Recently, I came across multiple memes which joked about how easy it is to spend hours scrolling through TikTok and also about how dejected you feel when you realise you’ve wasted hours on social media. It got me thinking about how social media companies and their users have such opposing views of app engagement. Companies would see this increased session time as a positive engagement metric, whereas clearly the user doesn’t feel the same way. As a product person, I started thinking about alternative ways of defining engagement which would align these two points of view.

Most social engagement actions can be broken down into two types of engagement – consumption and contribution of content on the app. We can break down our set of active users into users that do either of these actions, as shown by this graphic. The outer ring are active users that neither consume nor contribute – these are users that have downloaded our app but aren’t really interested in it. The middle ring are consuming users that regularly consume content on our app but don’t contribute to it. The inner ring are contributing users that don’t just consume on our app but also contribute content to it.
It is normal for companies to optimise for engagement metrics at the high level, overlooking how this affects the different types of engagement. For example, metrics like Stickiness Ratio and Retention Rate are engagement metrics which companies are fixated on improving. Both of these are calculated using active users, which is a measure how many users are using the app. They do not give any consideration for what these users are actually doing in our app. Are they simply consuming content on it or are they putting in extra effort to create content, contribute to and grow our product.
engagement ~ consumption + contribution
The three graphs below show how optimising at this high level could results in very varying levels of consumption vs contribution since you don’t have visibility over which side of the line your users fall. Case 2, where the majority of users become consuming users, is the most likely to occur, since consuming requires much less time and effort from the user than contributing. This could lead to app content being created by a small proportion of contributors and the majority of users purely consuming on the app.

Optimising at the lower level, however, allows us to have control over which type of engagement we’re trying to improve and, in the process, improve our overall engagement. If we were to focus on improving our contribution metrics, naturally the overall engagement metrics would improve, but we’d also be able to tell which of the above scenarios has caused this improvement. We’d have more visibility over what proportion of our users are creating and sharing content, and thus feeling more fulfilled than if they were just consuming.
How do we actually do this?
We need to start of defining our engagement metrics from the contribution point of view. Wherever we use metrics defined using Active Users, we also define similar metrics using Consuming and Contributing Users. For example, when looking at Retention Rate, which is the number of users that were active X days after installation, we can look at Recurring Contribution Rate – the number of users that posted on the app X days after installation. Similarly, when looking at the Avg session length per DAU, we can look at Avg session length spent performing contributing actions vs consuming actions.
Whether we want to simply monitor these metrics or use them in your AB tests and ultimately shape our product strategy with them is dependent on the product and what we’re trying to achieve. If we were focused on building a product that encourages users to create, contribute and feel fulfilled in doing so, then it would make sense to incorporate these metrics into our experiments. We’d also want to monitor consumption metrics to understand the effect our new product strategy has our consuming users.
Caveats

- Higher volume of data at the higher level than the lower level. Generally, the more effort needed from users the fewer of them will do it. This concept is best explained by Dr BJ Fogg’s Behaviour Model. In our use case, we can consider "Prompts" to be different features of our app. Consumption features are more likely to trigger a user action because consuming is easier to do and requires less motivation than content creation and contribution. So we’re likely to see a higher volume of data for high level engagement and consumption metrics than we are for lower level contribution metrics. Higher volumes of data allow for more faster and more statistically sound experiments, making it easier to optimise for the high level and consumption metrics.
- Traditional monetisation strategies favour consumption. Most social Apps monetise using native ads. This favours consumption since the higher the number of consuming users, the more opportunities for them to see ads and the higher the revenue generated for the company. In order to truly prioritise contribution, we need to find alternative models that fit contribution better. We could potentially borrow monetisation models from gaming, or use subscription models that reward contribution, like Medium.
While the caveats above make it harder to optimise for contribution rather than consumption, if we were to make it work consistently, we’d be able to build a social product that is truly fulfilling and engaging for users.