Best metrics for measuring user engagement

Posted by | December 27, 2013 | Analytics | No Comments

Achmed Awad, our CEO and business architect, knows a lot about social media and it’s techniques. Time to share his thoughts about the best metrics for measuring user engagement. It started with raw metrics then went to interpreted metrics. Now it is time for predictive metrics. This his how Achmed puts it.

en·gage  (n-gj)

v. en·gageden·gag·ingen·gag·es
To involve oneself or become occupied; participate: engage in conversation.

Let’s face it: websites are modern day business cards with just limited abilities to actively engage in conversations. It’s an excellent platform to portray your solution, your message to your audience. But it’s severely limited in ways to start a conversation. That’s why social networks like Facebook are so successful. They fill in the latent need of consumers to engage, converse with their beloved brands.

When we started Media Injection 4 years ago our first customers wanted to know how well they did on Facebook, Twitter, Youtube etc. It’s funny to see how the needs on reporting changed across the past few years. The goal always remained the same: How well do we succeed in engaging our fans?

Raw metrics

The first metrics were all about the amount of fans a brand acquired in a given time period, how many likes or shares a piece of content received. In the beginning I’ve often been at meetings with customers from Fortune 500 companies where the marketing team responsible for social media enthusiastically reported that they got 10.000 new fans, or 100k likes! The absolute amount sounds great of course, but how does this relate to past performance, or what does it mean in terms of percentual increase? We think that these metrics should always be relative to what you already have.

Interpreted metrics

From there we quickly grew to interpreted metrics. So you acquired 10k new fans? But you already had 1m, then it’s still OK by all means but not great. On the other hand if you started your measurement with just 5k fans, and tripled this amount up to 15k, this is an awesome increase of 200%. So back to engagement.
Back in 2010 some startups started talking about IPM (interactions per mille) as the way to go for measuring engagement for your social network page:

Knipsel1
This basically tells you the average chance of engagement per fan/follower. We’re surprised how many marketeers just blindly jumped into this metric and started using it as the holy truth about their social performance. And my argument is quite simple: who says that all my fans actually saw my content? And who says that the way Facebook and Twitter etc show my content to my fans/followers doesn’t change over time? The latter is a given fact, rendering this metric utterly useless.

Predictive metrics

So if we really want to measure engagement accurately then you only want to take into account a couple of things:

  1. How many unique people actually “engaged” or otherwise have taken an action based on your content
  2. How many unique people saw your content before engaging

And that’s how we ended up with IPV (interactions per unique view) which is measured on content level, but can easily be scaled up to a brand level serving as an average of all underlying content.

IPV

This is a really simple and accurate way to measure (social) engagement in it’s most pure form –> How many people end up taking a desired action based on my content, compared to the total amount of unique people who saw my content?

This is even more interesting if you combine it with the View Rate.

Vrc

View rate is especially helpful in Facebook, because publishing a message doesn’t necessarily mean that all your fans end up seeing it. And the view rate helps you to understand how many people of your total fan base saw your content in the end. And thus how well it performed in Facebook’s Edge Rank algorithm.

So far for metrics now. Shortly we will present part 2 of predictive metrics: our prediction widget.

Leave a Reply