The question I get asked most often by brands shopping for a new loyalty platform is how to make existing points do more work. It is almost never the right question. In almost every case the problem is not that the points are underpowered; it is that the programme has been quietly funding the wrong behaviour for years.
Starbucks is the public example this year. A programme that had drifted into a coupon engine, 34 million members strong but shaping its busiest customers into discount seekers, became the centre of a turnaround that an incoming CEO ran as a definition problem rather than a technology problem. The fix cost no new platform. It cost a quarter of executive time and the willingness to stop rewarding the wrong thing.
I was sceptical when the “Back to Starbucks” plan was announced in September 2024. New CEOs arrive with slogans and leave with restructuring charges. What makes this one worth your attention is the arithmetic on the far side. Two years in, the comparable sales line has turned, the membership count is a record, and the piece that moved the needle was not pricing, store refit, or a new menu. It was the Rewards programme, reframed.
Before Niccol, Starbucks Rewards Was Earning the Wrong Members
Laxman Narasimhan left Starbucks on 13 August 2024 after a little over a year, in the middle of a sales slump and a public backlash over pricing (Boston Globe, 13 August 2024). Brian Niccol, then Chipotle’s CEO, took the chair on 9 September 2024 and published an open letter in his first week committing the company to get “back to Starbucks” (coverage from the time collected in the same Boston Globe piece and in follow-up reporting from NBC News). Howard Schultz later told NBC he “did a cartwheel in his living room” when he first heard the phrase. That is unusual behaviour from a predecessor.
The diagnosis Niccol published, and has repeated on every earnings call since, is that Rewards had drifted into a coupon engine. Members redeemed discounted drinks on days they would have visited anyway. The light or infrequent customer, the one who could have added the next percentage point of comparable sales, was not in the programme at all. The programme was subsidising its best customers to do what they were going to do, at a lower margin, and doing nothing for the ones who could have moved the needle.
By the Q2 FY2026 earnings call on 29 April 2026, Niccol had put the frame in plainer language. “We had to stop doing all that discounting. This needed to be about engagement,” he told analysts, as Restaurant Dive reported on 30 April 2026. He added the sharper line a few minutes later: “you don’t have to make our rewards program a coupon book.”
A Loyalty Programme Is Not a Discount Engine, and the Two Should Not Share a Budget
Niccol had four obvious levers on arrival: price, menu, store experience, and loyalty. He is working on all four, but the one he put at the centre is Rewards, and the reason is worth understanding before you run your own loyalty review.
Price moves margin per transaction. Menu moves ticket. A store refresh moves footfall in a radius. Only the loyalty programme moves frequency per customer, and only sustainably if it is not paying the customer to show up. A flat points-per-dollar programme does the opposite. It pays your busiest customers extra, at the expense of margin, for behaviour that was already baked in. The headline membership figure looks good. The underlying unit economics corrode quietly.
Niccol’s line from the Q2 call, which Restaurant Dive quoted in the same piece, is the sentence to carry back to your own team: “You have to win both with your Rewards customers and, call it, the light or infrequent customer.” The implicit admission in that sentence is important. Rewards can win the first group. It cannot win the second, because the second does not read the programme, is not earning stars, and does not care. Trying to make one programme do both jobs is how most loyalty programmes end up where Starbucks had ended up: optimised for no one.
The decision that followed is the one most mid-market brands will not make. Niccol separated the two problems. Rewards would carry the frequency job, and it would do so by paying for engagement rather than subsidising a visit. The light-customer problem went to marketing, menu, and store experience, where it belongs. The Rewards budget stopped quietly bankrolling the discount pile.
The March 2026 Revamp Put Tiers Back and Discounts Out
Starbucks announced the reimagined Rewards programme at its Investor Day on 29 January 2026 (investor.starbucks.com, 29 January 2026). It went live on 10 March 2026 with three tiers (Green, Gold, Reserve), faster star-earning as spend rises, stars that never expire for Gold and Reserve members, free monthly customisations at every tier, and a 60-star redemption that had already become the most-used benefit in the old programme, accounting for roughly one-third of all redemptions.
Tressie Lieberman, Starbucks’ global chief brand officer, framed the change in the Investor Day press release: “Our Rewards program is strong, and we’re building from a position of leadership.” Her operational line from the same release is the one worth stealing: “we identified clear actions to unlock the next generation of loyalty.”
The mechanics of the revamp read as commodity loyalty-platform features. They are not. What tiers do that flat points cannot is make frequency legible to the customer. A member who crosses from Green to Gold has crossed a threshold the programme named out loud. The customer recognises a change in status; the programme recognises a change in value. Flat points buy a running total that nobody experiences as a milestone. That is the piece Starbucks got wrong under the old regime and has quietly re-engineered.
Stars that do not expire for Gold and Reserve members are the less obvious lever. Expiry is often sold as a retention tool, on the theory that a soon-to-expire balance pulls customers back. In practice it punishes exactly the customers you want to keep, and it creates a liability conversation with finance every quarter about carrying unredeemed balances. Removing expiry at the top tiers signals permanence of the relationship. It costs margin on paper. It buys retention at the only tier where retention is worth paying for.
Two Years In, the Numbers Settle the Argument
Membership figures are the clearest indicator because they are reported on a consistent basis. US 90-day active Starbucks Rewards members moved from roughly 34 million in Q3 FY2025, to 35.5 million in Q1 FY2026, to 35.6 million in Q2 FY2026, to a record 36 million in Q4 FY2026, reported on the Q4 call that followed the 29 October 2026 earnings release. The Investor Day release in January confirmed that Rewards drove nearly 60% of US company-operated revenue in fiscal 2025. The subsequent year has, if anything, reinforced that share.
The sales line tells the same story. Q4 FY2026 global comparable coffeehouse sales were up 7.2%, with the full fiscal year up 6.2%. Non-GAAP earnings per share for the quarter were $0.80, a 54% increase. More than 80% of North America company-operated coffeehouses reported positive comparable sales in the quarter, up from fewer than half a year earlier. Full-year revenue of $37.9 billion was up 2%; Q4 revenue of $9.1 billion was down 4%, a figure the release attributes to the mid-year transition of the China business to a licensed model rather than to the underlying trading trend.
Niccol’s summary in the release is characteristically brief: “Back to Starbucks” started as a turnaround plan and is now the company’s growth plan. He is allowed the sentence. The underlying numbers earn it.
The Light-Customer Problem Has Not Been Solved by Rewards, and Was Not Supposed to Be
The story is not finished, and the unsolved part is the interesting one. The restructuring announced alongside the Q4 FY2026 release includes closing roughly 250 North American coffeehouses that, in the company’s own phrase, were not delivering “the coffeehouse experience” the brand requires. That is not a Rewards problem. It is a store-portfolio decision that would have come whether or not the programme had been revamped.
The Fiscal 2028 targets Starbucks set at the January Investor Day, namely 5% or more consolidated revenue growth, 3% or more global comparable sales, 13.5 to 15% non-GAAP operating margin, and non-GAAP earnings per share of $3.35 to $4.00, are not loyalty-programme targets. They are company targets. Rewards is a load-bearing wall in that plan, not the whole building. If your takeaway from this piece is that a loyalty programme can carry a turnaround, you have read the wrong half of it.
What the programme did do, and what we would want ours to do for an organisation evaluating a similar move, is remove a quiet source of margin erosion and replace it with a frequency lever that pays for itself. The light-customer problem goes back to marketing, menu, and physical experience, which is where it should sit.
What This Looks Like Inside a WooCommerce Store
If your business runs on WooCommerce and you are considering retiring a flat points programme in favour of something tier-shaped, there is a concrete data model and a specific set of decisions that need to land before any plugin is configured. We have done this reset a handful of times. The cost is not where buyers expect.
The primary entity is not the point. It is the customer-period: a rolling window (90 days is Starbucks’ choice, not a default) over which a customer’s order history, visit count, and basket composition are aggregated. On WooCommerce that aggregation lives most cleanly in the order statistics table that the analytics system already maintains, with a materialised view refreshed nightly. Tier assignment, star accrual, and benefit eligibility are then functions of that aggregate. Plugins that make points the primary entity, and let the customer-period emerge as a by-product, quietly lock you into the coupon-engine configuration Starbucks has just walked out of.
Tiering requires two decisions that are usually left to the plugin’s defaults, and should not be. The first is the threshold rule: trailing-period spend, trailing-period visits, or a composite. We prefer visits over spend for mid-market retail, because spend-tiers reward the one annual splurge and visit-tiers reward the behaviour you want to compound. The second is the demotion rule. Most WooCommerce loyalty plugins let a customer sit in a tier indefinitely once they have earned it. Starbucks does not, and neither should you; a tier that cannot be lost is a status gift, not a retention lever.
Benefits that are not discounts are the editorial work that costs the most and takes the longest. Free customisations, member-only product access, early release windows, and priority support are built in WooCommerce with a membership plugin layered over conditional product rules; the technical piece is well understood. The deciding piece is which benefit you give away at which tier, and that is a quarter-long conversation between marketing, merchandising, and finance. The technology is three sprints. The agreement is three months.
Star or point expiry is the one decision we would ask you to make on day one, before any plugin is picked. Expiring stars at the top tier buys very little retention and generates a finance conversation every quarter about unredeemed balance liability. Removing expiry at the top tiers, the way Starbucks has, moves that liability conversation from a recurring annoyance to a one-time accounting change and signals permanence to the customers you most want to keep. For a mid-market store carrying a seven-figure point liability, this is a board-level decision, not a plugin setting.
Rough shape of the work on a WooCommerce stack already running a flat points programme: one sprint for the customer-period data model and tier logic, one sprint for benefit configuration and the membership plugin wiring, one sprint for the migration of existing point balances into the new scheme with a grace period. Three sprints of engineering. The decision work around what Starbucks took a quarter to resolve, which is what you stop subsidising, is the schedule risk. We have never seen that part finish faster.
The Starbucks lesson a buyer should take from this is not that your programme needs three tiers and non-expiring stars. It is that a loyalty programme is a definition before it is a configuration, and the definition that matters is which customer behaviour you are willing to pay to produce. Niccol’s answer was engagement, legible through tiers. Whether that is the right answer for your business depends on how you compete, and we would rather have that conversation with you over the shape of your existing order data than over a plugin comparison chart. If a loyalty review is on your roadmap for 2027, that is the conversation worth starting now, and we would be glad to walk through what it would look like against your stack.
Last modified: November 7, 2026
