One date correction before the analysis, because it changes what the story is about: Subway’s move from tokens to points, and its introduction of a three-tier membership, is not a 2026 event. MVP Rewards launched on September 9, 2023 across the United States, Canada and Puerto Rico, replacing the MyWay program. Roughly 30 million MyWay members were migrated automatically and their unspent tokens converted into points.
What happened in 2026 is the more interesting part, and it only makes sense against that 2023 baseline. Subway relaunched its loyalty proposition as Sub Club, gave it one of the most aggressive reward structures in the restaurant industry, and then pulled the centrepiece back within months. The full arc — tokens, to points and tiers, to stamps, and back to points — is a three-year natural experiment in what a franchised system can actually afford to give away.
The 2023 Structure
MVP Rewards did two things at once, and it is worth separating them.
The first was a currency change. Tokens, which accumulated on a per-visit or per-item basis, gave way to points earned per dollar: 10 points for every $1 spent, with additional points for ordering in the app or online, and 400 points converting to $2 in Subway Cash usable against future purchases.
That is a 0.5 percent effective return at the headline rate, before promotional multipliers. It is a modest number by fast-food standards, and it is deliberately modest, because a points-per-dollar system scales with the check in a way a per-visit token does not.
The second was the introduction of status: three tiers — Pro, Captain and All-Star — earned on annual spend, each unlocking more earning power and perks. Converting existing token balances into points on migration day gave the incoming 30 million members a head start toward those tiers, which is a well-understood tactic for making a new program feel immediately worth engaging with rather than immediately worth abandoning.
Together those changes moved Subway from a punch-card mentality to a spend-based one. That is the direction essentially every large restaurant program has travelled, for the same reason: per-visit rewards pay the same amount to a customer buying a drink as to one buying lunch for four.
The 2026 Reversal
Sub Club launched in December with a benefit that broke sharply from that logic: buy three footlongs, get one free. Measured against a 0.5 percent points return, a free fourth footlong is an enormous giveaway — roughly a 25 percent discount for a customer who orders footlongs consistently, delivered as a stamp card rather than as a percentage of spend.
It did not survive. Franchisees representing more than 5,000 locations signed a petition arguing the program was too generous to customers and that they stood to lose money on it. Subway ended the free-footlong benefit on April 1, 2026 and converted Sub Club to a purely points-based system — back, in substance, to the 2023 structure of 10 points per dollar and 400 points for $2 in Subway Cash.
The wind-down details matter for anyone assessing how the reversal was handled. Stamps toward a free footlong were forfeited on April 1. Customers who had already earned free subs before February 23 kept 12 months from the date earned to redeem them; those earning between February 23 and March 31 got 60 days. In the interim, stamps could no longer be earned on discounted subs at all.
That is a compressed redemption window on a benefit customers had been actively working toward, and the customer reaction was predictably poor.
Who Actually Pays for a Loyalty Program
The reason this is a franchising story rather than a marketing story is that in a franchised system, the brand designs the reward and the operator funds it.
A corporate marketing team evaluating a buy-three-get-one offer models incremental visits, larger baskets, and share taken from competitors. A franchisee evaluating the same offer models the food cost and labour on a sandwich handed over for nothing, against a customer who in many cases was going to buy the first three anyway.
Both analyses can be internally correct. They diverge because they are measuring different things — system-level growth versus store-level margin — and when the operators carry the cost, the store-level number is the one that decides whether the program lasts. More than 5,000 locations signing a petition is not a marketing disagreement; it is the franchise base declining to fund the offer.
This is also why the stamp mechanic was the part that failed rather than the points. A points program has a self-limiting cost: the reward is a fixed fraction of spend, so the liability grows only as fast as revenue. A buy-three-get-one stamp card has an unbounded ratio — a customer who only ever buys the cheapest qualifying footlong extracts the maximum, and the operator has no lever to adjust.
What This Signals for Fast-Casual Loyalty
Three transferable conclusions.
Currency design is cost control, not customer experience. The move from tokens to points in 2023 was framed publicly as giving members more ways to earn. Its more durable function was to tie reward cost to spend. The 2026 episode confirmed the point by demonstrating what happens when a program temporarily abandons it.
A franchised system has a lower generosity ceiling than a company-owned one, and that ceiling is not negotiable after launch. Any program a franchisor cannot get its operators to fund will be withdrawn, and withdrawal costs more goodwill than never launching would have. The consent has to be secured before the announcement.
Reversals are more expensive than the offer they cancel. A customer who never had a free footlong is neutral. A customer who was three stamps in and had them voided on April 1 is worse than neutral, and has been given a concrete reason to distrust the next thing the program announces. Subway spent roughly four months buying a lesson about its own franchise economics that a pre-launch operator consultation would have supplied for nothing.
For operators designing loyalty in 2026, the practical takeaway is to model the program against the customer who exploits it best, not the customer who uses it typically — and then to check whether the people paying for it agree with the number.



