User Engagement Drives Subscriptions - New RFV Engagement Scores from Deep.BI

The hottest topic over the past year was finding the holy grail of the subscription business. Many leading media companies have proven that the key user revenue driver is engagement. There are many ways to calculate this behavioral metric, and strangely, the most popular still seem to be "MAU" and "DAU" introduced years ago by Facebook. However, they’re not too actionable.

We, at Deep BI (inspired by the Financial Times success, which also promotes this way of measurement) recommend a different approach: the RFV engagement score - which is a combination of Recency, Frequency and (content consumption) Volume metrics.

Here’s why. First, a single score is like the North Star - easy to follow, easy to compare and easy to use. Second, every part of the combined RFV score tells us something we can react to actionably:

  • Recency - is the number of days since a user has used the product (or, in the month view, the average of the maximum number of days users have not been using the product). The higher the value, the more you should try to bring users back.

  • Frequency - is the number of days a user has been using the product. E.g. a monthly frequency of 15 means that the user used the product 15 days a month. This metric is critical in order to evaluate if a user has a habit of using the product. And the weaker the habit, the higher a user's churn propensity. The goal is to bring back users often until they learn a usage routine.

  • Volume - this should be the most important usage indicator. For news publishers it represents the number of articles read; for radios, it can mean time spent listening; for music services, number of songs played etc. This can be also a combination of different usage interactions. The key point here is to make sure that the product provides good value and users do not quit their sessions unsatisfied.

Leveraging our powerful data streaming and analytics technology, we have just released RFV metrics on the Deep BI platform. Our system calculates, in real-time, the engagement scores each time user interacts with a digital product (app, service, website etc.), and augments that interaction with current engagement metrics.

Then these scores can be used to:

  • Define custom engagement segments (e.g. heavy users, loyal users, fly-bys)

  • Define custom churn risk segments (loyal user has not used the app for 3 days)

  • Count the number of users in each segment

  • Calculate dynamics (flow) between segments (how many engaged users are becoming less engaged)

  • Find key engagement drivers. You can apply the RFV score to each attribute from user interaction data, for example: content types (genres, etc.) to find content consumed by the most engaged segments, locations to find cities or countries with the most loyal users, etc.

  • Intersecting engagement segments with other types of segments e.g. subscription products. In this way, you can immediately see who is not engaged among your paying subscribers, or how many trial subscribers are likely to subscribe because of high engagement scores.

Below are the main metrics tracked, which can give a good overview of the RFV score, as well as more in-depth insights related to it:

1. Global, average RFV scores Global RFV Score

2. Engagement segments vs. subscribers

Engagement Segments by Number of Subscribers

3. Number of engaged users over time

Number of Engaged Users Over Time

4. Engaged User Segments and those at Risk of Churning

Engaged User Segments and At Risk

5. The most attractive content categories for engaged users

Attractive Categories for Engaged Users

6. Days of the week with the most engaged users

Weekdays with Most Engaged Users

7. Cities with the biggest number of engaged users

Cities Most Engaged Users

The application of these engagement scores is the way to grow a paying, loyal user base. Some applications include re-engagement strategies using newsletters, push notifications, or ads, or even product improvement and a better recommendation system, among other methods.