AI News, Predicting Churn without Machine Learning

Predicting Churn without Machine Learning

In a business context, churn or churn rate refers to the number of customers leaving your business.

In this post I will describe a way of predicting churn based on customers' inactivity profile that I've applied in various client engagements.

Before we start predicting churn, we need to track our customers' activity over time.

Once we decide on which definition to use, we can apply it to calculate a binary value (1 or 0) for each month: If in month x, customer is active, then set activity value to one, otherwise leave as zero.

I've typically used the total monthly ARPU as the definition for activity, but this largely depends on the context of your analysis.

Now that we have a monthly activity flag for each customer, we need to use this to build an inactivity profile.

Now that we have our inactivity profile, we can use this to produce a monthly view of the number of customers for each level of consecutive inactive months within our window of analysis.

Each value represents the number of customers in each month that have been inactive for x consecutive months.

Using this data, we can compute the percentage of customers consecutively inactive for each month in our window of analysis.

The highlighted value (74%) represents the percentage of customers in February that have been inactive for two consecutive months.

We want to get the most representative values, so we take the average of the last four months in our window of analysis.

So let's see what these two terms (numerator and denominator) are, and then compute them for each of the inactive months to give us the probabilities we want...

Here n is a limit on the consecutive months of inactivity where we've noticied that the probability has stabilised.

For simplicity, we're going to assume that being inactive in consecutive months are independent events.

Now we have worked our the numerator and denominator for each month of inactiviy we can finally calculate the churn probabilities (yay!).

Note that the probability of churn for active customers (zero inactive months), is simply the multiplication P(1) x P(2) x P(3) x ...

Also, we've been analysing churn in months, but it would be useful to have a more granular view, weekly or even daily.

If you've enjoyed this article, or have any questions or feedback (or want to offer me lots of money to solve your business or data problems), feel free to get in touch: jmsmistral@gmail.com

Predicting Churn without Machine Learning

In a business context, churn or churn rate refers to the number of customers leaving your business.

In this post I will describe a way of predicting churn based on customers' inactivity profile that I've applied in various client engagements.

Before we start predicting churn, we need to track our customers' activity over time.

Once we decide on which definition to use, we can apply it to calculate a binary value (1 or 0) for each month: If in month x, customer is active, then set activity value to one, otherwise leave as zero.

I've typically used the total monthly ARPU as the definition for activity, but this largely depends on the context of your analysis.

Now that we have a monthly activity flag for each customer, we need to use this to build an inactivity profile.

Now that we have our inactivity profile, we can use this to produce a monthly view of the number of customers for each level of consecutive inactive months within our window of analysis.

Each value represents the number of customers in each month that have been inactive for x consecutive months.

Using this data, we can compute the percentage of customers consecutively inactive for each month in our window of analysis.

The highlighted value (74%) represents the percentage of customers in February that have been inactive for two consecutive months.

We want to get the most representative values, so we take the average of the last four months in our window of analysis.

So let's see what these two terms (numerator and denominator) are, and then compute them for each of the inactive months to give us the probabilities we want...

Here n is a limit on the consecutive months of inactivity where we've noticied that the probability has stabilised.

For simplicity, we're going to assume that being inactive in consecutive months are independent events.

Now we have worked our the numerator and denominator for each month of inactiviy we can finally calculate the churn probabilities (yay!).

Note that the probability of churn for active customers (zero inactive months), is simply the multiplication P(1) x P(2) x P(3) x ...

Also, we've been analysing churn in months, but it would be useful to have a more granular view, weekly or even daily.

If you've enjoyed this article, or have any questions or feedback (or want to offer me lots of money to solve your business or data problems), feel free to get in touch: jmsmistral@gmail.com

Maximize User Retention

In this Project, you’ll learn how to forecast the likelihood of user churn and how to proactively communicate with those users to keep them engaged before they churn.

No matter the app size, category, or business model, retaining app users is a big problem...but also an opportunity.

The market for app downloads is additionally characterized by cut-rate switching costs -- free-to-download apps and ever-decreasing download times are among the most prominent drivers of shrinking barriers to install.

And even with a robust, data-driven marketing solution, churn can be a particularly tricky problem to diagnose and treat because it trumps the logic of descriptive analytics: once you’ve observed users to churn, it’s already too late to save them.

The following steps will walk you through creating and analyzing predictive insights to identify users at risk of churn and their associated behaviors and attributes.

Generally churn Predictions will forecast whether users will not behave in some way while conversion Predictions will forecast whether users will behave in some way.

For both churn and conversion Predictions, specific Events can be added or removed to detail the behavior you want to forecast for your users.

For churn Predictions only, you can define the number of consecutive days that the Event(s) will not occur in order for a user to churn.

We recommend naming your Prediction by describing the type of user behavior you intend to forecast.

Related Behaviors: Related Behaviors are the usage patterns that serve as lead indicators of future behavior like churn or conversion.

Facebook’s former head of growth discussed how important it was for them to discover their keystone metric: Getting any individual user to add 7+ friends in their first 10 days.

For each attribute, you will see the proportion of active users observed to have each attribute, and the relative difference in churn or conversion between users observed to have that attribute vs.

The Hour of Honor | Critical Role | Campaign 2, Episode 24

Thanks to D&D Beyond for sponsoring this episode! Pre-order Waterdeep: Dragon Heist here: Pre-order ..

How Credit Cards Affect Credit Score: Utilization

There's lots of myths about how credit card utilization can affect your credit score. Should you leave a small balance on your credit card every month? Should ...

ОТ АТЕИСТА К СВЯТОСТИ

Ответы на главные вопросы атеистов, верующих, священников, людей, стремящихся обрести путь истинного бессм...

Proverbs While you Sleep (FULL NKJV CLEAR AUDIO with Binaural Beats)

This voice you hear speaking the Proverbs of God's Word is a voice connected to a passionate heart for God's Word. "LIKE, SUBSCRIBE, COMMENT, AND ...

Ellen's Helping Out with Homework!

Is there anything she can't do? Ellen offered to help her viewers with their homework. This is how it turned out!

Engine stop start systems: the truth about fuel & emissions | Auto Expert John Cadogan | Australia

Welcome to another 'What the FAQ' - where questions and comments, sometimes, as in this case, completely without rational thought, need the briefest ...

I Made $106 From $297 Spend On A Credit Card (Fidelity Card)

From spending only $297 on my Fidelity Visa Signature Card I was able to make $106 delivered straight to my Fidelity Roth IRA. Current Reading ...

Cost savings with Azure Virtual Machines - BRK3178

In this session, you will learn about new and existing options in the Azure Virtual Machines and VM Scale Set services to save costs. We will explore the best ...

Revelation Flannery O'Connor (Audiobook)

A reading of the short story "Revelation" written by Flannery O'Connor and read by Michael DuBon. Full Text: ...

Thyroid Problems | Foods To Eat and Foods to Avoid #EveryDayTips

Hi Friends, Watch More Top Five Health Care Videos Click here : Digesting these vegetables can block the thyroid's ability to absorb iodine, ..