Customer Churn: How To Limit Your Losses
April 15, 2018
Author: Avora Team
When does a paying customer become a churned customer? That’s not as easy a question as you might think. Do you consider a customer as having left after a week? A month? A quarter? Or is the metric in relation to previous buying behaviour?
For example, if customer A frequented your store once a week, but then fails to show up during the next month, they could well be classed a churned customer. Whereas for customer B, a visit once a month might be their normal shopping pattern. The RFM (Recency Frequency Monetary) model varies from business to business, but it’s essential to determine where losses are occurring and why.
Calculating your churn metric can be particularly difficult in the fashion retail sector where purchases may be dictated by all sorts of variables. Unlike in the supermarket sector, where weekly shops are the norm, buying patterns in the fashion retail market can fluctuate from season to season for the same customer. For this reason, being able to track individual customers is vital. Coupons, redemption codes, and loyalty cards can all help with customer tracking, while online it’s possible to track customers via their accounts.
However, some of these customers will complete checkout as ‘guest’ on a routine basis making it difficult to classify them. Assigning users a unique identifier the minute they land on the site – and maintaining this identifier even when they go off site to other franchises of the business – is therefore preferable, which is exactly where Avora’s TrueTag solution comes into its own.
The standard method of calculating churn is to segment your customer base to perform cohort analysis. For example, it’s a given that new customers will churn more quickly than old ones, so you might want to include a term in which to class customers as ‘stayers’ once they have remainder a customer for a certain period of time.
Focusing on the resident customer base at the beginning and end of a period is often used as a way to assess churn, but where this gets messy is when you have to factor in new customers that join during the period. The period in question can also throw up anomalies as it assumes that churn is equally spread over that timeframe. There’s plenty of interesting analysis available as to what is the best equation to use.
Deciding on how you will calculate your metrics matters and this is where detailed analysis of your data can add value, allowing the finance, marketing and sales departments to drill down and see cohort analysis not on just a quarterly or monthly basis, but down to the last day or even hour. That kind of analysis can then be fed to sales, providing targets for the number of new customers that need to be recruited. It can also help marketing focus on which customer segments needs attention to improve customer retention, and will also be useful to finance to provide a prediction of how much profit the company is on course to make.
Without good visibility of customer churn you can’t really see what’s going on in the business and you can’t delve deeper to find out why customers are churning. What’s the story behind the data? When customers churn is dependent on RFM which reveals the significance of churn. Avora can show the relationship between the two by helping interpret this data to reveal the data story.
By far the biggest cause of churn is poor customer service or the perception of poor service. There are now numerous channels that can be used to gauge the customer’s perception of the service they are getting, with social media channels such as Twitter providing a steady stream of data that can be analysed to determine who had a bad experience, where and when, and the causes (staff/product/returns process etc.). Address those complaints and the chances are you will have created a churned customer into a loyal one and maybe even a brand ambassador.
Other causes include problems with accessing information or completing a purchase through too many friction points; negative brand publicity; intense competition from a rival provider; and natural churn through customers moving out of the demographic/area you serve. Look at the time in the product lifecycle that they churned. Did they leave right after the free trial ended? Or was it at the end of a long subscription or membership? Did they routinely abandon their shopping cart at a key point in the process?
But this form of analysis is retrospective. Many of these customers will not come back, although it’s still worth investing in strategies that seek to reactivate dormant or lost customers. What you really need is the ability to predict and prevent churn. You need to know which customers are at risk of churning and can catch those before they churn. You can try to meet expectations, turn the relationship around and retain their business.
Preventing at risk or future customer churn is all about knowing your customers. You need to identify why existing customers have churned, what the telltale signs were to indicate churn and when these occurred, and what you can do to head-off their departure. And that requires predictive churn analysis and an analytics solution that can understand patterns over time using machine learning.
Predictive churn analytics can be used to analyse customer online behaviour and the likelihood to purchase, but it needs an evolving churn model so the data has to be instantaneous, in fast or near real-time. You’ll need in-depth customer profiles that provide individual data on frequency, dates and the value of purchases and the access method used as well as details on the products purchased. And you’ll need data on the interactions with that customer. How many times has the business reached out to them and over what touchpoints?
Avora is the only solution to offer online analytics developed from the ground up to capture customer transaction patterns. This is complemented by predictive analytics capabilities that utilise machine learning to present data that can be used to determine the likelihood of churn and identify which customers are at risk. This information that can then be used to take corrective actions. That may take the form of adjustments to the service offering, onboard process or customer service etc.
Armed with that information, the business is then able to invest resource where it is needed and with particular customer segments to improve perceptions and ultimately recover custom and reduce churn.