As of November 7, 2026, Amazon has turned a delivery fee into a membership feature. Since November 5, US Prime members can order Uber Eats restaurant delivery inside Amazon.com and the Amazon Shopping app with a $0 delivery fee on qualifying orders of $25 or more, with no separate Uber account required. Amazon is also taking $20 off each of the first two qualifying orders during the first 14 days and paying 5% back to Prime Visa cardholders through December 5. None of that touches the menu price, which is precisely the problem for anyone monitoring price gaps across platforms.
A fee waiver is a price change
Three things are happening at once, and only one of them is visible on a product page. The delivery fee is waived, conditional on a $25 basket and on where the order is placed. An introductory credit reduces the first two orders by $20 each. A card benefit returns 5% for a limited period. Each of those is a price movement in the only sense that matters to a shopper, which is what leaves the account at the end of checkout, and none of them appears in a menu price scrape.
The entry point is part of the offer
Amazon's own copy tells members to search for Uber Eats inside Amazon rather than open the delivery app, and that instruction is the commercial point. The same restaurant, the same dish and the same courier produce two different totals depending on which app the order starts in. A price comparison that ignores the entry point is therefore comparing a price that no shopper pays with a price that only some shoppers pay, and it will misattribute the difference to the restaurant rather than to the platform's membership strategy.
The $25 line is a threshold, not a discount
| Cost line | Order routed through Amazon | Order placed in the Uber Eats app | What monitoring must record |
|---|---|---|---|
| Menu price | Set by the restaurant | Set by the restaurant | Identical, so it cannot explain a gap |
| Delivery fee | $0 above a $25 basket | Charged unless a pass applies | Fee plus the threshold it depends on |
| Service fee | Still applies | Still applies | Percentage base and minimum |
| Promotional credit | $20 off the first two qualifying orders | Varies by account | Credit value and expiry window |
| Loyalty return | 5% back on Prime Visa through December 5 | Depends on card and pass | Effective rebate as a share of basket |
The threshold deserves its own note, because it converts a fee into a basket rule. Below $25 the shopper pays delivery, at $25 and above the shopper does not, which means the cheapest way to buy a $22 meal is to add something. That is not a discount on the meal; it is a change in the shape of the basket. Monitoring teams that compare unit prices across platforms will see the same $22 price on both and conclude there is parity, while the shopper experiences a $3 fee on one route and a $3 upsell on the other.
Comparison rules that survive bundling
Bundling breaks the assumption that a product has one price. The working rule is to compare landed cost for an identical basket, at an identical time, from an identical entry point, and to record the conditions attached to each fee. Four fields carry most of the value: the fee amount, the threshold that removes it, the credit applied, and the expiry date of the promotion. Everything else can be reconstructed later; those four cannot, because promotions disappear and thresholds change without notice.
There is a second rule that matters for governance. A member benefit that applies only inside one company's app is a distribution incentive, not a market-wide price movement, and it should be logged as such. Mixing the two produces the worst outcome for a brand: a report showing that one platform is systematically cheaper, when what actually happened is that one platform subsidised a fee for its own members during a launch window. The correct conclusion is narrower and more useful, namely that the price a shopper sees now depends on membership status and entry point as much as on the listed menu.
What to instrument before Black Friday
The launch lands three weeks before the heaviest promotional period of the year, which leaves just enough time to fix instrumentation rather than rewrite it in December. The immediate task is to make sure every price capture includes the fee and the condition behind it, and that membership-dependent offers are tagged separately from general discounts. Without that separation, the Black Friday read-out will attribute a membership subsidy to a market price shift and send teams chasing a gap that closes when the trial ends.
Where the credit expires
A second task is to watch what happens after the opening credits expire. The $20 introductory credit runs for 14 days, the card benefit runs to December 5, and the fee waiver is described as a membership benefit rather than a promotion. Which of the three survives into ordinary trading is the test of whether Amazon has changed the price of restaurant delivery or simply bought a trial. Until that is visible, the defensible statement is the narrow one: on one entry point, for one membership tier, one cost line has moved.
FAQ
What changed for Prime members on November 5?
US Prime members can order Uber Eats restaurant delivery inside Amazon.com and the Amazon Shopping app with a $0 delivery fee on qualifying orders of $25 or more, using their existing Amazon login. Amazon also takes $20 off each of the first two qualifying orders and pays 5% back for Prime Visa cardholders through December 5, 2026.
Is this a price cut on the food itself?
No. Menu prices are set by restaurants and service fees, taxes and tips still apply. One checkout line moved, and only for orders placed through Amazon rather than the Uber Eats app. A comparison built on menu prices will therefore show no change at all.
How should a monitoring programme record it?
By capturing landed cost: menu price, service fee, delivery fee, the threshold that removes it and any promotional credit, for the same basket at the same time on each entry point. Without the fee and the condition, the comparison is both incomplete and unstable.
Does the $175 annual saving claim hold?
It is Amazon's estimate, annualised from August 2026 order data and compared against ordering on Uber Eats without Uber One. It assumes a typical order frequency and guarantees nothing per order. Treat it as a marketing figure rather than an audited benchmark.
