The second visitmatters too.
Value created for the team
Gave the team a concrete, discussable version of group-wide retention — including cross-brand logic that turns one venue's regular into another venue's first-time guest.
Achievement
Designed a complete membership system — group accounts, points wallet, status, brand-specific rewards — from a one-line concept.
In Week 2 we spoke with Avocado about loyalty programmes. In Week 7 I built a mock-up of one for the whole group. Between those two points, loyalty stopped looking like a discount and started looking like a design problem.
The mechanic that started it
The example that stayed with me was a top-up incentive: a customer adds $100 of stored credit and receives $30 in bonus credit. It works because it moves money before the meal. The customer has already committed, so the next visit is a decision they have partly made — and the restaurant holds the cash earlier.
The same mechanic, read from four sides
The business
Repeat customers are cheaper to reach than new ones, and committed spending brings revenue forward and makes demand more predictable.
The customer
People participate when the reward is worth the commitment and simple enough to understand at the table. If explaining it takes a paragraph, it will not be used.
The brand
A rebate on a casual venue reads as good value. The same rebate on a premium venue reads as a restaurant that needs to discount — the reward structure has to match the positioning, not fight it.
The finances
$30 given against $100 committed has a real margin cost. The incentive has to change behaviour by more than it gives away, which depends on redemption rates and how often people genuinely return.
What I actually built
In Week 7 I mocked up a Cantera-wide membership: a group account, a points wallet, membership status, and restaurant-specific campaigns and rewards. Designing the screens forced me past acquisition into what happens after someone joins — how status is earned, what a reward looks like when it is redeemed, and how a casual visit and a premium visit can sit in the same account.
A loyalty programme is not a discount mechanism. It is a behaviour-design mechanism.
Every element chooses a behaviour: top-ups buy commitment, tiers buy frequency, cross-brand unlocks buy exploration. Deciding the behaviour first is the whole job.
Where a programme can fail
- The reward is not worth the commitment.
- The structure assumes a visit frequency the venue does not have.
- Discounting makes a premium brand feel cheaper.
- The rules are too complex to explain at the table.
- It rewards behaviour the business did not want — discounting visits that would have happened anyway.
- It launches as a finished system with no way to adjust it.
Evidence from the internship
The specific moments this case study is built on.
We were introduced to the recommended loyalty approach, including the $100 top-up for $30 bonus credit structure.
My first thought was that it was a strong offer and would obviously drive returns.
I started asking who pays for it. An attractive reward still has to clear margin, redemption conditions and actual spending behaviour. Marketing decisions have to create value for the customer and the business, or they only look successful.
The group's restaurants range from casual to premium, and the same loyalty structure was on the table for all of them.
A single consistent programme seemed obviously simpler and fairer.
Simplicity for the operator can be damaging for the brand. Value-based rewards suit casual concepts; premium venues are better served by exclusivity, priority reservations or personalised privileges. I stopped treating consistency as automatically correct.
I compared how other groups structure loyalty while building the proposal slides.
I was looking for the best model to adopt.
There was no single best model — rebate-heavy systems work where visits are frequent and erode positioning where they are not. That pushed me to a structure with one group identity and per-brand flexibility in the rewards.
There was limited transaction data to set reward costs or tier thresholds.
I wanted to keep refining until the numbers could be justified properly.
That was the overanalysis my supervisor had identified, showing up in a real task. I reframed the deliverable as a launchable first version designed to be adjusted with real behaviour, and said so explicitly in the proposal rather than implying certainty I did not have.
The trade-off
What pulled against what
Customer appeal against margin; group consistency against brand positioning; simplicity against segmentation.
Why the obvious answer was not enough
A reward that changes behaviour on a casual venue can damage a premium one, and any incentive risks paying for visits that would have happened anyway.
Why I chose this approach
One Cantera membership with a shared identity, segmenting restaurants by customer behaviour rather than cuisine, and allowing rewards to differ per brand — plus cross-brand unlocks so loyalty to one venue can introduce a customer to another.
Going deeper
Loyalty as a group-level marketing tool
The most useful shift in the Week 7 work was realising a membership does not only bring a customer back to the restaurant they already like. Rewards and unlocks can move a Chicco regular towards Il Toro or Humo, which turns retention into acquisition for the other venues at close to no additional media cost.
Why the first version should be imperfect
With limited data, a programme designed to be final is a guess with no exit. A programme designed as a first version has a built-in reason to collect behaviour and adjust — which is a better answer than waiting for certainty that was not going to arrive.
The problem as it was handed to me
Offer a reward, customers come back more often.
The trade-off underneath it
A reward that changes behaviour on a casual venue can damage a premium one, and any incentive risks paying for visits that would have happened anyway.
My responseProposed next test
I would set reward costs and tiers from actual spend and visit-frequency data, and define upfront which numbers would tell us the programme was working before any of it went live.
What I contributed
Gave the team a concrete, discussable version of group-wide retention — including cross-brand logic that turns one venue's regular into another venue's first-time guest.
I developed a proposed Cantera-wide loyalty programme: a mock membership system covering group accounts, a points wallet, membership status and brand-specific rewards, supported by research into other restaurant groups and slides setting out the reasoning, segmentation and cross-brand logic behind it.
Decision rules I would carry forward
Design the behaviour first — commitment, frequency or cross-brand exploration — and only then choose the mechanic that buys it.
Pilot one mechanic per behaviour on a single venue for eight weeks and compare repeat-visit rate against a matched venue with no programme.
Ask for whatever transaction data exists before reasoning towards thresholds, and treat any threshold set without it as provisional.
Recalculate the proposed tiers from real average spend and visit frequency, then check how far the reasoned thresholds were off.
Every first version ships with its success measures and its review date written into the proposal.
Define three metrics and a 90-day review for the membership mock-up, and test whether stating them changes how stakeholders respond to an unfinished proposal.
Design retention backwards from the behaviour and the margin it has to clear, and treat launch as the start of the data collection rather than the end of the design.