On November 1 2025 Microsoft deleted the programmatic volume discount that rewarded seat count on Online Services, so a former Level-D enterprise now pays the same list price as a 500-seat buyer. This page quantifies the per-user delta by seat band and tells you which levers still recover the margin you just lost.
On November 1 2025 Microsoft deleted the programmatic volume discount that rewarded seat count on Online Services, so a former Level-D enterprise now pays the same list price as a 500-seat buyer. This page quantifies the per-user delta by seat band and tells you which levers still recover the margin you just lost.
Microsoft announced on August 12 2025, and enacted on November 1 2025, the elimination of the volume-based "waterfall" price levels (A through D) that historically discounted Online Services by seat count across the Enterprise Agreement (EA), the Microsoft Products and Services Agreement (MPSA), and the China-specific OSPA. The new floor for every commercial buyer is Level A, which Microsoft defines as the public list price shown on its website. In plain terms: whether you deploy 500 seats or 25,000, you now start negotiations from the same posted rate. The automatic reward for scale is gone.
The scope matters. This affects Microsoft 365, Dynamics 365, Windows 365, Azure, GitHub, and the security, compliance, and identity SKUs. On-premises software and Software Assurance keep their B, C, and D banding, so the pricing logic is now split down the middle of a single agreement. U.S. Government and Education price lists are excluded. Our full buyer guide to the EA list-price shift covers the program mechanics in depth; this page focuses narrowly on the dollar cost of losing your level.
Timing of impact is not uniform. If your EA started on or before October 1 2025, the no-discount rule only bites when, after October 31 2025, you add an Online Service you had not previously purchased. If your EA started on or after November 1 2025, every purchase prices at Level A. Products you already bought under an active term are locked in for the remainder of that term. That protection is real but temporary, and it is the single most important piece of leverage most buyers are failing to use. See our early-2026 price-lock window analysis for the exact renewal timing that preserves old pricing.
A former Level-D enterprise now pays exactly what a 500-seat buyer pays. Twenty-five years of volume being rewarded, deleted in one product-list change.
Before quantifying the delta, a warning that shapes every number below: published seat-band thresholds for the old A-D levels conflict across reputable sources, and you should treat any single schema as indicative rather than gospel. Your own historical EA order forms are the only authoritative record of what level you actually held and what discount it carried. Pull them before you model anything.
| Level | Common schema (seats) | Alternate schema (seats) | Typical discount off list |
|---|---|---|---|
| A | 500+ | 5,000-9,999 | 0% (list baseline) |
| B | 2,400+ | 10,000-24,999 | 3%-5% |
| C | 6,000+ | 25,000-49,999 | 6%-9% |
| D | 15,000+ | 50,000+ | up to 12% (some sources 18%) |
The table above reconciles three published schemas. The "common" column reflects the widely cited 500 / 2,400 / 6,000 / 15,000 breakpoints. The "alternate" column reflects a materially different set of thresholds reported by at least one source. The discount range is the most consistent element: from roughly 3% at Level B up to around 12% at Level D, with one outlier citing 18% at the top. In our negotiation experience across large EAs, the effective programmatic Level-D benefit typically sat in the 10% to 12% band, not 18%. Use 12% as your planning ceiling unless your order forms prove otherwise.
Strip out the July 2026 list increase for a moment and look only at the discount removal. Microsoft 365 E5 carried a list price of $57 per user per month before July 2026. A Level-D buyer receiving the full programmatic 12% paid approximately $50.16 at that layer. Removing the level restores the full $57. That is a $6.84 per user per month delta, or $82.08 per user per year, attributable solely to losing your level and before any negotiated discount is layered back on.
| Seat count | E5 annual at 12% (old Level D) | E5 annual at list (new Level A) | Annual delta from level loss |
|---|---|---|---|
| 2,400 | $1.44M | $1.64M | $197K |
| 6,000 | $3.61M | $4.10M | $492K |
| 15,000 | $9.03M | $10.26M | $1.23M |
| 25,000 | $15.05M | $17.10M | $2.05M |
These figures isolate the discount removal at the $57 list price and assume the full 12% Level-D benefit. The 25,000-seat delta of roughly $2.05 million per year aligns with the independently reported figure that discount removal alone adds about $2 million for a 25,000-user E5 estate. If your actual Level-D discount was 9% or 10%, scale the delta down proportionally. If you held Level C at 6% to 9%, the recurring exposure is roughly half of what a Level-D buyer faces per seat, but it is still six figures at scale.
Two million dollars a year, per 25,000 E5 seats, from the discount removal alone. That is before Microsoft's July 2026 list increase lands on top.
The discount removal does not arrive alone. Microsoft announced on December 4 2025 a list-price increase effective July 1 2026: E3 rises from $36 to $39 (8.3%) and E5 rises from $57 to $60 (5.3%). For a buyer coming off Level C or D, these two events compound. The list increase raises the number you were already paying, and the discount removal strips the reduction you used to apply to it. The combined effect on a Level-D renewal typically lands at 15% to 20% above the prior rate, not the 5% to 8% the list increase alone implies. Our deep dive on the compounding math walks the arithmetic step by step.
The 25,000-seat E5 worked example makes it concrete. A buyer who renewed before November 2025 at Level D paid roughly $15 million annually. The same buyer renewing after July 2026 without negotiated discounts pays approximately $18 million, close to $3 million more per year, or about 20% higher. The decomposition: the list increase from $57 to $60 adds about $900,000, and the discount removal adds roughly $2 million. Neither change in isolation predicts the total. That is the trap for finance teams who budgeted for an 8% increase.
E3 buyers get hit harder in percentage terms because the 8.3% list increase exceeds E5's 5.3%. A 25,000-user E3 estate with former Level-D pricing rises from about $9.5 million to $11.7 million, an increase near $2.2 million or 23%. If you are still on E3 and assumed you were insulated from the E5 headlines, you are not. Model your own SKU mix; do not extrapolate from the E5 example.
The in-term protection is narrower than it reads. Any Online Service you add after November 1 2025 that was not already in your agreement prices immediately at Level A, even mid-term. This is the new-SKU trigger trap. A common failure pattern: a buyer with locked E3 pricing adds a Copilot or a Defender SKU to a subset of users, and that addition prices at full list with no level benefit, even though the base estate is protected. Treat every mid-term addition as a full-list purchase and budget accordingly.
The programmatic discount is dead, but negotiated discount is not. Microsoft still grants deal-specific concessions; the burden has simply shifted from automatic to earned. In our practice, the levers that actually move the realized rate on a post-2025 EA are commitment growth, Copilot attach, multi-year term, and competitive tension (genuine willingness to shift workloads to Google Workspace or to CSP). Enterprise buyers at 1,000-plus seats are still landing E5 in the $42 to $50 range against the $57 list, which is a negotiated discount of 12% to 26%. That range now has to be manufactured deal by deal rather than inherited from your seat count.
Concrete steps, in priority order. First, exercise the price-lock window: if you can renew or add committed SKUs before your term boundary and before July 1 2026, you preserve both the old list price and, where in-term, existing discounts. Second, benchmark hard against the new baseline so you know what a defensible negotiated rate looks like now that everyone starts at list; our list-price benchmarking guide gives the reference points. Third, treat Copilot as a negotiation asset, not just a cost, because Microsoft's field is heavily incentivized on attach and will trade base-SKU discount for it. Fourth, review whether CSP or MCA-E delivers a better realized rate than a renewed EA for your specific mix; the EA's structural advantage was always the volume levels, and those are gone. Fifth, build the recovery playbook from how to manufacture discount now that volume levels are gone.
For the strategic overview of every Microsoft licensing change across 2025 and 2026, including the cloud commitment floors and Copilot economics that interact with this change, see the 2025-2026 licensing guide. If your immediate question is simply the aggregate cost impact by seat band, the volume discount removal impact analysis is the fastest reference.
Pull your last three EA order forms and confirm your actual level and discount percentage. Do not assume 12%. Model your specific SKU mix at both the current list and the July 1 2026 list, applying zero programmatic discount, to establish your true renewal exposure. Identify every Online Service you plan to add in the next 18 months and price it at full list today, because that is what you will pay if you add it mid-term. Then map your term boundary against July 1 2026: if renewing before that date preserves meaningful pricing, that decision has a hard deadline and a quantifiable value. The buyers who lose most are the ones who budget for the 8% headline and discover the 20% reality at signature.
This change removed a lever Microsoft controlled and handed the remaining leverage to buyers who prepare. The seat count that used to earn discount automatically now only earns it if you make it part of a negotiated deal. That is more work, but it is not less discount, provided you show up with the model, the benchmark, and a credible alternative.
Only the automatic, programmatic price levels (A-D) on Online Services were removed, effective November 1 2025. On-premises software and Software Assurance keep their B, C, and D banding. Negotiated, deal-specific discounts still exist; they now have to be earned rather than inherited from seat count.
Isolating the discount removal at the $57 list price and assuming a full 12% Level-D benefit, the delta is about $6.84 per user per month, or roughly $82 per user per year. For a 25,000-seat estate that is approximately $2 million per year from the discount removal alone, before the July 2026 list increase.
Partly. Pricing for Online Services you already purchased is locked for the rest of your current term. But any new service you add after November 1 2025 that was not already in your agreement prices at full Level A immediately, even mid-term, and all protection ends at renewal.
Two events compound. The July 1 2026 list increase (E3 up 8.3%, E5 up 5.3%) raises the base price, and the discount removal strips the reduction you used to apply. For a former Level-C or Level-D buyer the combined effect typically lands at 15% to 20% above the prior rate.
No. The change applies to commercial organizations worldwide, with exclusions for U.S. Government and Education price lists. On-premises software pricing is also unaffected.
Not automatically. The EA's structural advantage on Online Services was the volume banding, which no longer exists. CSP and MCA-E may now deliver a better realized rate for some mixes, so benchmark all three against the new list-price baseline before renewing.
White paper: what a Copilot task really costs in dollars, the pre-purchase credit tiers, and the same work priced three ways for buyers.
Gated with a work email on the download page. No sales follow up you did not ask for.
Get the White Paper →
500+ enterprise clients. 11 vendor practices. Industry recognized. One conversation can change what you pay for the next three years.
One buyer side briefing a week. Renewal signals, audit moves, and the levers that work. No vendor spin.
