Best Practice for Analysing Customer Retention

Good, long term relationships with customers are vital for the vast majority of businesses, but the art of customer retention sometimes risks taking a back seat to “just bringing in the sales” online.

Introduction

From a web analytics perspective, success in retention can seem a struggle to benchmark compared to the black and white world of customer acquisition. Two common challenges are:

1)The most valuable customer retention data is spread across CRM (Customer Relationship Management), back office and web analytics systems, and often some form of data integration is required to get the best insight out.

2)The metrics themselves are harder to define and interpret. How much retention is “good” retention? What period do you measure customer churn over?

No two businesses are exactly the same, so in considering retention we need to look at the best areas to focus on and how we can use analytics as a driver for positive change in each.

Loyalty

Loyalty is an obvious trait that good long term customers exhibit. So how can we benchmark it?

Active loyalty tends to imply continued transactional behaviour, and we may have the choice of benchmarking purchase frequency with our web analytics tool and or CRM systems. CRM systems tend to have a more reliable view of purchase frequency as orders are tied against names and addresses rather than cookies (and deleting cookies or switching PC is much easier than moving house!).

Visit frequency is a sound measure of passive loyalty/long term interest, particularly for subscription services or sites with high value purchases that will have a long research cycle.

Here we will tend to focus on our web analytics tool, and need to bear in mind the skew from users switching devices or clearing out their cookies. Both will lead to an understatement of repeat visit frequency as users will appear as “new visits” whenever they switch to a new device or delete their cookies - this effectively resets the frequency clock.

For sites that have registered users, using a registration ID rather than a cookie will increase the accuracy of results (obviously only for registered users that bother to sign-in, so there needs to be an incentive to do so to get exclusive/tailored content).

When interpreting frequency trends, don't focus on an average figure - this is prone to skew from outliers (e.g. one user that incessantly visits the site every hour). Some sensible buckets will give far more insight into behavioural patterns.

Lifetime Value

Loyalty is all very well, but what is it worth to us in monetary terms?

Lifetime value (LTV) benchmarks the long term profitability of a customer, either from their historical transactions or a predictive forecast of their value based on prior behaviour. For example, some gambling sites can take a pretty reliable statistical stab at the LTV of a customer based on the amount they put down as their first deposit.

If we are using the historical approach, we effectively need to glue together multiple purchases to calculate LTV, and often the CRM or back office systems are the best sources for this. But LTV figures in themselves are of little use unless we can tie them back to online behaviour and discover what drives (or more importantly prevents) retention. This means exchanging data (e.g. customer or order references) between web analytics and CRM systems to glue the data together.

It's possible to do LTV within the web analytics sandbox, but this is often too skewed by cookie deletion/device switching to be of any use if we are looking at a long term picture (e.g. value over the course of a year).

A more pragmatic measure can be to approximate LTV with a short term measure of spend. For example, “Average Revenue per Visitor per Month” for a supermarket or “Average Revenue per Visitor per Quarter” for a book retailer. This smooths out the issue that the best customers may be those that make multiple modest purchases over time rather than a one-off splash.

LTV data is best applied as segments: we want to understand what our most valuable customers respond to and use that to expand the size of those higher value segments.

Acquisition Source

But we're talking about retention, right?

Acquisition and retention are inextricably linked, and the customers that are easiest to acquire are not necessarily the easiest to retain.

Consider a campaign focused on a cheap product on special offer - it may well bring in new customers very cheaply, but is only an effective loss-leader if those customers can be retained and monetised over time.

The easiest route to evaluating this picture is to take LTV segments (e.g. the best and worst spenders) and link them back to campaign source. This may change your long term view of the most profitable areas to invest in.

The data challenge here is linking LTV to source, normally by using a common customer identifier between your web analytics data and CRM systems. You can then import the customer identifiers associated with a particular LTV segment into your web analytics environment and profile your data based on this.

Engagement Metrics

Retention might ultimately be about repeat custom, but getting customers to keep coming back may mean more than just punting offers at them constantly.

Engagement with content or community is typically the active means of driving retention - if we can maintain interest across the sales cycle the chances of competitors jumping in are reduced.

The value of content has traditionally been hard to quantify, but when we start to link engagement with content to purchase behaviour it's much easier to evaluate (and hence optimise).

Up-lift in purchase frequency or LTV is a good ultimate KPI for engagement - do visitors that view or post to a blog end up spending more money over time, or is it a wasted investment? Traditionally web analytics tools have lacked the ability to glue together behaviour over time, but a customer-focused approach to data over time can start to unlock these vital insights.

A Note on Conversion Rate

Conversion rate is a headline figure in a lot of web analytics reports, but it's worth bearing in mind that it can present a very muddy view of what is going on.

Retention is about the long term picture - value over time, not in the course of a single visit. An effective retention strategy can actually lower the short-term conversion rate, with users visiting the site more frequently when they are in non-purchase mode.

For retention purposes, customer churn and LTV are better indicators of what is going on.

Summary

Customer retention is a key focus for analytics, and there are a variety of insights that can help us maintain good relationships with our customers online and maximise long term profitability.

Loyalty metrics are a good benchmark of how well our retention strategy is working in the immediate term. Lifetime value is key to evaluating the big picture and understanding who our best customers really are and where they came from. Understanding engagement will help us optimise that picture and keep our best customers coming back for more.

Comments? Please leave them on our Twitter stream.