Win Back Lost Customers

Pursuing defecting customers might seem too expensive, but the data says otherwise.

• Keeping an existing customer costs one-fifth of what it costs to acquire a new customer. (Lee Resources 2010)
• It is 600% to 700% more profitable to win back past customers than to acquire new customers (CustomerSure)
• Marketing Metrics reported these probabilities of selling to three types of customer:
• Existing customer – 60%-70%
• New prospect – 5%-20%
• Lost customer — 20-40%

All three types of campaigns are important, but the probability of recapturing a lost customer is much higher than that of acquiring a new customer, and it is also more profitable.

In other words, your “lost” customer database is extremely valuable.

Why customers you win back are likely to be very profitable

Why are customers you win back so much more profitable than new customers?

• They already have a relationship with your brand and don’t need to be targeted for extensive awareness campaigns
• You already have collected a significant amount of data about them

• They also have a demonstrated need or desire for your offerings

All of this information can be used to personalize effective Win-Back offers. AI enables an even deeper level of segmentation for even better personalization and more profitable results.

Not all customers are equal, and not all are worth winning back.

If you run a Win-Back campaign for everyone who has churned (#3 above), you’re likely to:

• Waste money on customers who are very unlikely to come back
• Reactivate customers who are not very profitable

• Attract customers who will churn again as soon as they’ve taken advantage of your offer

For your Win-Back campaigns to be successful, you need to know which churned customers hold the highest potential value and which have little or no value. AI based analysis can identify them for you.

01 How long should you wait before adding a customer to your Win-Back campaign?

There is no one-size-fits-all answer for this question. Each business is different. Let AI determine at what point after their last purchase you can assume that a customer has stopped buying from your business.

02 How long should you wait before adding a customer to your Win-Back campaign?

This question begs several other questions:

• How long will a reactivated customer stay once they are won back?
• How difficult will it be to win them back?

• How much will they spend in their second lifetime?

Generally, a customer that remains for a long time, even if they don’t spend a lot, will be more profitable than one who spends a lot but defects after a short period of time. Also, the easiest customer to win back is the one who left over price.

So, the customer that left because of price who will remain a customer for a long period of time will likely be your most profitable second-lifetime customer. But for all the customers you win back, a comparison of their first-lifetime and second-lifetime value will illustrate how important your Win-Back campaign is.

03 Will your churned customers consider it a big risk to return?

Depending on the circumstances of their departure, a customer might very well think that your Win-Back offer isn’t worth the risk of doing business with your company again. This is especially true if their departure was prompted by an unresolved service issue or what they considered to be poor treatment from someone in your organization.

04 How will you track and test your Win-Back campaign?

Without tracking and testing your Win-Back campaign, you won’t be able to measure its profitability or know how to improve on it.
 

Without tracking and testing your Win-Back campaign, you won’t be able to measure its profitability or know how to improve on it.

For testing, establish a control group of records for each different cluster you intend to target (see more below). Test alternative messaging against the control group to determine which is most effective. The longer you run your campaign, the more Machine Learning will help it be more effective.

To know which customers to win back, and how best to do that, you need a complete view of your churned customers. This is a very time-consuming and painstaking exercise if you attempt to create this view manually. AI is much faster and more thorough than any database analyst can be on their own.

Take a multi-dimensional approach to churn

For any given time period, the churn rate can be calculated in several ways:

• Total number of customers lost
• Percentage of customers lost

• Total value of recurring business lost
• Percentage of recurring value lost

This multi-dimensional approach gives you a fuller perspective of the problem you’re having with defecting customers.

Segment by life-cycle stage

Artificial Intelligence can analyze customer behavior and automatically segment your customer base by life-cycle stage in real time, without the need of manual intervention. Life-cycle stages could include:

• Existing customers (first life cycle)
• At risk of leaving

• Churned (no longer buying)
• Reactivated customers (second life cycle)

This high-level segmentation is critical to maintaining an understanding of your churn rate
and the effectiveness of your Win-Back campaigns.

Segment by why your customers left

When customers stop using your business, it can be for a number of reasons, including:

• The product or service was too expensive for the customer
• Lacked value — wasn’t worth what it cost
• Product didn’t function properly
• Moved out of your service area

• Service issues
• Prefers a competitor brand
• No longer needs your product or service

This is very valuable information. Each of these reasons helps determine which customers are likely to return, and which are not. Price and service issues are usually the easiest to address. Customers that have moved out of your service area or no longer need your product have a very low probability of returning.

Not all companies collect information about why customers defect, but they should. No business will achieve a 100 percent response level to its surveys, but Machine Learning, a subset of AI, can study customer behavior patterns and accurately attribute defection reasons for many churned customers beyond those who responded.

Customers who complained before leaving is another measurable factor to be explored. The information should be captured in your CRM. Are complaining customers more or less valuable than those who do not complain? Were some complaining customers more valuable than others, and does the reason for their complaint give an indication as to their value?

Segment churned customers by various measurements of their value

From the data you already have about your customers, you should have at your disposal:

• How frequently they purchased
• How recently they purchased
• The value of their average order
• The total value of all their purchases

• The products and/or services they purchase from your company
• How long they were a customer before they defected
• How many people they referred (if any), and the value of those referrals

By examining each of these attributes, AI determines which customers were most valuable as new customers.

AI uses all of the information in the 360° customer view to create clusters of customers who are similar in terms of how valuable they were, what they bought and why they left. From there, AI calculates how worthwhile it is to pursue each cluster and what your Win-Back offer should be.

Now it should be easy to understand why it makes more sense to do this work with AI and Machine Learning than it does to do it manually. And the 

The Win-Back programs of companies that don’t have AI at their disposal look something like this:

Obviously, that’s a lot fewer steps than using AI/Machine Learning to create a 360° view of each customer, but that’s where the comparison ends.

 

Here are examples of how AI/Machine Learning outperformed earlier reactivation campaigns in the non-profit sector that relied heavily on RFM alone.

• Out of a pool of 170,000 donors that previous campaigns had failed to reactivate, Machine Learning identified 25,000 prospects. Mailing to those prospects generated more than $250,000 in net revenue in 18 months.
• The response rate for the 25,000 prospects identified above was 1.60%, three times the response rate from cold acquisition lists.
• For one organization that was skeptical of the value of AI, Machine Learning algorithms produced major lifts in every segment, including higher than 50% for many segments

Why does this work so well?

AI/Machine Learning identifies exactly:

• Which lapsed customers to target and which to ignore
• What to offer them
• How to personalize the offer

When dealing with databases of 5,000 records or more, AI is an absolute necessity for any serious Win-Back campaign.

Once Machine Learning becomes part of your Win-Back program, it gets better and better at predicting which active customers who are in danger of becoming lost customers. We discuss that more in an article on predicting customer or prospect behavior.