Revisiting the Engagement Pyramid

By Angela, January 24, 2017

Back in 2014, I blogged about building a user engagement pyramid for social platforms, inspired by Fred Wilson’s 100/10/1 rule.

The premise of that post was simple: since user-generated content is king, the top of the pyramid (aka the most engaged users) are those who post status updates, photos and links. And, these creators are inspired when they are recognized for their contributions via shares, comments, and favourites. As such, the user engagement pyramid should look something like below and the goal of social networks is to move users up the pyramid.

In that original post, I offered two non-mathematical steps for creating an engagement pyramid:

  1. At the top, place the activity that you determine to be a sign of someone who is most engaged with your product (this is usually the factor that drives your most important metrics).
  2. Next, rank the activities in decreasing order based on level of friction (from hardest to easiest).

A few months ago, one of our founders reflected on this advice, “Conceptually, this is all good but is there a ‘scientific’ method of building this hierarchy of engagement?” As a lover of all things data, these comments inspired me to revisit the post and see if I could add a more solid mathematical framework to the pyramid.  

Upon further reflection, I think the first point about placing your most engaged activity up top is still valid. However, for point 2, I think that friction (as measured by how hard/easy an activity is) may not be the most important factor for determining engagement. Here’s why:

Individual vs. Systematic Engagement

Let’s consider the following: is the user who logs on everyday and simply lurks more or less engaged than the user who only logs in once that month but posts content?

On a purely DAU/MAU basis, the former is more engaged than the latter. However, there is a difference between the individual’s engagement and “systematic” engagement (how the individual’s actions influence all other actions of the platform). The pyramid above represents the latter because posting content, in theory, will attract more people to the platform than just lurking or liking. In other words, posting has higher systematic engagement than logging in.

It’s a nuanced distinction, but important to understand.

Can we create an index for engagement?

If we revisit point #1, there’s some low-hanging fruit to help us figure out individual engagement:

  • Of those users with the highest DAU/MAU, what are actions are they doing?
  • On the flip side, of those users who post or comment frequently, what is their DAU/MAU?

These two exercises should inform what the systematic engagement pyramid looks like. Most likely, you already have a strong intuition of what action should be on top and this exercise is just a matter of testing your hypothesis.

Here’s an example. Let’s say you believe that posting an image engages the most users. How do you test this to be true? You’ll need to take a step back and define the influence or reach of a given action.

In this case, the engagement “index” of an image can be computed as:

# of likes + # of comments + # of shares + # of views

Likewise, the engagement index of a comment is:

# likes + # of views

Simply put, think of the pyramid as ranking the potential reach of all users. Every step down the pyramid has a smaller reach. Based on the above, the reach of an image is greater than the reach of a comment. You can assign value to every possible user action (posting a status update, sharing a link, etc.) and then rank the actions according to total potential value.

Keep in mind, that your index will be more complex, as I made a lot of assumptions on linearity for simplicity’s sake.

Key takeaways

What can we take away from all this? First, systematic and individual engagement are two different things: systematic engagement is a function of all individual engagement combined. It’s important to understand this distinction when building your engagement pyramid. Second, for as much as we try to quantify engagement, this process is still very much more art than science.   

And please, keep the comments coming. Each discussion further refines the model and helps us all.

  • Kelly Kuhn-Wallace

    Consider modeling the gross impact of the content posted by the social platform “account owner.” This inquiry translates across platforms (a necessity in 2017) and will provide critical feedback on that content.

  • atkingyens

    Good suggestion! Thanks, Kelly!

  • Hi Angela,

    Thanks for sharing your thoughts – insightful as always. 🙂

    I’m curious why you would limit engagement to a linear model. If you look at other domains, e-commerce for example, one of the most common models is the recency-frequency-monetary (RFM) model of engagement. That’s a 3d model of engagement.

    You would then create a ranking function as a weighted average of the dimensions, depending on the use case. You would have a ranking function for brand loyalty, a ranking function for the big shopper who comes in once in a while and so on.


  • atkingyens

    Hi Ganesh! Great feedback. The intent was definitely not to limit it to a linear model at all; it was to illustrate the concept in a simple manner that everyone can absorb quickly (so linear seemed the easiest).

    But yes, I totally agree with you – there are many more sophisticated models that you can apply that weight different actions, etc. I might just have to get my hands on some data, test a methodology, and then blog about it. So thanks for the inspiration, and for also pointing me to RFM 🙂

  • Alex de Bold

    I think you’re on the right path and it’s not easy. We’ve lead over 200+ influencer campaigns for tier 1
    brands, tracked over 400 million impressions and tons of engagement via our HEFT platform.

    # of likes + Views is easiest as it relates closest to the paid media industry (clicks & impressions)

    Another aspect to engagement, equally if not more important, is the economic value of engagement relative to paid media. It’s the 1% of people who generate the content that people engage with vs the .01% of people who accidentally click on a banner ad responsible for billions in ad revenue. Happy to walk you through our thinking if you’d like and data sets.

  • atkingyens

    Yes – would love to chat! Any chance you can email me directly – angela at versionone dot vc? Look forward to it 🙂

Subscribe to our Blog via Email