Twitter Analytics will leave you wishing for more
- 29 August, 2014 08:16
This week Twitter began letting anyone use its Analytics service.
Go to analytics.twitter.com and you'll see metrics that show how your followers interact with your tweets as well as data about your followers.
Plus, many of the graphs that Twitter offers will make you wish that you could drill down a bit. Given their limitations, it's hard to figure out how some graphs might be useful.
When you visit your analytics page now, you'll see a graph showing "impressions" as well as a list of your tweets and the number of "impressions" and "engagements" as well as the "engagement rate" for each.
The original version of Analytics showed a graph that plotted mentions, follows, and unfollows over the past month. It also listed your tweets and for each, a count of faves, retweets, and replies. I didn't take a screen shot of the old layout but you can see it in this story on Search Engine Roundtable.
With the old version, at least I knew what the metrics were. But "impressions"? Hovering over the column, Twitter offers this description: "Number of times users saw the Tweet on Twitter." Really? Based on the way I use Twitter, I just can't imagine how the company knows which tweets I see. That makes this metric pretty meaningless to me.
According to twitter, "Engagements" means " Number of times a user has interacted with a Tweet. This includes all clicks anywhere on the Tweet (including hashtags, links, avatar, username, and Tweet expansion), retweets, replies, follows, and favorites."
That's a lot to add into one bucket. You can drill down a bit by clicking on the tweet, although it's not at all intuitive that you can do that. I didn't figure that out until I happened to see this Information Week story with a screen shot. Clicking on a tweet lets you see numbers on replies, detail expands, retweets, favorites, and impressions. That's helpful because you're likely more interested in retweets and replies then the vague "impressions."
Most other graphs don't let you similarly drill down into the data, making some of them less than useful. For instance, a graph on the main page shows you how many retweets you've had over the past 28 days and how today's number of retweets compares. But if you click on a bar showing a high number of retweets two weeks ago, it doesn't show you what tweets drove the retweets. If your goal in using these analytics is to increase engagement by looking at the kinds of tweets that increase engagement, knowing you had a lot of retweets on Aug. 12 isn't particularly useful without knowing what you tweeted.
You can also hit a tab called "followers" to see some nice metrics about followers. For instance, I'm not surprised but I am disappointed to find that 78 percent of my followers are men. Given that I tweet most often about tech and this field skews heavily male, it's not that surprising but a more equal split would be ideal.
I also see the percentage of my followers who follow certain other people on Twitter and that 18 percent -- the largest portion -- live in California.
The page also shows the top "interests" of my followers, including topics like technology, tech news, computer reviews, and startups. How Twitter knows this, I'm not sure -- it doesn't describe how it compiles that one.
This is another section that leaves you wishing for more. If I were a business and wanted to up my followers in a certain category, I might like to know who is in each. That might help me come up with a marketing plan that would attract other similar types of people.
Since the redesign a month or so ago, I find myself consulting Twitter Analytics a lot less frequently. The information is less useful to me, particularly at a glance. Now I see "impressions" and "engagements" and have to drill down to get some specifics. In the past, I saw retweets, faves, and mentions -- metrics that are easy to digest and meaningful -- at a glance.
I think it's realistic to hope that Twitter will continue to refine its Analytics service, adding more types of data and giving users more tools to try to get the most intelligence out of them. If it does, I expect to consult it more often.