FBA got more expensive this year, which made self-fulfillment look better. Amazon raised the bar for self-fulfillment in the same year, which made it harder. Most brands have modeled exactly one of those.
Seller Fulfilled Prime has spent years being described as theoretically interesting and practically impossible for anyone below enterprise scale. That reputation was largely earned. The program demands the Prime delivery promise from your own operation, and the Prime delivery promise is the hardest thing Amazon does.
What changed is that the alternative got more expensive. Amazon ended prep services in January 2026 and added a fuel and logistics surcharge in April, on top of storage and aged inventory structures that already punish slow-moving stock. For certain products, the FBA number moved enough to reopen a question that had been closed.
Then Amazon tightened SFP in July. This post covers both sides, because a decision made on one of them will be wrong.
Why This Is Back On The Table
The case for reconsidering is specific rather than general, and it applies to a narrower set of products than the enthusiasm suggests.
FBA fulfillment fees rise steeply with size and weight. Storage is charged on cubic feet, so bulky items pay disproportionately. Aged inventory surcharges compound on anything slow. Stack those against a product that is large, heavy, slow-moving, or high-value, and the FBA cost per unit can exceed what you would pay to ship it yourself, particularly if you already run a warehouse for other channels.
Add the 2026 changes. Prep is now your cost regardless of which path you choose, which removes one of FBA's convenience advantages. The fuel and logistics surcharge applies to fulfillment fees. Neither is enormous alone; together they shifted the comparison.
SFP is not a cheaper version of FBA. It is a different operating model that happens to be cheaper for a particular product profile. If you do not already have fulfillment capability, the cost of building one to satisfy Prime requirements will exceed anything you save, and this analysis does not apply to you.
Where it genuinely fits is brands already running a warehouse for a direct-to-consumer channel, with products FBA prices badly. For those brands, SFP puts a Prime badge on inventory they are already holding and shipping. That is the real proposition.
Delivery, Not Dispatch
This is the conceptual key to the entire program and the thing sellers evaluating it most often misunderstand.
When Amazon relaunched SFP enrollment in October 2023, it replaced the previous on-time shipment and Buy Shipping obligations with an on-time delivery requirement. Accountability moved downstream, from the moment you hand a package to a carrier to the moment a customer receives it.
You are no longer being graded on whether you shipped on time. You are being graded on whether the carrier delivered on time, which is a thing you influence and do not control.
Everything difficult about SFP follows from that sentence. Carrier performance becomes your performance. Weather in a region you do not operate in becomes your problem. A carrier's rural coverage gap becomes a metric you are measured against.
The operational consequences run deep. Carrier selection stops being a rate negotiation and becomes a reliability decision. Handling time settings stop being a convenience and become a promise you are graded on, which means conservative settings protect the badge at the cost of some conversion. And the geographic distribution of your inventory determines your compliance, because transit time is a function of distance.
The Performance Thresholds
Several numbers circulate. Here they are with their sourcing made explicit, because one of them is disputed.
| Metric | Reported Threshold | Sourcing Note |
|---|---|---|
| On-time delivery | 93.5% or higher | Widely reported. One source says sellers "generally need at least 90%". Operate to 93.5%. |
| Valid tracking rate | 99% or higher | Consistently reported across sources. |
| Seller-initiated cancellation | 0.5% or lower | Consistently reported. Far stricter than standard seller-fulfilled. |
| Delivery speed | Varies by size tier | Standard, Oversize, and Extra Large are evaluated separately. |
| Late shipment rate | Below 4% | Prequalification context. |
Two of these deserve emphasis. The cancellation threshold of 0.5 percent is brutally tight, which means a single stockout event on a low-volume SKU can breach it. Anyone considering SFP needs inventory accuracy considerably better than they probably have.
And the separate evaluation by size tier is a planning fact rather than a detail. Standard, Oversize, and Extra Large are effectively three different programs with three different operating models, so a brand can reasonably run SFP on one tier and FBA on another.
None of these figures are confirmed on an Amazon-owned public page, since program requirements live inside Seller Central. Verify against your own account rather than planning from any published article, this one included.
What Changed In July 2026
This is the change that decides whether SFP is available to you at all, and it is more consequential than its incremental appearance suggests.
Reporting indicates that from July 6, 2026 Amazon raised the standard-size delivery-speed requirement so a one-day delivery date must display on 40 percent of Prime page views, up from the 30 percent set at the 2023 relaunch. Amazon has reportedly confirmed that sellers already meeting the new thresholds need take no action, and that other eligibility requirements are unchanged.
Ten percentage points sounds incremental. Work out what it means geographically and it is not.
A one-day delivery date requires the customer to be inside a carrier's one-day zone from your facility. No amount of operational excellence extends that radius.
The threshold is share of Prime page views showing a one-day date, so it reflects where your customers are, not where you wish they were.
Reporting is direct on this: single-warehouse SFP sellers face a structural disadvantage, and distributed networks are now the practical baseline for nationwide coverage.
Compliance comes from splitting inventory across regions, which reintroduces the fragmentation cost SFP was partly meant to avoid.
There is a genuine irony here. The brands most attracted to SFP are those wanting to avoid Amazon's fulfillment network, and the requirement to satisfy it is building something structurally similar to Amazon's fulfillment network.
Reporting also indicates a new tool launching in September 2026 that lets sellers feed shipping times, weekend availability, and cut-off times at individual delivery zip code level directly into the customer-facing delivery promise. Today that translation from your shipping templates is opaque, so more direct control could recover speed-metric performance without new warehouses. Worth investigating before concluding you cannot comply.
The October 17 Weekend Deadline
If you take one action item from this post, it is this one, and it is three weeks away as this publishes.
Reporting indicates weekends are excluded from SFP speed metric evaluation until October 17, 2026, after which they count again. Weekend pickup has been an SFP requirement, but the speed metric has not been counting those days. That changes.
If your warehouse does not operate Saturdays, your one-day and two-day rates are currently being measured on days you do operate. After October 17, weekend page views enter the calculation. A seller comfortably above the speed threshold today can fall below it without changing anything, because the measurement window is what changed.
The compounding problem is timing. October 17 lands immediately before peak season, which is when operations are most stretched and least able to absorb a new requirement. A seller who discovers a speed shortfall in late October is discovering it at the worst possible moment.
Practical responses, in rough order of cost. Confirm your current speed metrics and estimate what they look like with weekend days included. Establish weekend carrier pickup if you do not have it, since USPS is reported to have the broadest Saturday coverage nationally. Staff at least Saturday dispatch. And if none of that is achievable before the date, plan for the possibility of losing the badge and decide now whether that is acceptable rather than reacting in November.
Verify the date in Seller Central. It comes from industry reporting of an Amazon announcement rather than a public Amazon page, and dates of this kind occasionally move.
Prequalification And The Trial
Entry has two stages and the second one is graded.
Prequalification. Reported requirements include a professional selling account, a domestic US default shipping address, at least 100 seller-fulfilled packages shipped in the previous 90 days, cancellation under 2.5 percent, valid tracking above 95 percent, and late shipment under 4 percent. In other words, a functioning self-fulfillment operation with a track record, not an aspiration.
The trial. Sources disagree on its shape. Amazon's public program page is described as specifying 30 days, while one source describes it as four complete performance weeks running Sunday through Saturday. Either way you ship at least 100 Prime trial packages, your offers do not display the Prime badge during the trial, and you must meet the delivery-speed and performance thresholds throughout.
Reporting on a June 2025 policy update indicates a cap of three trials per year, a minimum of 100 monthly Prime shipments, and restricted graduation periods before major sales events. The trial cap matters more than it sounds. Three attempts a year turns a failed trial from a setback into a meaningful loss of optionality.
It is not a learning period. Carrier mappings, weekend staffing, and tracking integrations should all be proven before you start, because you are being measured from the first order. Ship several test orders through your intended carrier first and confirm tracking registers correctly in Seller Central. Tracking scans arriving twelve hours late can quietly sink valid tracking rate before anyone notices.
On timing, reporting suggests the better trial windows are January through early June and August through mid-October, with late Q3 and Q4 high risk because they either block graduation or consume an attempt when operational pressure is highest.
Buy Shipping And Tracking
The 2023 relaunch removed Buy Shipping as a standalone obligation, and Buy Shipping still matters. Both statements are true and the distinction confuses people.
Reporting indicates sellers must purchase shipping through Amazon's Buy Shipping service or through carriers that integrate with it, because that integration is what allows Amazon to confirm delivery against its own tracking systems. It underpins valid tracking rate rather than existing as a rule for its own sake.
Carriers must also be on Amazon's list of eligible shipping providers, with most national carriers qualifying. Verify against current requirements for your marketplace rather than assuming, and test before the trial rather than during it.
The practical failure mode is a carrier that appears suitable and produces tracking events Amazon reads late or not at all. Delivery genuinely happened on time, the record says otherwise, and the metric moves against you. This is checkable in advance with a handful of test shipments and is one of the more avoidable ways to fail a trial.
Using A 3PL For SFP
Given the distributed-network requirement, a third-party logistics provider is the obvious route to compliance, and it comes with a caveat worth stating plainly.
You can use a 3PL for SFP, and the provider must meet the same performance standards an in-house operation would. The badge attaches to your seller account rather than to the provider, so their missed deliveries are your metrics. You have outsourced the operation and retained the accountability.
That makes provider selection a different exercise than a normal 3PL evaluation. The questions that matter are whether they have a confirmed SFP track record, whether they genuinely operate weekend fulfillment, whether their network gives you the regional coverage the speed thresholds now require, and whether their carrier integrations produce tracking Amazon reads reliably.
A provider who is excellent at general ecommerce fulfillment but has never run SFP is a real risk, because the tolerances are tighter than standard fulfillment and the failure mode is losing a badge rather than a late delivery. Our comparison of 3PL against FBA and self-fulfillment covers the general evaluation, and SFP adds these requirements on top rather than replacing them.
The Ecom Profit Box
Our collection of ecommerce growth resources, including the unit economics frameworks this decision needs.
Get It FreeModel It Before October
We do not sell fulfillment. If you are weighing SFP against FBA, we will look at your numbers and tell you honestly.
Book A CallThe Margin Math
Build it per SKU. The comparison only works at that level because SFP wins on specific product profiles rather than across a catalog.
That last line is where SFP analyzes go wrong. Every other term is a cost you can estimate. Badge risk is a probability, and it is not small, because the thresholds are tight and partly outside your control.
The asymmetry matters. Saving thirty cents a unit while carrying a real chance of losing Prime eligibility on those units is a poor trade, because the badge is worth considerably more than thirty cents in most categories. SFP has to win by a margin that compensates for the risk, not by a rounding error.
If your per-SKU economics are not built yet, that comes first. Our contribution margin playbook covers the structure and the true cost of an Amazon sale covers what belongs in the FBA column.
Which Products Actually Fit
The profile is narrow and worth stating precisely, because SFP is frequently attempted on catalogs it was never going to suit.
- Large or heavy items. Where FBA fulfillment fees and storage scale punishingly and your own freight rates compare well.
- Slow movers with steady demand. Products that would accumulate aged inventory surcharges in FBA but sit fine in your own warehouse.
- High-value items. Where you want control over handling and the fulfillment cost is small relative to unit price.
- Products you already stock for direct sales. The strongest case by far, since the warehouse, labor, and carrier relationships already exist.
- Made-to-order or customised goods. Which FBA cannot handle at all.
The profile that does not fit is small, fast-moving, low-value items. FBA prices those efficiently, they turn quickly enough to avoid storage penalties, and the per-unit shipping cost of sending them individually from your own warehouse rarely competes.
A mixed approach is normal and sensible. Running SFP on oversize items while keeping standard-size in FBA is a legitimate structure, helped by the fact that Amazon evaluates the size tiers separately. Our guide to omnichannel fulfillment strategy covers how those splits work alongside other channels.
Losing The Badge
Enrollment is the start of ongoing measurement rather than an achievement you keep.
Amazon reviews SFP performance weekly. Reporting indicates that sustained performance below the on-time delivery threshold for around three consecutive weeks results in the Prime badge being removed automatically, with no negotiation. Recovery means fixing the cause, demonstrating consistent improvement for two to four weeks, and then running the trial again, subject to the annual trial cap.
Reporting also indicates Amazon has moved toward more targeted enforcement, deactivating a poorly performing listing individually rather than disabling all seller-fulfilled offers at once. That is better than the alternative and it means a single problem SKU can be removed without taking the program with it.
The monitoring cadence that follows is twice weekly at minimum on on-time delivery and valid tracking. Rolling metrics with tight thresholds punish delayed detection, and a metric caught drifting is recoverable where a breached one is not.
Because these metrics interact with wider account standing, the general discipline in our guide to account health and suspension prevention applies here too, with tighter tolerances.
A Readiness Test
Answer these honestly before starting a trial. A no anywhere in the first five is a stop rather than a problem to solve later.
- Do you already ship 100 or more seller-fulfilled orders in 90 days with cancellation, tracking, and late shipment metrics clearing prequalification?
- Do you operate weekends, or can you before October 17?
- Can you display a one-day date on 40 percent of Prime page views for standard-size items from your current footprint?
- Is your inventory accuracy good enough to hold seller-initiated cancellations at or below 0.5 percent?
- Do your carriers produce tracking Amazon reads reliably? Test this before the trial, not during.
- Does the per-SKU math win by enough to compensate for badge risk rather than by pennies?
- Can you monitor twice weekly and act inside a week when something drifts?
When The Answer Is Just No
For most brands it is, and that is a finding rather than a failure. If you do not already run a fulfillment operation with weekend coverage and multi-region inventory, SFP in its current form is not accessible to you, and building toward it purely to earn a badge is a large investment justified by a benefit you can obtain by shipping to FBA instead.
The narrow case remains real. A brand with an existing direct-to-consumer warehouse network, oversize or slow-moving products FBA prices badly, and the operational discipline to hold tight metrics can make SFP work well. Everyone else should read the July speed change as Amazon signalling who the program is for, and take the signal. Rebuilding your FBA unit economics with the 2026 fee changes properly modeled is a better use of the same afternoon, and our FBA fee breakdown is the place to start.
What To Remember
- Weekends re-enter SFP speed metric evaluation on October 17, 2026 per industry reporting, which can push a compliant seller below threshold with no change to their operation.
- Reporting indicates the standard-size one-day requirement rose from 30 to 40 percent of Prime page views on July 6, 2026, which single-warehouse sellers are structurally unable to meet.
- Since 2023 Amazon measures on-time delivery, not on-time shipment, making carrier performance your performance and turning SFP into an operating model rather than a setting.
- Reported thresholds are 93.5 percent on-time delivery, 99 percent valid tracking, and 0.5 percent seller-initiated cancellation, with one source citing a lower 90 percent delivery figure. Operate to the stricter number.
- Trials are capped at three per year, which makes a failed attempt a loss of optionality rather than just a delay, so treat the trial as an audit of a proven operation.
- A 3PL can run SFP but the badge stays attached to your account, so their missed deliveries are your metrics and their SFP track record matters more than their general capability.
- Badge risk belongs in the margin comparison. Saving pennies per unit while risking Prime eligibility is a poor trade, because the badge is worth more than the saving in most categories.
Where This Came From
- Amazon, Seller Fulfilled Prime program page, and Fulfillment by Amazon, for public program context. Detailed requirements live in Seller Central behind a sign-in wall.
- Industry reporting on the July 6, 2026 delivery-speed increase from 30 to 40 percent one-day for standard-size items, Amazon's confirmation that compliant sellers need take no action, and the September 2026 zip-code-level delivery promise tool.
- Industry reporting on the exclusion of weekends from speed metric evaluation until October 17, 2026.
- Industry reporting on the October 2023 relaunch replacing on-time shipment and Buy Shipping obligations with a 93.5 percent on-time delivery requirement and initial speed thresholds of 30 percent one-day and 70 percent two-day.
- Industry reporting on the June 2025 policy update introducing a three-trial-per-year cap, a minimum of 100 monthly Prime shipments, and restricted graduation periods before major sales events.
- Industry reporting on prequalification thresholds, trial structure and length, valid tracking and cancellation requirements, Buy Shipping integration, 3PL eligibility, and automatic badge removal after sustained underperformance. Sources disagree on the on-time delivery threshold, with most citing 93.5 percent and one citing 90 percent, and on trial length, with Amazon's program page described as specifying 30 days against one source describing four full performance weeks. None of these figures are confirmed on Amazon-owned public pages.