So does Twitter work? How can you measure results? Which numbers matter? Or don’t matter? Let’s look at Followers, Impressions, and Engagement.

Followers and Impressions

Twitter Analytics offers numbers for followers, impressions, engagements, and engagement rate. What do these mean? Here are Twitter’s definitions (source: screen shot from Twitter Analytics):

Twitter-Definitions To put that in searchable text:

  • Impressions: Number of times users saw the Tweet on Twitter.
  • Engagement: Total number of times a user has interacted with a Tweet. This includes all clicks anywhere on the Tweet (including hashtags, links, avatar, user name, and Tweet expansion), retweets, replies, follows, and favorites.
  • Engagement Rate: The number of engagements (clicks, retweets, replies, follows, and favorites) divided by the total number of impressions.
  • Followers: The number of followers for your account.

You can see the problem: An impression means your tweet showed up in someone’s account. Just as you don’t read every tweet from everyone you’re following, they’re not reading your tweets either. It’s also very easy to inflate both impressions and followers (just go to Fiverr.com or similar spammer sites to buy hundreds of thousands of followers).

Followers, Engagements, and the Truth

Let’s look at data from Danny Sullivan, who has 390,000 followers (June, 2014). During the World Cup, he got the following engagement: 25, 77, 119, 26, 20, 230, 69, 103, 23. That means for the first tweet, he got 25 engagements, 77 for the second tweet, and so on. The total is 735 for ten tweets, or an average of 73.5 engagements per tweet. Divide 73.5 by 390,000 followers to get 0.0001884. Multiply by 100 to get 0.01884%. Round off and you get a 0.02% engagement rate. That’s not 1%. That’s not a tenth of 1%. That’s 0.02%. That’s how few of his followers are reacting to his tweets.

If you invite 5,000 people to your birthday party and only one person shows up, that’s 0.02%, The Worst Birthday Party EVER.

sad birthday

(Note: Danny’s numbers are tweets about the World Cup, a wildly popular global event, so I suspect his numbers are elevated. I wonder about engagement numbers for his usual topics.)

Okay, is 0.02% good or bad? When I send out my monthly newsletter via MailChimp, the newsletter tool shows me the number of opens (how many people opened the newsletter) and the number of clicks (how many people clicked the link in the newsletter). If the link leads to my website for a form or purchase, I can also count those registrations or purchases. An average newsletter gets 10-15% engagement. My andreas.com newsletters get 46-54% engagement every month.

What about a normal Twitter account, which means not 400,000 followers, but only a thousand followers? Although I know the tricks to jack up follower numbers (these are described in my #TwitterBook), I don’t do that. My follower count is natural. I have around 1,100 followers. Per the Twitter analytics data for my account, my engagement rate is 1.4% (over 18,000 impressions in the last 30 days.) That’s 700X better than an account with 390,000 followers.

Nevertheless, it’s 1.4%, not 50%, so compared to a newsletter, Twitter is pretty bad for engagement.

However, we’re comparing engagement against the number of followers, and as I wrote, the follower number is unreliable, which means… engagement ratio based on followers is worthless as a metric. (Actually, it’s worse than worthless: if you tell your boss that you’re getting 0.02% engagement, she’ll shut down the whole thing.)

What matters are the business metrics: leads, sales, and revenues, which means the results. We’ll get to this in a moment.

Delayed Engagement

Several reviewers of this draft pointed out tweets often go out, get some engagement, and then, days or weeks later, pick up again as people discover them in their news stream. This means you have to wait perhaps 10-20 days to be sure to have all the data for a tweet.

What about Hashtags?

What if you have only 1,000 followers but you tweet with a hashtag that was used by 400,000 people in the last 30 days? In other words, 400,000 people wrote tweets in the last 30 days that included the hashtag.

That indicates the active crowd for that concept. (But just because they’re posting with the hashtag, it doesn’t mean they’re reading the tweets.) So it’s useful to know that #HashtagA has 400,000 uses and #HashtagB has 500 uses, but it’s not a business metric.

(How to find these numbers? Topsy.com shows this, but for some reason, their site hasn’t been working for the last 60 days.)

So What’s an Engagement in Twitter?

What counts as an engagements at Twitter? In Twitter’s (extremely) desperate need to show results (any results at all) they count everything as an engagement. Here’s the list of what Twitter considers to be engagement:

  • Retweet: Someone retweets your tweet.
  • Reply: Someone replies to your tweet.
  • Favorite: Someone “Favorites” your tweet.
  • Follows you: Someone follows you (after clicking to your profile).
  • Click on your Twitter Name:  Someone clicks on your Twitter name to see your profile.
  • Click on your Twitter Photo:  Someone clicks on your Twitter photo to see your profile.
  • Click on a link: Someone clicks on a link in your tweet.
  • Click on Share: Someone shares your tweet.
  • Click on Embed: Someone uses embeds your tweet somewhere else. (Do this by going to a tweet, click More and then Embed)
  • Card engagements: If you’re using Twitter Cards (a type of ad), someone clicks on the card.
  • Click on a hashtag: Someone clicks on a #hashtag in your tweet.
  • Click on a photo: Someone clicks on a photo in your tweet.
  • Click on a video: Someone clicks on a video in your tweet.
  • Pulse: The person looking at your tweet has a pulse, which rules out cabbage.

Okay, that last one is a joke.

Twitter considers any possible action on a tweet as engagement. However, most of these are irrelevant, so the engagement number itself is inflated. What really counts are leads, sales, and revenue.

BTW, if you advertise in Twitter, you will be charged for any of these actions, even when someone just clicks on your profile photo.

Which Engagement Metrics Matter?

The only metrics that matter are business metrics, which are top line and bottom line numbers. This means: Does it produce revenues? Does it save money? Does it produce a qualified lead, sale, or revenue?

The other numbers (views, follows, retweets, shares, etc.) may lead to KPIs, so they’re secondary metrics.

How to Increase Engagement

So how do you increase engagement? Twitter reviewed several million US tweets to see how a tactic affects engagement:

The effect on Twitter engagement by various tactics.

The effect on Twitter engagement by various tactics. Summary: It depends on the topic.

For example, if you’re tweeting about politics and you include a photo, you’ll get 62% more engagement. If you include a video, you get only 14% improvement. However, if you’re posting about music, a video will increase engagement by 35%. So you need to study your topic to determine the tactic.

(The “digit” in the charts mean that the tweet included a number. Some people really like to look at numbers.)

So… which tactic to use? It depends on your topic. Look at the charts above (and go to Twitter Blog’s posting for more details).

Twitter’s research covers only a few actions. There are several more things that you can do to increase engagement:

  • Ask a question.
  • Add a link
  • Write short tweets (less than 100 char)
  • Ask for a retweet (RT)

(I don’t include numbers for these because I don’t have authoritative data. If you know of research for these, please let me know.)

By increasing engagement, more people will see your tweets. If it’s qualified traffic (your target audience), it’ll lead to an increase in business metrics (leads, sales, revenue).

So Does Twitter Really Work?

After all of this, you’re wondering: does this really work? In September, I published another book at Amazon. I used tracking tags in the URL to see how many people clicked the link. Twitter produced 76 clicks to get the book (see who-really-reads-your-tweets), which was twice as much as what I got from Facebook or LinkedIn. That’s my KPI: actual sales. So, yes, Twitter works.

I’m looking for more data. All I need is engagement numbers for a 30-day period for accounts with significant activity. If you can share your Twitter data, I’d appreciate it. Just email me: andreas@andreas.com.

Note: I appreciate comments from @KasThomas and @JFouts on early drafts.