Measuring Return on Loyalty
Why guess when you can know...
Investments in customer loyalty are booming. Marketing teams are spending big on rewards, partnerships, agencies, data capabilities and marketing technology. Loyalty has become table stakes in most consumer facing sectors. Having a loyalty program is becoming a cost of doing business, a hygiene factor. 35% 1 of all credit card purchases in Australia earn Qantas points, 25%2 of all credit card purchases in New Zealand are earning Airpoints. Myer claim 68%3 of their sales are linked to the Myer One program and even life insurance companies are on the points bandwagon, with AIA Australia4 launching a points based program and Qantas following suit with their own version call Assure5. Australian companies are spending more than $5 Billion dollars6 a year in loyalty points and rewards. This number is dwarfed by the $9 Billion 7 in the UK and over $52 Billion8 in the USA… Companies are investing significant amounts of money into customer facing programs and activities…but what are we trying to achieve?
- Loyalty is a long-term phenomenon; so the techniques we use to measure campaign returns – where we hold-out a control group, attribute conversions to the campaign, measure sales lift against the control and report incremental margin achieved versus costs, to arrive at ROI for the campaign – are problematic; How do we apply these disciplines to multi-year customer retention targets? Especially if a loyalty program is part of your strategy. Customers do not react well to being told “you cannot join this brilliant reward program because we need a control group and you’re in it”.
- Self-selection bias; this is the big loyalty program measurement challenge. It is a fact that the customers most likely to enrol in your program are those already most engaged with your brand, products or services. This makes sense. They know they are going to benefit because they intend to keep buying from you, will be rewarded the most and are more likely to notice you have a program in the first place.So comparisons of sales, value or loyalty for members versus non-members do not necessarily measure program effectiveness. Yet we regularly see statements like ‘program members spend 200% more than non-members’. This translates to, ‘my biggest spending customers are more valuable than my lowest spending customers’. It does not translate to ‘my loyalty program caused members to spend twice as much’. Self-selection bias makes program ROI measurement difficult, but it does not invalidate offering a program. If you do not have a good, competitive loyalty customer value proposition, customers may well self-select into your competitor’s program.
- Myopia; in competitive markets, customers are usually promiscuous (three quarters of Australian grocery shoppers visit at least two competing brands each month11) , this means at least some of your best customers are also shopping frequently with your competitors. Analysis of loyalty program data is a useful way of understanding the value of your customers to you, but the critical missing insight is how that value compares to their entire wallet. An example to consider… A large retailer sees an increase in spend size and frequency in a segment of mid-value customers over a given period…this is good, right? Well what if your share of wallet with these customers went backwards in that same period? Not so good now, they are spending a bit more with you, but potentially spending more with your competitors. How is it possible to see this? The data sharing economy (and coalition loyalty programs) are now unlocking new insights for brands – enabling you to look beyond your own view of the customer. Credit card companies are now opening up their data to help brands build a market-wide view of their customers’ behaviour, including (in certain circumstances) the ability to append external data at an individual customer level.
Solution to the measurement challenge?
- Before & After: (Generally possible for online programs where customer identity is known even before the customer enrols in the program.) This compares customer sales behaviour (categories purchased, average basket size, time between purchases etc.), at the individual customer level for identical periods before and after enrolment to avoid any seasonal distortions. With consideration given to major differences in marketing and promotional calendars during the two periods involved, increases in customer sales activity indicate the program is having a positive impact on results.
- Statistical Pairing: This approach does compare members with non-members, but at the individual customer level matches customers on every available attribute, attempting to make program membership the only significant difference between the two. This hopefully produces a set of paired customers large enough from which it is possible to extrapolate differences in value and churn to the larger program membership, to determine ROI.
- Redeemers Vs Non-redeemers: This is perhaps the easiest measure of program return to implement and is really a variant of statistical pairing. For programs with a loyalty currency (e.g. points) that is accumulated by members and redeemed for rewards, there is almost always a portion of members who qualify for a reward, but because they are not engaged or influenced by the program do not bother to ‘pick it up’; non-redeemers. Contrasting the behaviour of Redeemers with Non-Redeemers provides a measure of program effectiveness as the only difference between the two cohorts is the influence of the program. Differences in value can be attributed to the program investment. This is a conservative measure of return as it does not consider any positive program effects on members who do not accumulate enough to qualify for a reward, but does count all the program’s costs.
- Share of Category: By working with a large-scale data partner, we can compare program investments to changes in the market or category share of the program owner. For example; what impact on our share of category sales did our $1m investment in frequent flyer points, as incentives for members, have? This return is measured by secure tokenised access to payment card data for your members and non-members shopping in your category. Share of category is critical for any business who faces the arrival of a foreign entrant. With the arrival of Amazon to Australian shores, retailers of all sizes will be able to see how their market share at a customer / segment level, is impacted over time. Sporting and footwear retailers are equally at risk with the arrival of the French sporting giant Decathlon. Being able to accurately measure current levels of loyalty (internal view), with the additional share of wallet perspective (external view), provides critical insight for marketers to best direct customer investments. For example; which customer segments are most at risk? Where will I get the best ROI from directing marketing resources?
- Qantas Investor Day presentation 5 May 2017
- Visa & Ellipsis market research 2017
- Myer 2014 Full Year Results ASX release and presentation
- AIA Australia
- Qantas Assure
- Ellipsis research
- Wise Marketer research
- Customer Strategy Network research
- Don Peppers
- Reinartz & Kumar HBR 2002
- Roy Morgan - Supermarket loyalty: what’s that?
- Forrester - Age of the Customer