The Data Analytics Blog

Our news and views relating to Data Analytics, Big Data, Machine Learning, and the world of Credit.

All Posts

Cash Vs Credit For Retailers: A Principa Case Study

February 17, 2020 at 9:00 AM

cash vs creditPrincipa’s Cash Vs Credit offering uses advanced analytics to help retailers answer a variety of questions. We have undergone such projects for many of South Africa’s leading retailers, and this case study describes the methodology Principa utilises to help retailers in the fashion industry.

The two big questions answered by a Cash Vs Credit solution are:

  1. How much more profit can be expected by converting cash customers to credit customers?
    Retailers are interested in exploring whether converting traditionally cash-only customers to store cards cannibalise their cash base with little return on the bottom line.
  2. What proportion of sales margin can the Financial Services department include in their profit calculation?
    A Financial Services Department includes interest, fees, insurance, etc. as their revenue streams, and excludes any margin from retail sales. Although this is in line with current industry practice, there has always been a general belief that having a credit offering available (which gives the customer additional convenience) does potentially encourage spend. Therefore, a retailer’s Financial Services department could include a percentage of the sales margin when calculating profit.

How do we do it?

A primary objective of a Cash Vs Credit project is to determine how much more a customer would spend if given a facility to enable spend, compared to having no facility. This helps to ascertain the financial benefit (from the business’s point of view) of having a credit vehicle available.

Since we cannot observe the same customer both having and not having a credit facility at the same time, Principa makes use of matched sampling and regression techniques. The matched sampling method enables us to find customers who are very similar across various profiles so that we can remove the effects of other “unwanted” variables and more accurately determine the impact of having a credit facility.

“…Principa makes use of matched sampling and regression techniques.”

Where there are no cash customers to match to credit customers for a particular profile, Principa makes use of a regression technique. This allows us to extrapolate to a cash customer’s spend for these profiles, based on what we know about cash customers close to these profiles. Using cash customers nearby ensures a more accurate spend prediction.

To conduct the matched sampling and regression, Principa follows four key steps, as outlined below:

  1. Principa takes demographic and bureau data, as at the observation date, relating to the retailer’s in-store card base and cash base. This is combined with spend data relating to the 12 months before the observation date to better understand the transactional behaviour of the customer. The population is then segmented into credit and cash segments on a customer-level.
  2. Using this data, we are able to determine the most important variables that distinguish credit customers from their cash counterparts. These variables are then used as the profile variables to find the matched samples.
  3. We want to ensure that the effect of having a credit facility takes the entire credit base into account. If the matched samples in the previous step do not capture the entire credit base, we use regressions to be able to predict what the cash spend would have been, had there been a cash customer with the same profile as the un-matched credit customer.
  4. After finding the extrapolated cash customer, we are able to match all credit customers to a cash customer and determine the additional spend of a customer when given a credit facility.

What have we learned?

“…the spend of a customer with an in-store card is likely to be between 1.5 and 3.8 times higher than a customer without the in-store card…”

Previous analysis has shown that once the differences between the profiles of cash and credit customers have been removed, and the only remaining difference is the credit facility itself, the spend of a customer who happens to have an in-store card is likely to be between 1.5 and 3.8 times higher than a customer without the in-store card in the 12 months following the observation date.

Knowing how much more a credit customer spends because of their access to the in-store card can help answer the questions initially posed. Unless the costs and risks associated with servicing an in-store card exceed the additional spend achieved from credit customers, bringing on more credit customers should have a positive impact on the bottom line. In addition to this, growing the credit customer base will also allow for a larger base for future product diversification, value-added products, etc.

Determining how much more a credit customer spends because they have the facility to do so also allows us to calculate the percentage of sales margin the Financial Services department could include in their profit equation. This gives a better reflection of financial results and allows retailers to better self-fund any further initiatives.

Megan Wilks
Megan Wilks
Megan Wilks is a consultant within the Decision Analytics team at Principa. Prior to joining in 2016, she worked in the consulting and data analytics field, covering a range of industries, including retail, credit and non-profit in Southern and East Africa.

Latest Posts

Shift happens: Top tips on Scorecard re-alignments

Principa employs a variety of best-practice credit scorecard building techniques including mathematical programming, regression modelling, optimal segmentation-seek genetic algorithms and reject inference parceling, amongst others. Through our credit risk scorecards businesses can look to improving their credit risk decisioning by 5-30%.

The Retaining and Reviving of Customers

What is the true value of losing a customer?

How Quick-Step Machine-Learning Models will help you through COVID-19

In a previous blog, we looked at assessing your credit models and the challenge of building and deploying models representative of the COVID-19 crisis. At the crux of the challenge was that the fact that: