On December 11, 2014, Stone Temple Consulting published a
study on Twitter engagement. In this study we examined nearly 2M original tweets in western languages (US and European languages), to see what factors drive engagement on Twitter, where engagement consisted of Retweets and Favorites.
Impact of Social Authority
One of the unique factors in this study was the examination of how Twitter behavior varied by Social Authority. Followerwonk was used to measure Social Authority, and in the study one of the first things we did was that we measured its impact on the chances of engagement (e.g. getting at least one RT or favorite) get per tweet. Here is what we found:
The number of Retweets and Favorites you are likely to get vary greatly by Social Authority. For example, users in the 30 to 39 range get Retweets on only about 17% of their tweets, and about 1.7 Retweets per Tweet. On the other hand, users in the 80 to 89 range get Retweets on 90% of their Tweets, and about 435 Retweets per Tweet. That's a huge difference! Here is what Rand had to say about the way Followerwonk's Social Authority works:
"Peter Bray, who founded Followerwonk before Moz bought it, always had a passion around discovering what does true authority mean on twitter and how can you identify someone who is a real influencer on that platform. One of the challenges that he encountered was the frustrating vanity metric of followers and the idea of using just raw followers to indicate how important or influential someone was. The problem is, on Twitter like many other things, that's pretty gameable through a followback or unfollow strategy, you can grow your follow count pretty massively. There's also weird anomalies where certain people have huge numbers of followers, but they are not influencers or they rarely tweet.
So, Peter came up with a brilliant idea to create a number that was not based solely on follower count, but based on every input we could possible get; through a machine learning algorithm, mapped to and correlated with the number of retweets that the average tweet sent by an account receives. So, your graph there, I look at that graph and I'm very proud of social authority because it shows that it's performing exactly as it should: the more social authority you have, the higher engagement you can expect to get. Social authority is exactly that, it's a measure of what is the likelihood that if this person sends a tweet they will get, in particular many retweets, but I think favorites and replies are probably well correlated to that."
Images Are the Biggest Key
Of all the factors we tested, by far the biggest influence on engagement came from including an image in your tweet. In particular, for Social Authority levels, from 1 to 59, including images resulted in 4 to 9 times as many Retweets and 5 to 12 times as many Favorites!
That would suggest that you'd be crazy to not include an image with all your tweets, or does it? Consider what Rand Fishkin said at
11:46 of the show (the link will take you straight to that point in the show, as will the other links referencing the show below):
"I think the interesting thing for me is this data is useful, but I think (over-optimizing based on it) can lead to dangerous behaviors. ... I tried a few experiments over the summer with tweeting images; specifically image based tweet storms. ... When I attempted those, I found that by filling up my followers feed with multiple image blocks, I had my most lost followers during times when I tweeted images the most heavily.
I would say images on Twitter work great when they stand out from the crowd, but when you're overwhelming the timeline, if you think "I should only ever tweet images" or "I'm going to tweet image after image after image", I think you may shoot yourself in the foot. I certainly shot myself in the foot and I could see my stats in followerwonk that my lost followers were the highest on those days."
So yes, images can have an impact, but it's important to use them in the right time and place, and that does not mean every time and place. With that in mind, consider what Neal Schaffer said about
why they have an impact:
"I'm not surprised by that data at all. We've seen similar data studies with images, whether it be in tweets or how photos are more engaging on Facebook, or what have you. Then you have these pure visual social networks that have come out of nowhere whether it's Pinterest, Instagram, or Snapchat. We engage more with visual, it's shown more prominently in the timeline and therefore it's not surprising that it's going to get more engagement, in this case retweets."
You can see more of the live action in this 5:16 video extracted from the show, including a great set of tips on how to get more out of your images:
Character Count and Time of Day
The study also found that increasing character length correlated strongly with more RTs and Favorites. For example, Tweets that are 121 characters or longer are more than twice as likely to get at least one RT than those that are 40 characters or less. That's a pretty significant difference! You can see the discussion on the video
about character count here.
A common factor we have heard over and over again is that there is a right time of day to tweet. Our study showed something entirely different:
When we
discussed this in the video, the consensus of the panel is that there are many factors that could be behind this being the case:
- Twitter is used by many as a news feed, and it's easy to get up in the morning and review all the incoming info for the past 8 hours. As a result, the impact of when you sent a Tweet out is lessened. The some of those people Retweet it and that helps make it fresh in the stream at other others anyway.
- During peak business hours, it's harder to get people's attention, so Tweets in off hours have some advantages in terms of people responding in real-time
- Twitter is not really used as a discussion platform. RT and Favorite levels are much higher than Reply levels. As Rand said: "Twitter would be overwhelming and almost unusable if you received the same number of Replies as you did Favorites and Retweets"
Other Factors
The study, and the video contemplated many more factors and their impact on Twitter engagement. These included including hashtags, links, and mentions of others. For most users, the data suggests that there is a solid gain in total Retweets and Favorites that you will get by including Hashtags (about a 30% gain in total RTs on average). On the other hand, only some users get a gain by using Links, and most users will get a negative result by including Mentions of others.
You can see that none of these factors stack up to the impact of images, which is clearly the most powerful factor in play in Twitter engagement. Hashtags are a clear positive, so it may make some sense to get in the habit of using a Hashtag. But, Links and Mentions should only factor into your Tweets where their use truly fits the content and context of the Tweet.
Conclusions
Twitter is a powerful platform that is used largely as a social newsfeed by many. For building your own personal brand, or the reputation and visibility for your business, it's important to understand how others are using it, and tune your behavior accordingly. As Rand noted during the show, Twitter is not generally used as an interactive discussion platform. Rand, Neal and I, each offered some closing thoughts at the end of the show, as you can see in this 2 minute video:
Tailoring how you publish content on Twitter to the needs of your audience makes sense. As Neal Schaffer said during the final few minutes: "If you're already doing really well on Twitter, don't change what you're doing", but also "Always be experimenting with new things". This just makes sense.
However, the study can provide you with a keen understanding of how Twitter is being used by others, and what works. As Rand Fishkin noted: "when I talk to people and marketers ask me for advice, about Twitter, I can say not only am I giving you this advice, but here's a large agggregation of data to backup these statements, I think that's really healthy and helpful because one person's opinion about how Twitter works is not nearly as powerful as that same opinion backed up by reams of data, and analysis". Rand also noted: "one of the things I haven't done at all is use hashtags, I think this clearly suggests I need to try that".
For my part, I believe the content and context of what you are tweeting should always come first. If it makes sense to mention someone, because they are the creator of a piece of content, or you learned about it from them, you should do that. The demands of correct social behavior, and of building relationships online, trump any issues with how that might impact total engagement with an individual tweet.
I also notes in the closing: "with Rand's note earlier that you don't shove an image in every single Tweet you ever do, if you have something where you are really looking for major engagement, you would be nuts to not include a good compelling image, well-crafted to reinforce the content of the tweet". I do think this is important, but I'd suggest taking it even a bit further.
For those tweets where you include an image, don't just throw something quick together. Invest some serious time in it. Visual communication is a mix of art and science, so don't sell it short. In addition, measure the results, so over time you can learn what works and does not work for you. Since they are such a powerful factor in engagement, don't short-sell the effort you put into making the image communicate.