. When Usain Bolt was run over by a Segway last week, 7 Twitter results displayed in the pack while it was trending. Within 24 hours, the Twitter pack was replaced by a news result pack which is some impressive machismo from Google considering the speed and depth in which they can make changes.
When the kids were swooning over the 22nd birthday of One Direction member, Liam Payne, a similar pack of eight Twitter results appeared – which seems to be the top-end limit for tweets in SERPs. The detail here is that I searched a hashtag which displayed the Twitter Pack on the SERP. The pack can change when the query is reloaded depending on the activity level of the tag.
How are these things ranked, and how can you get your Tweet to appear? I’m afraid it’s not entirely clear.
The Tweeters who had their tweet displayed in the results pack at the time of my sample had the following metrics:
@latelateshow has 111k followers and the tweet had 1.2k retweets and 1.1k favorites after two hours of life as a tweet.
@OneDrecti0nFans has 361k followers and the tweet had 1.5k retweets and 1.1k favorites after two hours of life as a tweet.
@Real_Amy_Jones has 825 followers and the tweet had 94 retweets and 59 favorites after 31 minutes of life as a tweet.
@irwinsahero has 24.5k followers and the tweet had 349 retweets and 88 favorites after 39 minutes of life as a tweet.
@stylescitylight has 22.5k followers and the tweet had 106 retweets and 55 favorites after 41 minutes of life as a tweet.
@Lookatnialler has 54.2k followers and the tweet had 219 retweets and 102 favorites after 43 minutes of life as a tweet.
@Sofiapinja has 235 followers and the tweet had 2 retweets and no favorites after 24 minutes of life as a tweet.
@Liam0Leary has 81 followers and the tweet had no retweets and 2 favorites after 26 minutes of life as a tweet.
Five of the accounts that were displayed had followers in the thousands while two had less than a few hundred followers.
Retweets were even more dispersed with two tweets having less than two retweets, four having between 94 and 349 retweets and two of the tweets had retweets above one thousand.
Favorites were similar to retweets with two tweets having less than 2 favorites, four tweets had between 55 and 102 favorites and two had favorites above one thousand.
Bear in mind that this represents an extremely small dataset; conclusions are foggy at best. Taking advantage of trending topics appears to be the strongest opportunity in the near-term. So far, it seems that velocity of engagement has something to do with appearing in the Twitter SERP Pack for a tending topic, but more testing is needed.
Additional tracking of how Google transitions from Twitter pack results to news results, the timeline and what sources are being used to populate each type of result would make an interesting follow up test.
How Twitter SERP Packs Track
My deep dive started with one simple question: “Does a Tweet found in the SERP get credited as Organic or Social?”
How it should get credited is a debate in itself. Clicking the link created from the username or @handle takes you to the user profile. Clicking the image (if one appears) takes the user directly to the status message. Clicking any links which are part of the Tweet’s composition takes the user to the linked resource.
My analytic smoking gun is found when you click the hyperlinks:
What the tracked URL is tracking is uncertain, but with Twitter Source, Google, TW Camp, Ctwgr and Eauthor, we can make an educated guess. At the very least, someone is tracking this via the Google Organic location! But how are they coming through in GA?
How Do URLs Report in Analytics?
With all the garbage coming through Analytics right now, this is a sticky wicket — yeah referral spam and “secure search”, I’m giving you the stink eye! We created a non-indexed landing page for the purpose of this test.
IP filters off, real-time analytics on, and with incognito mode activated, clicking the link directly in the feed reports as:
And when I navigate to the page from DragonSearch’s Twitter feed:
And logged in it still attributes correctly but with our old favorite buddy, (not provided):
Which gets back to the garbage aspects of Google Analytics. Our test link was not indexed and the only place it was advertised was on Twitter. Your traffic is going to change based on how a user is logged in and where the link was located when they click:
But what about a shortened Tweet using Hootsuite’s Owl.li prefix or another service?
Doesn’t matter what it is – custom URL shortener, Hootsuite etc. all reported the same via Google analytics – when the link is clicked from the SERPs, it’s Organic, Google. On the respective dashboards, each registered as a click, which makes sense because they’ve never measured where the click was. To these tools, a click is assumed to have happened on the specific social network. Twitter analytics reports similarly:
Do Pinned Tweets Stick in the Feed Longer?
The SERP pack is a feed which, as mentioned above from a brand’s perspective, is going to move at a speed dictated by the amount of Tweeting done by the brand, so pinned Tweets get no extra love off of Twitter.
In the example of news results, pinned Tweets don’t stick any longer, but the ones that appear in the Twitter pack appear to behave differently – sticking around longer and getting swapped out at a slower pace than what you’d expect to find with a trending hashtag.
Do Replies Appear?
It appears that no, they don’t show up within the Twitter SERP feed. I replied to one of our SEO’s tweets both directly and by putting a period in front of it, thinking that if I was blasting all of our followers, maybe it would get picked up in the SERP feed:
Neither of the above examples appeared in the SERP feed, but they did appear in the Twitter feed. This is rational — a user searching a brand or entity will likely not care to see what other interactions that brand/entity has had. This is also good news for, say, airlines doing reputation management and responding to irate customers, since those interactions won’t be broadcast any further than originally intended.
Do Retweets Appear?
When you retweet and add a comment, your comment appears with a link to the original tweet, as seen in the example on the left, below:
But if you retweet without adding a comment, the retweet does not appear in the SERP Twitter feed. I wasn’t able to find one example of a retweet appearing in the Twitter pack, only original tweets to the handle. I’m curious to see if posts pushed from Instagram, Facebook, Twitter or LinkedIn will come through on the SERP both with and without images (which I’ll share in a follow-up post).
How Do You Get Images to Appear in the Feed?
It’s literally as easy as adding an image to a Tweet. When a GIF is used, the preview image shows on the SERP. What doesn’t get displayed is social schema markup. If you are using cards on your content for images or a summary, it won’t show in the Twitter SERP feed. In the below screenshot, the Tweet on the left contains a blog post which has been marked up with a summary card, while the Tweet on the right is a simple image added to the Tweet:
I’m curious about app cards and player cards containing media other than video. I’ll also follow up on this.
So What is the Point of the Twitter SERP Pack?
While in its infancy, it appears that the Twitter SERP Pack is another cutesy experiment which may or may not stick around. Besides offering richer results and pushing other results further down (or off) the page, this snippet gives users the ability to review multiple touchpoints and get a feel for the brand in much less time. I’m most excited about the prospects of jumping into a trending topic. This is a clear win for the most agile marketers who can create content on the fly, as there’s a lot of traffic to be earned by appearing in the Twitter SERP pack.
The debate about how traffic should appear based on the Tweet’s location is one that I’m excited to watch play out, because it’s not going to end if these snippets disappear from the SERPs. As we debate the merits of social media as a ranking factor, the argument is going to get more convoluted. We’ve been losing the ability to track content by medium and are losing the ability to understand what metrics can be trusted, as well as which metrics are a little more than inflated. However unsuccessful Google+ was, it still provided another data layer and more depth to the algorithm. It’s hard to deny, the social signal helps the search engine identify news and trends in real time. The transition from Twitter Pack to News Pack is a muscle flex to what Google can do with search and how completely optimized you must be to play Google’s game.
Special Note: I would like to give some shout outs to Jacques Bouchard and Evan Auerbach for their valuable input into this blog, Greg Martinez for his web dev support and the rest of the Dragons for clicking on my links and helping with the tests.