The vehicle moved more money than the discount

Microsoft sells one product catalog through three contracts: the Enterprise Agreement locks, CSP flexes, and the Microsoft Customer Agreement never expires. The SKU prices are similar across all three, which is why buyers compare the wrong thing. In 40 to 55 agreement reviews, the lock, the flex, and the exit decided the bill.

Prepared by Redress Compliance · August 15, 2026 · Microsoft advisory. 40 to 55 agreement reviews, 2024 to 2026.

• Segment the estate by seat behavior first: stable base, variable population, and committed Azure, because each belongs on a different vehicle and the segmentation is the decision.
• Model all three vehicles on your real numbers, including the shelfware the EA lock would freeze and the flex CSP would only deliver if exercised, before any rate conversation.
• Give CSP a flex plan or do not go there: the 8 to 15 percent penalty in our reviews belonged to estates that bought flexibility and never used it.
• Put committed Azure on an MCA with a MACC sized to evidenced consumption, with the commitment sizing discipline applied before signature.
• Reconcile blended estates quarterly, because the 5 to 10 percent duplicate entitlement tax accrues silently wherever two order books cover one population.
• Engineer checkpoints into the MCA, since the evergreen term never hands you a renewal date; the MCA brief covers what replaces it.

Enterprise procurement is trained to compare products, and Microsoft has arranged things so there is no product to compare. E3 is E3 in every vehicle; the license is identical, the list price nearly so. What differs is everything around the license: how long you are held, when you can shrink, what happens at the end. Buyers who spend their evaluation comparing SKU rates are comparing the one dimension Microsoft has deliberately flattened, and ignoring the three it has not.

The review data shows what that misdirection costs. The estates pushed onto CSP paid for flexibility the way gym memberships pay for fitness: the capacity was real, the exercise never happened, and the 8 to 15 percent premium over a renegotiated EA bought a right nobody administered. The EA estates made the mirror image error, buying certainty and discovering it applies to mistakes too: 12 to 22 percent shelfware, frozen mid term, on two thirds of the estates we reviewed. Neither group chose a bad vehicle. Each chose a vehicle whose defining feature they had no operational plan to use.

The blend, which is where most large estates land, adds a third failure mode that neither single vehicle has: seams. Two order books covering one population will, absent active reconciliation, double cover part of it, and in our reviews that overlap ran 5 to 10 percent of seats, paid twice and used once. The blend is still usually right, because stable and variable populations genuinely belong on different paper. But a blend is a system, and a system needs an operator: a quarterly reconciliation, one owner, one entitlement view across both channels.

What makes 2026 the year to decide deliberately is that Microsoft is deciding for you otherwise. The EA floor keeps rising, the retirement wave keeps widening, and the MCA's evergreen structure quietly removes the renewal date that used to force a periodic reckoning. The vehicle question will be answered either by your model or by Microsoft's migration schedule, and only one of those prices your volatility correctly. Model all three on the estate you actually have, segment before you sign, and treat the renewal as the one date the whole structure can still be rearranged. The head to head arithmetic sits in the CSP versus EA comparison, and the wider library in the Microsoft practice.

Watch the briefing · 4:21How to Prepare for Your Microsoft EA Renewal in 2027Know which door you are walking through: the EA retirement wave, the MCA E and CSP paths, and the offers pattern that hides the vehicle decision.

Across 40 to 55 Microsoft agreement reviews, the vehicle layer carried the findings:

The patterns: flexibility purchased and never exercised, locks applied to volatile populations, and blends run without a reconciliation owner. In every case the vehicle decision had been made by default, by history, or by the seller, and never by a model of the estate's own behavior.

The buyer side move is to price the trade, not the SKU. The wider library sits in the Microsoft practice.

What is the difference between a Microsoft EA, CSP, and MCA?

Commitment versus flexibility. The Enterprise Agreement commits you for three years in exchange for volume pricing, with counts trued up annually and settled at renewal. CSP is the partner channel: monthly or annual subscriptions under New Commerce Experience terms, with the partner setting service and price inside Microsoft's rules. The Microsoft Customer Agreement is the evergreen digital contract Microsoft now prefers, signed direct or through a partner, and it underpins most new Azure buying.

When does the EA still make sense in 2026?

For large, stable estates that value price certainty over flexibility: headcount moving less than about 5 percent a year, a broad M365 platform base earning the deep tiers, and a finance team that wants a fixed three year number. Microsoft has pushed the practical floor toward 500 plus users and keeps tightening eligibility, so the population the EA fits is shrinking by design.

When does CSP beat the EA?

When headcount moves. CSP rarely wins the headline discount, but the ability to flex seats at term boundaries beats a deeper rate on seats you cannot shed. The caveat from our reviews: customers pushed onto CSP without a managed flex plan paid 8 to 15 percent more than a renegotiated EA would have cost, because the flexibility was never actually exercised.

Where does the Microsoft Customer Agreement fit?

It is where Microsoft is steering most buyers, and heavy committed Azure spend belongs there: an MCA with a MACC and a drawdown plan, not an aging EA commitment. The MCA is evergreen, which removes the renewal date, so price protection and review checkpoints have to be engineered into the deal rather than inherited from the term.

What does the wrong vehicle cost?

More than the discount debate. In our reviews, CSP without a flex plan ran 8 to 15 percent over a renegotiated EA, two thirds of EA estates carried 12 to 22 percent shelfware the lock had frozen, and mixed estates held duplicate entitlements on 5 to 10 percent of seats, paid twice and used once.

Should we run a single vehicle or a blend?

Most large estates end up blended, and the blend is where stranded licenses hide. The discipline is deliberate segmentation: stable base on the committed vehicle, variable population on the flexible one, Azure on the MCA with a MACC, and a quarterly reconciliation so the same seat never carries two entitlements.

How do we choose the vehicle?

From two inputs the quote never shows: your headcount volatility and your three year roadmap. Price the same estate under each vehicle, including the flex you would actually use and the shelfware each lock would freeze, and let the modeled totals decide. The SKU prices are similar across all three; the lock, the flex, and the exit are where the money moves.