Attribution and ROMI Case Study: Casino Loyalty Programmes
Casino loyalty programmes are one of the oldest large-scale attempts to link marketing spend to customer value. Every member's play is tracked, every offer has a cost, and the question "did this campaign make money?" has been asked for decades. That makes them an excellent case study in ROMI and attribution, including the traps.
The example below is an illustrative composite of a regional casino resort with round numbers. It is not based on any real company, but the mechanics are the ones casino marketers use.
The currency: theoretical win
Casinos do not judge a guest by what they won or lost on a given night, because luck swings both ways. They use theoretical win ("theo"): the amount the house expects to win from a guest's play, given the games, stakes, speed and time played.
A slot player betting $2 per spin, about 600 spins an hour, for three hours on machines with an 8% house edge generates a theo of roughly $2 × 600 × 3 × 0.08 = $288 for that trip. Tracking this through a loyalty card lets the casino estimate each member's average daily theo and visit frequency, which together give an expected value per member. It is a lifetime-value model that B2B marketers would recognise from the wider definition of demand generation.
Reinvestment
Casinos then decide how much of that expected value to return to the guest as points, free play, rooms, meals and event invitations. This reinvestment rate is the casino's marketing budget per customer. Set it too low and guests drift to a competitor; set it too high and the casino gives away its margin.
The campaign
Our resort wants to increase midweek visits from members who used to come monthly but have not visited for 60 days. Marketing proposes a mailer offering $50 in free play valid Monday to Thursday.
The tempting way to measure it is to count redemptions: 1,800 of the 20,000 members who received the offer visited and used it, and their play generated $360,000 in theo. That looks like a spectacular return on $90,000 of free play. It is also wrong, because many of those guests would have come anyway.
Attribution with a control group
So the resort held back a random 4,000 eligible members who received nothing. Over the same weeks, 6% of the control group visited anyway, against 9% of the mailed group.
| Mailed group | Control group | |
|---|---|---|
| Members | 20,000 | 4,000 |
| Visited in offer period | 9% (1,800) | 6% (240) |
| Average theo per visiting member | $200 | $200 |
| Theo per member | $18.00 | $12.00 |
| Incremental theo per member | $6.00 | n/a |
The incremental theo is $6 × 20,000 = $120,000, not $360,000. Costs are the free play actually redeemed ($90,000, though casinos often count free play at less than face value because part of it is played back to the house) plus printing and postage of about $15,000.
The campaign is modestly profitable on theo, before considering hotel and food revenue from the extra visits, and far less impressive than the redemption count suggested. The resort can now test a $25 offer, a different segment or a different channel, each against its own control group.
Online, the same logic runs faster
Online casinos and sportsbooks apply the same approach every day. Each promotion is compared with a holdout group, and the cost of bonuses is set against the incremental net gaming revenue they produce. Our article on marketing automation for betting operators explains how those tests are built into automated journeys.
Lessons for B2B ROMI
- Measure value, not activity. Theo is the casino's version of pipeline weighted by probability.
- Last touch flatters. Counting redemptions is last-touch attribution, and it over-credits campaigns that reach people who were coming anyway.
- Holdouts beat models. A random control group is the most convincing evidence of contribution. See revenue performance management for B2B attribution options.
- Report it simply. One incremental number, one cost, one decision, as in communicating with execs.
Related reading