Christmas in July: Organic Tweet Analytics
Who doesn’t love a new toy? Remember that feeling as a kid during the holidays - wondering what fun lurked behind the gift wrapping of each box you found? Shaking each one to figure out if it’s a sweater or a Gameboy...and what could this strange, Millenium Falcon-shaped thing be? What a glorious feeling.
Well, as a data person, getting new metrics is a lot like that moment when you finally get to open up your holiday presents. But it’s not every day that a new toy not only brings you new stuff - it also makes your old stuff better. Today is that day.
Twitter recently announced that they have rolled out more advanced metrics for organic Tweets. If you’ve signed up for Twitter’s advertising services, you can now see a more detailed view of success around your organic content, similar to some of the metrics that advertisers see for paid content.
These new metrics provide three main additions to organic Tweet analytics:
- A timeline view for impressions per hour
- New metrics
- Impression metrics that allow us to calculate engagement rates
I wanted to jump in and show you some concrete examples of these new metrics, but as of today, the data is only accessible by the owner of the Twitter account. So I’m left with no choice other than to demonstrate the awesomeness of these new metrics by throwing my own personal Twitter data into the ring to analyze what my followers are responding to.
That’s right, I’ve used Twitter’s new analytics functionality to download the archive of my personal Twitter account, @chriskerns, with all the new metrics we have at hand. So today we’ll learn if people out there care more about bourbon, Star Wars, the Seattle Mariners, obscure pop culture or digital marketing. I know, I know - you’re excited.
First let’s take a look at the new data that Twitter is providing.
The New Metrics
Here’s an example of a Tweet that I wrote last week, and the new analytics view on Twitter:
As you can see, there’s a timeline view in the top right showing when people saw my Tweet, and a mix of more traditional Twitter metrics with some new data. In addition to Retweets, Favorites, and Replies (which have always been available through one means or another), some of the new metrics include:
- User Profile Clicks: how many times another user clicked to find out more about the author
- URL clicks: how many people clicked on the link in your Tweet
- Hashtag clicks: same as above, but for any embedded hashtag in your content
- Detail expands: clicks on the “expand” link found in many Tweets
- Embedded media clicks: other users interacting with video or pictures
- Follows: when other users click “follow” on one of your Tweets
- and more (including nice tracking around how many people decided to email your Tweet.)
Some of the measures are available in the dashboard, as seen above, and others are available by exporting an Excel-friendly file that lets you sift through all your data.
I must have been a good kid this year, because we can do some pretty fun stuff with these new metrics. Follows from within a Tweet? Awesome. Hashtag clicks? Fascinating.
But it gets better.
The new metric I’m most excited about, though, is impressions. Tweet impressions - or the number of times another user has viewed your Tweet - has always been a tricky thing to figure out. Now you can see it for each Tweet, and with a timeline view to show us the reach our social content is getting. It's the gift that keeps on giving.
With this new data, we can now look at success metrics with ratios instead of raw numbers. Raw numbers can work in certain situations to gauge success, but ratios will almost always give you a better measure of success for your marketing tactics. If 20 people clicked on one of your links in a Tweet, I suppose that could be seen as good by some brands. But when you see that Tweet was seen by 2,000 people (meaning it was only clicked by 1% of people that actually saw the Tweet) and that your average engagement per impression is closer to 5%, then your takeaway changes completely. Ratios rule.
Let’s look at a few examples from my Twitter feed where the idea of success changes with these new “per impression” numbers.
When looking at which Tweet of mine had the most engagement - determined by number of clicks on the Tweet - the following Tweet (where I had discovered the location of the new Star Wars set via Instagram geolocation data) rose to the top:
Engagement Winner (Raw Count)
Just discovered that the Star Wars Ep VII Instagram teaser pic had geolocation enabled. Location? Abu Dhabi pic.twitter.com/72rupMhmmv
— Chris Kerns (@chriskerns) May 18, 2014
So that Tweet got the highest number of clicks (over 1,100), so it must be my “best” Tweet, right? Well, not really. Turns out while it was clicked on 1,100 times, it was Retweeted a lot and seen over 5,600 times, giving it an engagement rate of around 21%. That’s great reach, but to figure out which Tweet received the best engagement, I should be looking at the Engagement rate, or Engagement Per Impression.
When I sort my Tweets by Engagement PI, the following Tweet about real-time marketing actually shows up as the winner:
Engagement Winner (Per Impression)
— Chris Kerns (@chriskerns) July 14, 2014
This Tweet was only seen 95 times, but was clicked on 21 of those times, giving it a 22% engagement rate. That was just good enough to beat out the Star Wars Tweet, and makes it our new winner.
Let’s do the same thing but with our new User Profile Clicks metric. Which topics make people want to know more about the author of the post?
My Tweet with the highest raw number of Profile Clicks, it’s this one where I was preparing to measure real-time marketing during this year’s Super Bowl, which resulted in 35 clicks to see my Twitter user profile:
Profile Clicks Winner (Raw Count)
— Chris Kerns (@chriskerns) February 2, 2014
But while that was the largest raw number of clicks, that action was only performed by 0.8% of the users who saw the Tweet. When we look at the top content that drove User Profile Clicks Per Impression, we see that the following Tweet about Top Gun was the winner with around 4% of people clicking on my profile.
Profile Clicks Winner (Per Impression)
So some guy at @555uhz is tweeting Top Gun, frame by frame, right now
— Chris Kerns (@chriskerns) February 18, 2014
And if we look at Favorites and just used raw numbers to see which was the winner, we’d see that same Star Wars Tweet in the lead. But when we use Favorites PI, we’ll find that, believe it or not, the following Tweet (which talks about the ability to favorite your own Tweets) generated the most favorites - meaning that the ratio of people who saw the Tweet compared to people who favorited the Tweet was highest for this post.
Favorites Winner (Per Impression):
Just learned you can favorite your own Tweets. Seems a bit redundant but I'm totally going to do it now on every Tweet.
— Chris Kerns (@chriskerns) June 4, 2014
The irony is thick, don’t you agree?
We can also use the data to look for interesting patterns in our posts.
- With a quick glance, I can see that nine of my top ten Tweets with the highest number of impressions included hashtags, which makes sense (they can bring in an audience beyond just your follower base.)
- I can also see that media, and not links, drive most of the engagement on my Tweets, with eight of my top ten Tweets for engagement including media, but only one including a link.
By taking the extra step of classifying your content by category and tactic, you can run a full regression to see which factors impact the numbers you’re trying to drive. Which hour of the day drives the highest reach and engagement? What’s the impact of certain hashtags vs. others on impressions?
A good analyst, armed with better data, can answer a lot of key questions to drive a bigger impact for their efforts. And this move by Twitter brings us better data.
So What Did We Learn
Not only did we learn that I need some new hobbies, we also learned that Twitter’s new organic metrics can be a very powerful tool to not only track more things, but to track the same success we’ve been tracking for years in a better way.
Yes, we can now see some cool new data that gives us more detail about user behavior, but we can also find more accurate answers to our questions about what’s working in social. In the three examples noted above, we would have chosen the wrong winner without bringing in the impression data, and that’s a huge advantage for all marketers going forward.
So as the year continues, make sure you’re setting your social goals with the right metrics (ratios) and using all the new tools that are provided by the social networks to discover what’s working best for your brand. Using the right social measures will help you drive towards the right goals, learn more about what works, and most certainly land you on the “nice” list within your organization.
If you've made it to the end of this post, you clearly share my passion for all things social analytics. I'm participating on a webinar with Bitly tomorrow (Wednesday, July 23) on the Value of Social Data.