The way users behave in your product can help you understand who they are and what value you are providing to them. A behavioral persona describes a distinct way of using the product and can help you discover which user behavior drives retention and growth.
The Product Analytics Playbook by Amplitude provides a clear example about three distinct behavioral personas for YouTube:
- Creators: the small percentage of people who actually create videos and post them on YouTube.
- Viewers: the vast majority of the site’s traffic; visitors who are simply watching videos.
- Viewers + Commenters: people who view videos and leave comments.
Each of these groups of users, or behavioral personas, are using YouTube in a distinct way and for a specific reason. Creators use it as a platform to upload their content on the web and build an audience, while Viewers and Commenters use it for entertainment or to follow Creators they like.
Why You Need to Find Your Behavioral Personas
There are essentially two main reasons why you want to identify your behavioral personas:
- Understand the nuances of your product that deliver value and drive long-term engagement. Different personas may have totally different retention rates, which you won’t be able to see if you look at retention for all your current users put together.
- By understanding which aspect of your products drive more enagement, you can optimize your product design and marketing messaging around those features to provide them with the best user experience, and in so doing increase the odds that more and more users will become a successful (highly retained) persona that finds value in your product.
Depending on the stage or size of your organization, you might decide to invest all resources onto a single behavioral persona / use case (e.g. a startup) or you might uncover a new use case you did not think about before (e.g. a bigger, more established organization with bigger teams and resources).
Historical examples of uncovering new use cases are Instagram and Twitter. Before being known to the global population as Instagram, this app was called Burbn and it was a location-based app allowing people to check in at locations, earn point and post pictures, until the founders started to study their app’s user behavior and discovered that while most features weren’t being used, there was one aspect of the app that was constantly used by a small group of users: posting and sharing photos. They decided to drop everything else and focus on making that feature fast and seemless -Instagram was born!
While Twitter was originally launched as a social network, almost immediately their founders realized that many people were using it for something else: customer support. People were twitting complaints or questions to companies, which in turn used the tweets to respond to their customers. Twitter understood this use case was so key that they decided to add Direct Messages to the traditional Tweet!
In both of these cases, the companies noticed that a subset of their users had a unique way of using their product and decided to make product changes to support (or completely focus on) that use case to improve the user experience.
How to Find Your Behavioral Personas
According to the Product Analytics Playbook by Amplitude there are two ways to unearth your behavioral personas:
- Qualitative methods: mainly brainstorming and current knowledge of your product can make you formulate hypotheses about who your personas are; which can be validated by user interviews and user testing. Qualitative methods tell you why people do certain things and what they use your product for (the context of your Web data).
- Quantitative methods: use them to get to know what your users do with your product; for example, user segmentation based on user properties (demographics) and event properties (e.g. how much they spend per order), or based on frequency of use (e.g. how often they order).
Who is Your Most Successful Persona?
In order to understand which persona / use case drives most enagement, and hence current users growth, you need to establish your retention baseline. In other words, you need to know what the retention rate of all users lumped together is. Then you can compare that against the retention rates of your behavioral personas and see if you spot any drastic difference in retention rates between different personas and against the baseline.
In Adobe Analytics, you can do this in several ways:
- Calculated retention metrics: based upon the metric I have previously defined on this blog.
- Cohort tables: by making use of the dimension to display behavioral cohorts on cohort tables rather than acquisition cohorts (which I have explained here).
- Freeform tables: by simply defining a segment per behavioral persona, segmenting out the Unique Visitor metric by each of the personas and breaking the metric down by day/week/month (depending on which retention cycle your product has).
Method 2 is essentially out-of-the-box, however is limited in that it only allows you to define a persona by a single dimension, while a persona’s definition might require a little more complex logic than that; morover, cohorts group people by a starting period, while here we want to analyze all users belonging to a persona over time, regardless of the cohort.
Method 3 is also out of the box, however, like method 2, it can only allow you to see retention curves for cohorts of users.
Method 1 is according to me the most overcomprehensive and allows you to look at user growth, not just retention. Plus, with this method you are free to define user segments as complex as you need to capture all usage nuances of a given persona. I will explain this method in the remainder of this post.
Calculated Retention Metrics
In my previous post How to Account for User Growth with Adobe Analytics in 2019, I have explained how to split monthly unique visitors (MUV) by New MUV, Retained MUV, and Resurrected MUV. Also, I have provided a way to capture Churned MUV.
Now, there are two more metrics that are needed to assess user retention and user growth over time; namely, the MUV Retention Rate and the MUV Quick Ratio. The latter, in particular, was introduced by Social Capital.
MUV Retention Rate
This metric simply measures the portion of last months visitors that returned this month. In other words, the month-on-month retention rate is defined as:
(1) MUV Retention Rate(t) = Retained MUV(t) / Total MUV(t-1)
Now, because it’s hard to calculate a metric that pulls data from different time periods, we need to work out a formula whose components naturally provide us with that temporal lag. We will do that via the following identity, which always stands:
(2) Total MUV (t-1) = Retained MUV(t) + Churned MUV (t)
Which says that the total number of unique visitors last month either came back (hence they are retained) this month or did not come back (hence they churned) this month. From which we derive that:
(3) Retained MUV(t) = Total MUV(t-1) – Churned MUV(t)
Which entails that identity (1) can be rewritten as:
(4) MUV Retention Rate(t) = (Total MUV(t-1) – Churned MUV(t)) / Total MUV(t-1)
(5) MUV Retention Rate(t) = 1 – Churned MUV(t) / Total MUV(t-1)
Which not only provides us with the formula we need to calculate the retention rate but also shows that retention rate and churn rate are complementary to each other.
However, given the sequential segmentation in Adobe Analytics only allows us to look ahead time-wise (never behind), we can only calculate churn for next month (i.e. at time t+1). Therefore, we need to substitute t-1 with t and t with t+1 in identify (5) and then we obtain the actual formula for which all ingredients are in place:
(6) MUV Retention Rate(t+1) = 1 – Churned MUV(t+1) / Total MUV(t)
In fact Total MUV(t) is just the Unique Visitors metric, while Churned MUV(t+1) was previously calculated (see
How to Account for User Growth with Adobe Analytics in 2019).
MUV Quick Ratio
The quick ratio for the time period t provides the ratio of incoming users to outgoing users during that period. In other words, it’s computed as:
MUV Quick Ratio(t) = (New MUV(t) + Resurrected MUV(t)) / Churned MUV(t+1)
Here too we have all formula’s factors already previously computed.
This metric tells us whether we are growing our current (retained) user base. A ratio = 1 says that we are loosing next month as many users as we acquired this month (current user base remains constant); a ratio < 1 indicates that we are loosing next month more users than we acquired this month (we are shrinking our current user base); a ratio > 1 says that we are experiencing real growth since we are loosing next month less users than we acquired this month!
With these two metrics and with the ones previously defined, we are now able to look at the MUV growth profiles for both all users lumped together (the baseline user growth) and the MUV growth for the different behavioral personas. The charts that follow aim exactly at that; the left-hand side showing the MUV for a behavioral persona (i.e. a Visitor-contained segment) splitted into its constituents metrics and the right-hand side showing the retention rate and the quick ratio for each persona.
A behavioral persona is nothing more than a segment in Adobe Analytics. Whether you want to define it at Hit, Visit, or Visitor level, it is completely up to you. Each has pros and cons. For instance, if you use a Visitor container, you will capture all users that have done a certain thing at least once over a period of time you will want to analyse. This might be a good option if you want to capture user behavior that you are not expecting to happen more than once (e.g. making an account; or, having completed the onboarding video). If you are after a aha moment for your app or website, this might be a good approach. Hit containers might be suitable in case you want to constrain a given usage pattern to take place over and over on each and every day/week/month you are analysing retention for (e.g. listening to music for a music app). The following paragraph presents retention and growth for an example app with two behavioral personas.
Retention and growth outlook for all unique visitors lumped together (no segment applied):
Persona 1 Retention
Retention and growth outlook for the visitor segment defining persona 1:
Persona 2 Retention
Retention and growth outlook for the visitor segment defining persona 2:
The charts above reveal that Persona 1 has a higher retention rate than the baseline (about 10% higher) and Persona 2 has almost double the retention rate of the baseline and has experienced periods of real growth of its current user base (quick ratio > 1 for most of the time, with a peak of 2) as opposed to all current users, which have remained pretty stable over time (fluctuating between 0.8 and 1.2). This is also very clearly indicated by the purple bars of Persona 2, which represent its Retained MUV (i.e. its current users), and which completely overshadow the blue and turquoise bars; that is, the incoming users (New MUV and Resurrected MUV), while the churn of this persona is relatively low (orange bars).
All Personas Retention
Putting it all together, an easy and quick way to compare personas is by plotting their retention rates in one single chart, like in the following example:
It is quite clear from this chart that while all identified personas have a higher retention rate than the baseline (green bar), persona 4 and 6 are the most retained, hence the most successful personas for this product.
Behavorial personas capture all distinct ways in which your product can be used. They are a phenomenal tool to understand what user behavior and which use cases drive retention and growth. They provide you with the product analytics and insights you need to grow your current user base; informing your decisions as to which features to invest upon and which value proposition you want to promote moving forward.