Copilot, Gemini, or Amazon Q: The Buyer Side Decision for 2026
At June 2026 list, Microsoft 365 Copilot and Gemini Enterprise both price at $30 per user per month and Amazon Q Business Pro at $20. The winner is decided by your incumbent suite renewal, not the AI evaluation.
Prepared by Redress Compliance · June 2026 · Representative 10,000 seat estate scenario (benchmark scenario, not a quote)
Executive Summary
The enterprise AI assistant market has converged on one metric: per user per month. Microsoft 365 Copilot lists at $30, Gemini Enterprise at $30, and Amazon Q Business Pro at $20. At list, the three look interchangeable. In practice the prerequisite stack underneath each seat decides the real cost, and the incumbent suite renewal decides the real discount.
The full Copilot stack runs $66 per user per month once the Microsoft 365 E3 prerequisite at $36 is counted. Google moved the other way in January 2025: it retired the paid Gemini add ons and folded the assistant into every Workspace plan, repricing the base by roughly $2 per user per month.
Amazon attacks from below with Q Business at $3 Lite and $20 Pro, with no productivity suite prerequisite at all.
The waste pattern is consistent. In the blanket assistant rollouts we audited in 2024 to 2025, 40 to 60 percent of deployed seats showed low weekly active use. Buyers who staged the rollout, gated expansion on measured adoption, and held competing quotes landed 15 to 30 percent below the opening proposal.
The bundled discount on a blanket deal never paid for the idle seats it created.
This paper delivers the negotiation framework: the verified entitlement baseline, the prerequisite stack scoping, the EA, EDP, and CUD bundle math, the capability ladder, the five contract clauses that protect the budget, the counter moves to Microsoft's standard tactics, and the BATNA and side letter language we use.
Your deadline is your suite renewal date. Everything in this paper prices best inside that window.
Why the Assistant Decision Is Decided at the Suite Renewal
None of the three vendors sells the assistant as a standalone product in practice. Copilot defends the Microsoft 365 seat base, Gemini defends Workspace, and Amazon Q deepens the AWS commitment. Each vendor discounts the assistant in proportion to the estate it protects, not the value it delivers.
That is why the negotiation cycle matters more than the feature matrix. The vendor is most motivated when the suite renewal is open and a competitor's quote is on the table. Negotiate the assistant mid term, with no renewal leverage, and the quoted discount collapses toward zero.
The buyer side framework in this paper runs five moves in sequence. Baseline what you own, scope the prerequisite stack, separate the bundle math, stage the rollout on a capability ladder, and close with the five protective clauses. Each move earns the leverage the next one spends.
The Per User Metric and the Prerequisite Stack at June 2026 List
The headline rates converge. The stacks underneath do not.
| Assistant | List per user per month | Prerequisite stack | Commitment mechanic |
|---|---|---|---|
| Microsoft 365 Copilot | $30 | Microsoft 365 E3 ($36), E5 ($57), or a Business suite | Annual commitment per seat, even when billed monthly |
| Gemini in Workspace | $0 incremental | Any paid Workspace plan; base prices raised about $2 in January 2025 | Folded into the suite; cannot be detached at renewal |
| Gemini Enterprise | $30 ($21 Business tier) | Google Cloud account; Workspace not strictly required | Annual subscription, agent platform included |
| Amazon Q Business Pro | $20 (Lite $3) | AWS account and IAM Identity Center; no suite prerequisite | Monthly per user, tier assignable per user each month |
| Amazon Q Developer Pro | $19 | AWS account; targets the developer population only | Monthly per user subscription |
Source list prices: Microsoft 365 enterprise plans, Google Workspace pricing, Gemini Enterprise pricing, and Amazon Q Business pricing.
Three mechanics hide inside that table. First, a Copilot seat is an annual commitment even when billed monthly, so a seat added in month two is owed for twelve months regardless of use.
Second, Google's fold in cannot be unattached: the roughly $2 uplift is now permanent base price, compounding with every seat you add. Third, Amazon Q tiers are assignable per user per month, the only metric of the three that flexes downward without a negotiation.
Move One: Build the Verified Entitlement Baseline
Every assistant negotiation we have supported started with the same discovery: the customer did not know what AI entitlements it already owned. Microsoft ships Copilot Chat at no charge to Microsoft 365 subscribers.
Google now includes Gemini in every paid Workspace plan. Security and audit features the reseller quotes as upsells are often already in the E5 or Enterprise Plus stack.
The verified entitlement baseline is the document that survives Microsoft scrutiny. It reconciles three sources: the enrollment and order paperwork, the tenant admin center assignments, and measured usage telemetry. Where the three disagree, the paperwork governs the negotiation and the telemetry governs the sizing.
| Baseline element | Source of record | What it proves in the negotiation |
|---|---|---|
| Suite entitlements per user | EA enrollment, MCA orders, Workspace and AWS billing consoles | What you already pay for, including bundled AI you have not deployed |
| Assigned versus active seats | Tenant admin centers and identity provider groups | The gap between assigned licenses and people who use them |
| Weekly active assistant use | Copilot dashboard, Workspace reports, Q analytics | The defensible seat count for any expansion tranche |
| Agent and consumption charges | Azure and AWS billing exports, Copilot Studio meters | The spend the per user price quietly excludes |
The baseline has a second job: audit defense. The same reconciliation that prices your renewal also answers a vendor compliance review without scramble. A baseline the vendor cannot puncture is what turns your usage data from their sales asset into your negotiation asset.
Move Two: Scope the Prerequisite Stack Before Pricing the Assistant
The assistant price is the visible third of the bill. Copilot requires a qualifying Microsoft 365 suite, so an estate sitting on E3 pays $66 per user per month all in, and an account team will use the Copilot conversation to push E3 to E5 moves worth more than the assistant itself.
Scope the stack in writing before any assistant pricing discussion. Which users hold E3 versus E5, what Workspace edition each population runs, and which workloads already sit inside an AWS identity perimeter. Then price the assistant against the stack you have, not the stack the vendor proposes.
The reverse move matters as much. Do not let an assistant decision trigger a suite upgrade that was not independently justified. In the engagements we benchmarked, the E5 uplift attached to a Copilot deal was the single largest avoidable line, exceeding the Copilot fee itself in roughly one estate in three.
Move Three: The EA, EDP, and CUD Bundle Math
Each vendor wants the assistant inside its biggest commercial vehicle. Microsoft prices Copilot inside the Enterprise Agreement, Amazon retires Q spend against the Enterprise Discount Program commitment, and Google trades Workspace and Gemini Enterprise terms alongside committed use discounts on Google Cloud. The bundle is where the discount lives, and where the lock in compounds.
The mechanics differ in ways the quotes will not volunteer. Copilot seats inside an EA enrollment are added through the True Up, which only ever adds; reductions wait for the enrollment renewal.
Q Business subscriptions bill through the AWS account, so an unspent EDP commitment makes Q look free in budget terms while it accelerates your committed burn. Copilot Studio agent consumption can decrement an Azure MACC, blending the seat decision into the cloud commitment.
Bundle math also sets the exit cost. An assistant bought inside a three year EA, an EDP, or a CUD inherits that vehicle's term. The cheapest assistant at signature can be the most expensive to leave. Model the exit scenario before signing, because section 8 shows it is also your strongest discount scenario.
Move Four: The Capability Ladder and the Staged Rollout
The three catalogs are ladders, not single products. Microsoft runs from free Copilot Chat to the $30 seat to metered Copilot Studio agents. Google runs from bundled Gemini in Workspace to Gemini Enterprise at $21 and $30. Amazon runs from Q Lite at $3 to Q Pro at $20, with Q Developer at $19 for the engineering population.
The ladder discipline is to place each user population on the cheapest rung that covers its measured need, and to promote rungs only on usage evidence.
Most estates resolve to three populations: a small power user tier that justifies the $30 seat, a broad tier served by the bundled or Lite rung, and a developer tier that belongs on Q Developer or equivalent.
The worked estate: blanket versus staged
Take a representative financial services firm with 10,000 knowledge workers on Microsoft 365 E3, evaluating Copilot (benchmark scenario, not a quote). The account team proposes all 10,000 seats at list from month one. The staged alternative pilots 2,000 seats for six months, then expands to 5,000 once adoption clears the gate.
| Rollout option | Seats and timing | Year one cost | Versus blanket |
|---|---|---|---|
| Blanket deployment | 10,000 seats x $30 x 12 months | $3,600,000 | Baseline |
| Staged rollout | 2,000 seats x $30 x 6 months, then 5,000 seats x $30 x 6 months | $1,260,000 | 65% lower |
| Year one difference | Avoided idle seat spend | $2,340,000 | Funds the next phase |
Across the AI assistant deals we benchmarked in 2024 to 2025, buyers who staged the rollout, gated seats on demonstrated usage, and held competing quotes recovered 15 to 30 percent against the opening proposal.
In the estates we audited, 40 to 60 percent of assistant seats deployed estate wide showed low weekly active use after the first quarter. Benchmark ranges: Redress Compliance advisory engagement file, 2024 to 2025.
Move Five: The Five Contract Clauses That Protect the Budget
The discount headlines the deal. The clauses decide whether the budget survives years two and three. These are the five we negotiate into every assistant commitment, in the order they typically trade.
| Clause | What it secures | Negotiation anchor |
|---|---|---|
| 1. Pilot price hold | The pilot discount carries to full deployment for at least 24 months | Vendors fund pilots to anchor list at scale; remove the anchor |
| 2. Anniversary reduction right | Reduce assistant seats at each anniversary without repricing the remainder | True Up adds only; without this clause reductions wait for renewal |
| 3. Adoption gate | Expansion tranches trigger only when weekly active use clears an agreed bar | Your usage telemetry, from the Move One baseline |
| 4. Scope statement | What the per user fee includes: agent runs, Studio meters, premium models | Consumption meters sit outside the $30 seat unless captured |
| 5. Bundle firewall | Suite discounts are not contingent on assistant attach or growth | Protects the renewal baseline when the assistant underdelivers |
Clause four deserves the closest read. The per user price is not the ceiling of assistant spend. Copilot Studio agents meter messages and capacity packs outside the seat fee, and agent consumption can be billed against Azure. The scope statement caps that exposure or it compounds silently.
Microsoft's Standard Tactics and the Counter Moves
Microsoft runs the most disciplined assistant sales motion of the three, and it is predictable. Each tactic has a counter that works because it is documented, not argued.
| Standard tactic | What it does | The counter move |
|---|---|---|
| Quarter end discount expiry | Compresses your evaluation into Microsoft's December and June close | Open the negotiation early; let their deadline price your patience |
| All seats or no discount | Makes the discount contingent on a blanket commitment | The staged rollout with a pilot price hold, clause one |
| E5 uplift attach | Uses the Copilot conversation to sell a suite upgrade | Move Two: price the assistant against the stack you have |
| Agents are included framing | Leaves Studio meters and consumption outside the quoted price | The scope statement, clause four, with a consumption cap |
| Your stack will not fit Gemini or Q | Dismisses alternatives before you have priced them | Written cross quotes and a segmented BATNA, section 9 |
What the discounts actually settle at
The discount benchmarks below come from the assistant negotiations we supported and benchmarked in 2024 to 2025, across renewal and exit scenarios. They are achieved outcomes against the vendor's opening proposal, not list discount claims.
| Scenario | Achieved outcome versus opening proposal |
|---|---|
| Renewal, no competing quote | 0 to 8% |
| Renewal, two competing assistant quotes in writing | 12 to 22% |
| Staged rollout, usage gates, competing quotes | 15 to 30% |
| Funded exit scenario, workload portability proven | 25 to 40% |
Read the gradient, not the bars. Every step up the chart is a leverage artifact you control: a second written quote, a usage gate, a proven exit path. None of it requires actually switching vendors. It requires Microsoft believing you could.
BATNA Construction and the Side Letter Language
A whole estate migration threat is not credible, and the account team knows it. A segmented BATNA is.
The credible alternative is built population by population: the developer tier priced on Amazon Q Developer at $19, an AWS centric business unit piloted on Q Business Pro at $20, a Workspace subsidiary already holding bundled Gemini, an agent workload quoted on Gemini Enterprise at $30.
Each segment gets a written competing quote, even where you expect to stay. The quotes are not theater; they are the pricing instrument the discount benchmarks in section 8 depend on. The vendor prices the probability of loss per segment, and small credible losses move price more than one large hollow threat.
The concessions then land in a side letter or amendment, in contract language rather than account team goodwill. Three examples of the language pattern we use:
Anniversary reduction: "Customer may reduce its committed assistant seat count by up to 30 percent at each enrollment anniversary, without repricing of, or retroactive charge against, the remaining seats."
Adoption gate: "Tranche two seats activate upon Customer confirmation that weekly active usage across tranche one has met or exceeded 60 percent for two consecutive months, and not before."
Adapt the percentages to your estate; the structure is the point. Every clause is measurable, dated, and survives the account team's rotation. A discount in an email thread does not.
Verify entitlements and usage
Build the verified entitlement baseline. Reconcile paperwork, tenant assignments, and telemetry. Classify the estate into power, broad, and developer tiers.
Build the segmented BATNA
Collect written quotes across the three catalogs per segment. Scope the prerequisite stack and decompose every bundled figure into standalone prices.
Trade structure for price
Table the five clauses before the discount conversation. Stage the rollout, set the gates, and close inside the suite renewal window at the vendor's quarter end.
Our recommendation: pick the assistant with the ladder, not the demo. Stage the rollout, gate it on usage, and spend your leverage at the suite renewal where all three vendors price to defend the estate.
- Before any commitment: build the verified entitlement baseline and the segmented BATNA. An estate that knows its own usage and holds two written quotes starts 15 to 30 percent ahead of one that evaluates features.
- In the negotiation: secure the pilot price hold, the anniversary reduction right, and the scope statement before discussing seat count. Structure concedes cheaply early and expensively late.
Redress Compliance is a 100 percent buyer side advisory firm with 500+ enterprise clients and more than $2B under advisory across 11 vendor practices. If a Copilot, Gemini, or Amazon Q decision is on your desk, contact us or visit our Microsoft practice before you sign. We are glad to tie a meaningful part of the fee to delivered value.
