Alternative Payments and Credit Card Cannibalism

Prior to the advent of Alternative Payment Methods (APMs), consumers paid for online purchases with credit cards.  Just like consumers are likely to have multiple credit card brands, many of these same consumers are now also likely to have at least one APM in their payment arsenal.  When prompted to pay for a product a consumer may decide to use an APM as opposed to a credit card.  In this case, the APM represents a surrogate payment type and clearly does not result in an incremental sale.  In payments jargon, we say that the APM cannibalized the credit card sale.  On its face, this seems harmless, however we will learn further on that this behavior can be harmful to merchants.

How do you determine whether you company is enjoying real uplift due to the addition of an APM?  Well, there are two methods.  One way is simple due diligence; ask colleagues that use a particular APM about their results.  You should also ask the service provider to provide references.  In this latter case, you should ask the reference to provide analytics to support the results.

The second method is to actually field test the APM and use regression analysis to determine the level of cannibalism.  Merchants should perform a trial, offering the APM to the smallest percentage of their customer volume that will provide statistically meaningful data.  This will minimize financial harm if the APM turns out to be detrimental.  This subset, of course, must be tagged or “isolated” from the entire customer base such that they can be easily identified. The merchant should then compare the relative percentages of payment types in the test against historical payment type percentages.  The following table illustrates a range of results for such a test demonstrating an increasing APM success rate as you move to the right.

 

Historical

Case 1

Case 2

Case 3

Visa

60%

55%

58%

60%

MasterCard

40%

35%

38%

40%

APM

0%

10%

10%

10%

Comparative

100%

100%

106%

110%

Cannablism

 

10%

4%

0%

Uplift

 

0%

6%

10%

Historically, this merchant accepted only Visa® and MasterCard® for payment, so these forms represented 100% of sales.  Case 1 demonstrates a case of pure cannibalism, where the merchant saw no comparative increase in sales, while the APM cannibalized  of the credit card sales.  Case 2 demonstrates a 0% increase in sales while cannibalizing a 10% of credit card activity.  Case 3, ostensibly the most successful trial, demonstrates an uplift of 6% with no apparent cannibalism.


Credit Card Cannibalism, Recurring Payments, and Lifetime Value

Cannibalism has some fairly interesting effects outside of the simple displacement of competing payment types.  In particular, cannibalism can affect the lifetime value of customers.  Nowhere is this more profound than in businesses using subscriptions with recurring payments.  Companies operating subscription programs typically measure their customer retention rates (CRR) very closely.

One cancellation indication can be found on the merchant’s payment processing report in the form of refunds.  Refunds generally reflect some type of billing error or a pro-rata refund to compensate the consumer for the unused portion of the service period during cancellation.  In a well run company, billing errors are rare, so most of these refunds are owed to cancellations.  Not all cancellations, however, involve refunds.  Still, the refund rate can be used as a proportional indicator for the cancelation rate.

It is important for merchants to look at the refund rate – and not just the cancellation rate – because refunds are directly related to the payment type.  Refund rates for APMs can be significantly different than those for credit cards.  They can be lower or higher, the characteristics generally driven by the APM’s features, average product value, and customer demographics.  The important point is that APMs that exhibit relatively high refund rates decrease the lifetime value of recurring payment customers.  This can be measured directly by examining retention rates.  The following chart extends our example above by including the average monthly refund rates for credit cards and the three APM cases.

 

Historical

Case 1

Case 2

Case 3

Visa

60%

55%

58%

60%

MasterCard

40%

35%

38%

40%

APM

0%

10%

10%

10%

Comparative

100%

100%

106%

110%

Cannablism

 

10%

4%

0%

Uplift

 

0%

6%

10%

Refund %

1.0%

5.0%

5.0%

5.0%

Customers

1,000

1,000

1,000

1,000

Cancellations/M

10.0

14.0

11.6

10.0

Agg Refund %

1.0%

1.4%

1.2%

1.0%

12 Month CRR

89%

84%

87%

89%

24 Month CRR

79%

71%

76%

79%

36 Month CRR

69.6%

60.2%

65.7%

69.6%

In this example, we assume an isolation case where we are looking at 1,000 existing customers while excluding new customers.  We then study the retention rates over a 36 month period.  We see that the company’s traditional credit card refund rate was Refund %.  The refund rate for this APM, however, turns out to be significantly higher at 1.0%.  In cases with no cannibalization, we intuitively know that all billings are incremental, and the high refund rate is simply a cost of doing business – but, we are still way ahead of the game.

In cases with cannibalization, we see that the aggregate refund rate is greater than the historical credit card rate, and this of course, means that more people are cancelling their subscriptions.  An historical 1% refund rate resulted in a 36 Month CRR CRR after 36 months.  In Case 1, our worst cannibalization case, a 1.0% base refund rate translated into an aggregate refund rate of 1.0% (or 10.0 cancellations per 1,000 customers).  Over a 36 month period, this results in a 69.6% CRR, significantly lower than the credit card base case.  The monetary cost of this sizable  difference will be based on the mean customer lifetime, the average ticket value, and other factors.  This metric, and other related measurements, may be easily calculated using any of the free Lifetime Customer Value (LCV) calculators available on the Web.  An excellent LCV Calculator in excel format is available from the Harvard Business School at: http://hbswk.hbs.edu/archive/1436.html.

It should be noted, that this same analysis can be performed using “disputes” (or “chargebacks”) instead of refunds as the principal metric.  Generally this is not desirable for two reasons.  First, the number of disputes under most payment types is usually lower than the number of refunds, so there is less data to work with.  Also, the dispute processes for different APMs varies widely.  Nonetheless, if a merchant has sufficiently large volume, then an analysis including both refunds and disputes could be useful.