Some days I wake up with weird stuff in my head. This morning, for totally unknown reasons, it was Brewster’s Millions. You know, that 80s film with Richard Pryor, where he has to spend $30 million in 30 days, with nothing to show for it at the end, to inherit a much bigger fortune (the haircuts are best forgotten).
Then, fast forward a couple of hours and a few coffees later, I was reading Bikky’s follow-up interviews with five restaurant CMOs. A 400-unit QSR brand grew loyalty sales this year while cutting loyalty spend by 25%. The CMO’s explanation was simple: they got “really, really targeted about who we’re giving stuff to.”
Not the most technical answer you’ll ever read. Probably the most useful.
In the same piece, a 50-unit casual dining brand described the opposite. Redemptions rose, visit frequency softened, and loyalty ROI fell. Plain old meal deals ended up outperforming the programme as traffic drivers.
And there was Brewster again. A loyalty programme that rewards everyone, piles points onto visits that were going to happen anyway, and pushes plenty of money out of the door with not much to show for it when someone finally asks.
That’s not a dig at loyalty programmes, by the way. Most were built when “reward everyone” was the right call, and plenty have earned their place. If you run CRM or loyalty at a QSR brand, though, the gap between those two stories is a good place to start your 2027 plan. The question is whether yours is still spending like Brewster.
Read redemption and frequency together
Rising redemption looks lovely on a dashboard. Pair it with falling frequency, though, and it usually means you’re paying guests for visits they were going to make anyway. Very Brewster.
The only way to tell the difference is a control group: a slice of eligible guests who don’t get the offer, measured against the ones who do. Without one, last-touch attribution happily credits the programme for every redeemed visit, and the number looks great right up until your CRO asks for the real one.
Our CRM team at Massive Rocket has seen this first-hand. One programme they inherited was reporting more than three times its true incremental revenue, because the control group had been set up incorrectly. Nobody had done anything wrong on purpose. The measurement had just never been checked.
Spend on the guests you can actually move
Membership on its own is a weak reason to send someone an offer. A guest whose next visit you can actually change is a strong one.
The guests worth investing in are usually the ones showing early signs of drifting, with visits spacing out, baskets getting smaller and the app opened less often. A binary “lapsed after 12 months” rule finds them far too late. Meanwhile, your regulars who come in every Tuesday don’t need a discount to keep coming in on Tuesdays (they’ll be there anyway, probably ordering the same thing).
I’ve been on the receiving end of getting this wrong. I wrote recently about my local petrol forecourt sending me a fuel price alert every Friday, three years after I’d switched to an EV. Every. Single. Friday. My behaviour had changed completely, and the messages never noticed.
Which is why silence is a decision too. In one QSR programme, our team found a recency segment that was buying less because of the emails it was getting. They caught it and suppressed it within a week.
Treat every channel as its own market
Bikky’s CMOs found DoorDash barely reacted when ad spend was cut, while Uber Eats sales fell sharply under the same strategy. The 20-unit fast-casual team now runs the two platforms like separate paid media channels, each with its own budget, offers and audiences.
The same logic applies to your own channels. Someone who orders through the app behaves differently from someone who only ever comes through the drive-thru, and they respond to different things. If your reporting lumps them together, your offers will too.
Go looking for the outliers
Averages flatten everything. Bikky’s own example is worth borrowing: a brand trying to fix breakfast found only around half of orders in that daypart contained a breakfast item, and the top seller was a cheeseburger (respect). Reframing the problem as “more traffic from 7 to 11am” changed what they built.
Your CRM data has stories like that sitting in it, like an offer that works in one market and nowhere else, or a daypart where app orders jump for no obvious reason. Work out why before you try to copy it somewhere new.
Your 30-day check
Brewster had 30 days to spend it all. Here’s a 30-day look-back that takes about an hour, and it’s the quickest way I know to see whether your programme is spending like him:
- Pick your highest-volume campaign or offer from the last 30 days.
- Pull the revenue your platform attributed to it.
- Ask: how much of that would have happened if we’d never sent it?
- If you have a control group, run the comparison. If you don’t, the gap between the reported number and the real one is the size of your problem.
If that gap turns out bigger than you’d like, don’t panic. You won’t be the first, and it’s very fixable.
This is step two of our Agentic CRM framework, Fix the Reports, and our team does it before almost anything else with a new client. As more of the programme gets handed to automation and agents, those systems will optimise for whatever you tell them success looks like. Get that number wrong and they’ll scale the wrong thing very efficiently.
The 400-unit brand grew loyalty sales while giving less away, because it got clear about who needed the push.
So when did you last check what your rewards are actually buying you? If you’d rather not find out the Brewster way, talk to us about our Agentic CRM framework. We’ll help you spend less and end up with way more to show for it.
