Amazon's Subscribe & Save Top-Off Discount: What It Is, What It Costs, and Whether It Works

On July 23, 2026, Amazon extends Subscribe & Save discount pricing to reorders: extra units of a subscribed product, and reorders of eligible products. The seller funds it, and it is on by default. One question decides whether it is worth it. Does it make customers reorder more than they would have anyway, or does it just discount orders you were already getting? Here is what one brand’s data says.

What is the Subscribe & Save top-off discount?

Subscribe & Save has always discounted a subscriber’s scheduled deliveries. Top-off applies that same discount to a different order: an extra unit outside the cadence, or a reorder of an eligible product. Not a coupon, not a subscription delivery. Just the S&S percentage on a one-off reorder.

The announcement bundles two different programs, which Amazon has since clarified:

• Subscribe & Save Top-Off. A subscriber orders extra of a product they already subscribe to, at their S&S discount. Almost all of today’s activity, and the half that may be working.
• Buy Again & Save. A separate, new program. A Prime customer who bought a product before gets your S&S discount on a reorder without subscribing, as long as they buy five eligible S&S products in one order (which can be from different brands). This is what July 23 opens up, and the half to worry about.

Both run on your own S&S discount, with no Amazon funding. Everything measurable today is the first kind.

Wait, is this actually new?

No. July 23 is the terms change, but the discount has run since December 2025. This brand’s first top-off order landed December 19. Of eight other supplement and consumable brands we checked, three started December 18, one day earlier, and five had none. A coordinated rollout, staggered by account. If you have not seen it, you are in a later wave.

Amazon has also moved the rate: 5% at launch, doubled to 10% at a fleet-wide cutover in late April 2026. Same promotion, twice the rate, five months in. The rate turns out to be the whole game for the economics.

Who actually qualifies for it?

We reconstructed the purchase history behind every top-off order:

• About 95% came from customers with a prior S&S order of that exact product.
• Roughly 2% had bought it before but never subscribed.
• Under 3% had no prior purchase of it, most likely a data artifact.

Two takeaways. This is not acquisition: almost no new customers. And eligibility travels with “ever subscribed to this product.” The median gap to a prior S&S order is about 50 days, but there is no cliff at 180 or 365 days, and a real tail runs past a year. Buying something else in between does not reset it.

Who funds it?

Amazon’s email (“more repeat orders,” “you may fund discounts”) reads like free retention. It is not, and Amazon has confirmed it: both programs use your existing S&S discount, and Amazon contributes nothing extra. S&S discounts are seller-funded, the 5 to 15% comes out of your proceeds. Every dollar of discount is a dollar off your own price, so the question is whether it buys anything.

The one question: is it incremental?

A discount can be incremental two ways, and they are easy to confuse.

Acquisition: does it bring in new customers? No. At 95% existing subscribers, it wins none, and the Scorecard grades it low on that.

Frequency: does it make existing customers reorder more? That is the real test, and it needs a clean before-and-after. The trap is seasonality, since this launched in December and ran through the Q1-Q2 peak. So we compare year-over-year and lock the cohort: the same subscribers who were active before the discount, against what those exact people did in the same months a year earlier. We use May and June, before Prime Day, so the event does not distort either year.

The same subscribers reordered about 18% more at the 10% tier than a year before. A year-older cohort normally reorders slightly less, not more, so that 18% overcomes a real decline. The true effect is a bit larger. The lift is real and modest: roughly a fifth.

Does the lift pay for the discount?

A lift only counts if it clears the margin it costs. At a contribution margin around 60%, the discount comes straight off profit, because costs do not move:

A 10% discount costs about 17% of the profit on the order, not 10%. That 17% is also the bar. The share of discounted reorders that must be incremental to break even equals discount divided by margin: about 17% at the 10% tier, 8 to 9% at 5%. One truly new reorder for every five that were coming anyway.

The 18% same-customer lift lands right around that bar, above it once the cohort’s aging is netted out. Not a giveaway, not a windfall. At 10% it about pays for itself, a bit better. A small, real win on the subscriber half.

The watch-item is pull-forward. If some of the lift is stock-up, it borrows from a future full-price order. The lift builds across months rather than spiking once, which argues against it, but it is early. Watch for a later dip below the prior-year line.

But don’t top-off users spend more?

They do. Top-off users generate about 70% more contribution profit per customer than subscribers who never use it, and place about 90% more orders. That is not proof of incrementality. Top-off users are the heavy buyers who self-select into ordering extra. They place more orders because that is who they already were. Your best customers use every offer. To credit the discount you need the same customer buying more, which is the year-over-year test above, not this gap.

Measuring it: what the Scorecard tracks

None of this shows up on its own in Seller Central. Amazon has confirmed that top-off orders just count as subscription units, folded into your S&S numbers, and Buy Again & Save flows into “Total Reorder” on the S&S dashboard. Neither is broken out as its own discount, so you cannot see what it costs. So we measure it directly. ExpandFi’s Promotion Performance Scorecard breaks the top-off discount out and grades it A to F, on the same scale as every other promotion. The row gives you:

• Incrementality (acquisition): the share of new or reactivated versus already-active buyers. Top-off scores low, correctly.
• Net value per customer: incremental revenue minus the discount you funded.
• Retention lift against a matched baseline, and a letter grade.

Every figure in this piece comes from that measurement.

The real risk: the non-subscriber reorder discount

The subscriber half may work. The other half, Buy Again & Save, is the risk. It gives your S&S discount to a Prime customer who bought a product before and reorders it without subscribing, once they build a basket of five eligible S&S products.

That basket gate bounds it. This is not every reorder, so the exposure is smaller than the full returning-non-subscriber base, about 40% of this brand’s revenue. But it is real, and there is no reason to expect a reordering lift, because these customers never chose a cadence. It is your margin on demand you already had. And it only triggers when your S&S discount is 10% or more, which makes it a lever, below.

Where it lands, the cost is the same. A 10% discount is about 17% of the profit on the order. And it does not stop at the order. Most of a repeat customer’s lifetime profit is reorders, and returning customers are about two-thirds of revenue, so discounting their reorders lowers lifetime profit itself. Lower LTP means a lower CAC you can afford.

What it does to a $10 acquisition

Say you can spend $10 to acquire a customer, based on their 12-month lifetime profit. Illustrative, but the mechanic is the point. Discounting non-subscription reorders at 10% lowers 12-month LTP about 5%. So the most you can pay and hold your return drops from $10 to about $9.50. Fifty cents on every acquisition, on every channel. And you underwrote the ads already running against the old $10, so profitable-looking spend was priced on a number that just shrank.

The levers you still control

Amazon turned this on by default, but you have real controls, and Amazon has now spelled them out. The S&S discount is seller-set: 0%, 5%, or 10%, up to 15% in some categories.

• Cut to 5% to switch off Buy Again & Save. The non-subscriber program only runs at a 10% or higher discount. Drop to 5% and it stops, while your subscribers keep their discount. The cleanest lever if you want the subscriber half without the non-subscriber half.
• Opt out of Buy Again & Save at the account level. Amazon says you can turn the non-subscriber program off at the account (MCID) level without touching your S&S enrollment.
• Top-Off has no separate opt-out. The only way to turn off the subscriber top-off discount is to leave Subscribe & Save entirely, so that half is effectively bundled with S&S.
• Test the rate up, toward 15%, if the subscriber lift holds. At 10% it only about breaks even, so treat it as a test.
• Reprice. Absorb the discount in the base price. Protects margin, costs competitiveness.

The upshot: the two halves are not locked together. You cannot opt out of the subscriber top-off without leaving S&S, but you can shut off the riskier non-subscriber half, by holding your discount at 5% or opting out of Buy Again & Save at the account level. Move the rate, watch net value per customer, let the data pick.

Should Amazon reconsider this change?

Amazon’s goal is reasonable: make reordering easier, hand sellers more repeat orders. Shoppers get a better price, and on the subscriber half the seller may too.

The design is the problem, on the non-subscriber half. It is seller-funded, on by default, and reaches customers who committed to nothing. It hits the profit that funds advertising: lower lifetime profit means a lower CAC a seller can afford, which means less ad spend, which is Amazon’s own revenue. And it weakens Subscribe & Save itself, because if non-subscribers get the discount on reorders anyway, the reason to subscribe erodes. Three choices deserve a second look: on by default rather than opt-in, eligibility by product rather than by real subscription, and the full cost on the seller. At a minimum, let sellers scope it to their actual subscribers, the half that pays.

What we don’t know yet

• How durable the lift is. The clean evidence is the first months of the 10% tier. A later dip below the prior-year line would mean pull-forward, not new demand.
• How much of your reorder volume hits the five-item basket. That gate bounds the Buy Again & Save exposure, and how often your non-subscribers build a qualifying basket is a data question, answerable once it is live.
• The full Buy Again & Save mechanics, which Amazon is still rolling out.
• How it ripples into subscribe rate, cancellations, and your new-versus-returning mix.

The bottom line

Good or bad depends on which half.

The subscriber half works, modestly. It does not acquire customers, but the same subscribers reorder about a fifth more, roughly enough to cover the margin, a bit better. Barring a pull-forward reversal, a small, real win.

The non-subscriber half, Buy Again & Save, is the risk. It marks down demand you already had, from customers who never chose a cadence, and lowers the lifetime profit that sets your acquisition budget. That half is closer to a giveaway, and July 23 turns it on, though you can switch it off by holding your discount at 5% or opting out at the account level.

Take the win where the data shows one, guard the flank where it does not. In ExpandFi, the Scorecard breaks the discount out and grades it, so you can tell the half that pays from the half that does not, on your own numbers.

What comes next

The sharpest question is not yet answerable. The reorder discount for non-subscribers takes effect July 23; only when those reorders start flowing can we test whether they are incremental. When that data arrives, we will run the same cohort-locked analysis and update this piece. The subscriber half is a modest win today. The non-subscriber half is the open question, and we will answer it with numbers.

See the top-off reports for your brand

Three reports in ExpandFi break this out on your own data, no query needed:

• Top-Off Sales by Period (Promotions, Top-Off): top-off sales by day, week, month, and quarter, split by channel.
• Subscribers vs Top-Off Users vs Non-Subscribers (Promotions, Top-Off): lifetime value and lifetime profit for the four customer groups, so you can compare top-off users against everyone else.
• Promotion Performance Scorecard (Promotions, Overview): grades the top-off discount A to F on incrementality, net value per customer, and retention, next to every other promotion you run.

Run this on your own brand

Every number here came from ExpandFi’s MCP. Connect it to Claude and paste this to run the same analysis on your account:

Using the ExpandFi MCP, analyze my Subscribe & Save top-off discount:

1. Find the promotion tagged "Subscribe and Save Top-Off". Show when it started, monthly order volume, and the discount rate over time, and state clearly whether Amazon changed the rate and when.

2. What share of top-off users had a prior Subscribe & Save order of the SAME product? Is this reaching existing subscribers or new customers?

3. Incrementality, done right: lock a cohort of subscribers who were active BEFORE the discount launched, and compare their reorder rate during the discount to what those SAME customers did in the same months a year earlier. Apples-to-apples, so it controls for both seasonality and the newer subscribers who would otherwise inflate the rate. Report the same-customer lift at each discount tier.

4. Break-even: at my contribution margin, what share of discounted reorders has to be incremental to cover the discount (it equals the discount divided by the margin)? Does the same-customer lift clear that bar? Flag pull-forward risk.

5. Cost side: what share of my revenue is returning non-subscribers (the July 2026 reorder pool), and how much does a 10% discount on those reorders lower my 12-month lifetime profit and the CAC I can afford?

Then give me a plain-English read: for my brand, is this discount incremental and worth it, or a margin giveaway?

New to ExpandFi? This is exactly the kind of question it is built to answer, on your own data.

FAQ