Google's Post Discount Period clause quietly returns 100% of a 28 to 35% commit discount to list on the day your term ends, and it is the one redline most buyers never table

Google's standard commit paper reverts every negotiated rate to then-current list the moment the Discount Period lapses, which converts a $10M three-year commitment into a $14M run rate overnight if renewal slips a quarter. Workspace paper carries the same defect in a different shape: the March 2025 reset moved Standard from $12 to $18 per user per month with no opt-out, and nothing in the base agreement stops Google doing it again mid-term. Clause construction, not timing, decides which of those two outcomes you inherit.

Prepared by Redress Compliance · August 22, 2026 · Google Cloud advisory. EDP, CUD, and Workspace renewal engagements 2024 to 2026.

Every Google commit negotiation eventually collapses into nine pieces of drafting, and the rate card is not one of them. You will win 28 to 35% on a three-year $10M GCP commitment because Google is buying share against AWS and Azure, and it will hand that number over faster than either competitor.

What Google protects instead is the paper that governs what happens to that discount at the edges: when the term ends, when you undershoot, when a new meter switches on, when a SKU gets repackaged, and when you want out.

Those clauses are where the vendor recovers the margin it conceded on the headline rate.

Treat the discount as settled early and spend your negotiating capital on the nine below, because a 32% discount with a hard reversion clause and no true-down is worth materially less over 36 months than a 28% discount with month-to-month continuation and a 15% flex band.

Google's standard language is short enough to miss on a first read: if Customer continues to use the Services after the Discount Period, Customer will automatically move to then-current list prices. That single sentence undoes the entire commercial negotiation on a date certain.

At $10M of annual committed spend closed at a 32% effective discount, your true list exposure is roughly $14.7M.

The day the Discount Period lapses, that is what you pay, and you pay it against then-current list, not the list that existed when you signed, so three years of Google price movement lands on you at the same moment. A renewal that slips one quarter costs approximately $1.2M in reverted spend.

A renewal that slips two quarters because a CFO change or a reorg pushed the approval cycle costs $2.4M, and you will pay it before anyone in procurement notices the invoice shape has changed.

The reason this is the highest-value single redline on Google paper is the asymmetry of the outcome. Every other clause moves your economics by percentage points. This one moves them by 100% of the discount.

It also inverts your renewal leverage entirely: on the day you most need to negotiate hard, you are already paying list, which means Google's downside from a slow renewal is zero and yours is a seven-figure run rate increase.

NPI's position is that every enterprise customer should challenge this clause, particularly before committing significant spend, and that both AWS and Google have shown limited flexibility in amending it. Limited is not zero, but it does mean you cannot table this as a week-four cleanup item.

It goes on the table in the first session, with a walk-adjacent posture and a stated position that you will not sign a commit of this size against an automatic reversion.

The achievable ask is not permanent discount survival, which Google will refuse. It is continuity plus a defined runway.

Draft it as: discounts and special pricing continue on a month-to-month basis while the parties negotiate in good faith on a new Discount Term, and if no agreement is reached within a stated number of days, the discounts hold for a defined transition period regardless of monthly spend volume.

That last qualifier matters, because Google's deal desk will try to condition continuation on maintaining a monthly consumption floor, which reintroduces the same cliff in a different shape.

A strong outcome is month-to-month continuation with a 180-day transition; an acceptable one is 90 days with an express carve-out that the transition rate cannot exceed your final in-term rate.

Anything less and you are pricing the renewal from a position of zero leverage, which is precisely the outcome the exit rights you build at signature are supposed to prevent.

Google's commercial team will not raise shortfall treatment. It is not in the deck, it is not in the term sheet summary, and the rep will happily let you sign a three-year commit with a payment obligation for the unspent balance because that is the default posture and nobody asked.

That silence is the tell: shortfall on private agreements is handled on a negotiated basis, and we routinely see two entirely different constructions in paper signed the same quarter. One customer owes cash for every dollar under the number.

The next has an automatic commitment reduction right that trues the commit down to actuals with no invoice. Same vendor, same product, same fiscal year. The variance is not driven by spend size, it is driven by which buyer asked.

Google is buying share in this market, closing a $10M three-year GCP commit at 28 to 35% effective discount against comparable AWS and Azure paper, and a vendor buying share funds flexibility out of the discount line before it funds it out of the walk-away risk.

Set the ask at a year-three true-down of 20 to 30% tied to defined, objective triggers: divestiture of a named business unit, a migration of a named workload to another provider following a documented service failure, or a decline in a stated internal volume metric.

Triggers matter because an unconditional true-down reads as a request to unwind the commitment and Google will refuse it on principle.

Separately, cap the shortfall remedy at the discount differential, meaning you repay only the discount value on consumed spend recalculated at a lower band, not the full unspent commit. On a $10M commit with $7M consumed at 30% off, that is roughly $700K of exposure instead of $3M.

Google's counter will be to accept the differential cap and refuse the true-down, or to offer a one-time reallocation across years rather than a reduction. Take the differential cap first; it is the larger dollar item.

When true-down is genuinely refused (and it is, in perhaps half the deals we see), ramp shaping is the structural substitute and it is easier to win because it does not change Google's total contract value.

Push years one and two to 20% and 30% of the commit, leaving 50% in year three, and demand the discount tier be fixed at the aggregate three-year level rather than recalculated annually.

That combination converts shortfall risk into a timing problem you control, and it buys you the option to renegotiate before the year-three balloon lands. Pair it with a written rollover of unused year-one spend, which Google grants more readily than a reduction because the money still arrives.

The drawdown definition is worth more than two points of headline discount and almost nobody redlines it. Two points on a $10M three-year commit is $600K over the term.

A drawdown definition that excludes Marketplace, support, and CUD-covered consumption at full retail can leave you $2M short of the number on identical actual spend, which triggers the shortfall clause you just negotiated. The discount is what you pay per unit.

The drawdown definition is whether you hit the number at all, and the second question determines whether the first one ever matters.

Google's standard paper is deliberately thin here: it says qualifying spend without enumerating what qualifies, which means the answer is settled later by a billing team you will never meet.

The single highest-value item is written confirmation that CUD-covered consumption draws down at full retail value, not net of the CUD. Resource-based commitments discount eligible Compute Engine hardware up to 55%, and up to 70% on memory-optimized series.

If a workload running at 55% off draws down at the net figure, you need more than twice the consumption to retire the same commit dollar, and every CUD you buy actively works against your commit attainment. Get it in the schedule, in writing, before signature.

The same logic applies to Marketplace: insist third-party software and data purchases count at 100% of invoiced value, because Marketplace is where large estates find the last 15% of a commit in the final quarter, and a 50% counting rate quietly removes that escape hatch.

Our breakdown of CUD negotiation tactics covers how the stacking works underneath the commit.

Draft it as an enumerated inclusive list in the schedule with a catch-all covering any future SKU billed by Google to Customer or its affiliates.

The catch-all is doing real work: Gemini Enterprise seats, Agent Gateway metering effective July 2026, Sessions and Memory Bank billing from September 2026, none of which existed when most current commits were drafted.

Google's counter will be to accept the enumeration and strike the catch-all, arguing forward SKUs need separate commercial treatment. Hold the catch-all and trade the professional services line instead; PS is the smallest number on the table and the easiest concession to look generous with.

The March 2025 reset is the only Workspace case study you need, and you should put it on the table in the first meeting. Starter went from $6 to $8, Standard from $12 to $18, Plus from $22 to $28 per user per month, effective March 17, 2025 or at next renewal, whichever came first.

That is a 17 to 22 percent list increase, delivered mid-relationship, with no opt-out and no way to buy Standard or above without Gemini bundled in. Note the mechanic carefully, because it defeats the redline most buyers write: Google did not raise the price of the SKU you bought.

It changed what the SKU contains and priced the new contents.

A clause that says "the per-seat rate for Business Standard is fixed at $12 for 36 months" survives right up to the moment Google stops selling Business Standard in the form you bought it, which is exactly what happened to the standalone Gemini for Workspace SKUs (last billing day January 31, 2025.

Pro-rated credits issued in February 2025, then the base tiers absorbed the capability at a higher price six weeks later).

So the ask is three-part and you should table all three together. First, a fixed per-seat rate for every tier in your estate for the full term, written into the order form.

Second, a renewal uplift cap at the lower of CPI or 3 percent, applied to your net rate rather than to then-current list, because an uplift on a list price Google controls is not a cap.

Third, and this is the clause that does the work, an entitlement-scope provision stating that your rate attaches to the functional capability delivered on the effective date, and that any successor, replacement.

Or repackaged SKU delivering equivalent capability inherits the same rate for the balance of the term.

Google will push back hardest on the third, and it will offer the first two readily, which tells you where the money is. Our companion pieces on price hold and uplift cap construction and on AI repackaging protection carry the drafting detail.

Expect Google to counter with a capability-change notice period and credits; treat that as an opening, not an answer, and read it alongside the broader AI contract red lines that apply across every vendor doing this.

A three-year $10M GCP commitment routinely closes at 28 to 35 percent effective discount, materially deeper than a comparable AWS EDP or Azure MACC. Buyers read that number as generosity or as competitive desperation. It is neither.

It is a share-purchase strategy with a very specific accounting logic behind it, and understanding that logic is the difference between spending your leverage well and spending it on nothing.

Consider what the discount actually costs Google. It is reversible: the Post Discount Period clause returns every point of it to then-current list on the day the Discount Period ends, which means the 32 percent you won is a 36-month rental, not a purchase.

It is recoverable: as the Workspace side demonstrates, Google can restructure packaging and reprice capability without touching your negotiated rate at all. And it is invisible where it matters, because Google's internal metric is committed spend and workload share, not gross margin per account.

A deep discount that locks a $10M three-year commit moves the number the field is compensated on. The concession is cheap in the only currency Google's organization actually counts.

Now consider what the clauses cost. A true-down right converts a fixed forward revenue number into a variable one. A termination for convenience right converts it into an option Google wrote and you hold. An entitlement-scope clause forecloses the repackaging lever entirely for the term.

Each of these attacks revenue certainty, which is the thing Google is buying with the discount in the first place. That asymmetry, cheap rate and expensive structure, is the entire negotiation, and it runs in exactly the opposite direction to how most buyers sequence their asks.

The overtrade pattern is remarkably consistent in our practice.

A buyer arrives with genuine competitive leverage (an AWS proposal, an Azure MACC in hand, a workload that genuinely could move), spends it pushing 29 percent to 33 percent, signs standard paper because the discount felt like a win.

And then discovers at month 30 that four extra points on a $10M commit is roughly $400K over three years while the Post Discount Period reversion is worth $4M on a single renewal slip.

The discount was the only thing secured. Everything structural was left on Google's paper, unamended, because it never got tabled.

The Workspace record makes this concrete. Google retired the standalone Gemini SKUs on January 31, 2025, issued pro-rated credits in February, and reset base tier pricing on March 17.

That is a vendor demonstrating, in public, over seven weeks, that it will restructure packaging unilaterally and that credits are its chosen remedy. Credits are a remedy of last resort dressed as a right. Nothing in the base agreement prevented any of it, and nothing prevents a repeat.

If Google will do that to a bundled productivity suite where switching costs are high, assume the same instinct applies to every AI meter that switches on mid-term in your commit.

The sequencing implication is direct. Table the structural clause set (Post Discount Period, true-down, drawdown definition, entitlement scope, exit and swap rights) in the first meeting, before any number is discussed, and hold the discount conversation until the clause set is agreed in principle.

Google will resist the reordering, which is itself informative.

And treat a hard refusal to move on Post Discount Period as pricing information rather than as policy: it tells you precisely what Google's account team expects your renewal leverage to look like in 36 months, and they have modeled it more carefully than you have.

The seat prices are the part of the AI layer buyers argue about, and they are the part that matters least. Gemini Enterprise Business at $21 per user per month is not a price Google arrived at through cost modeling.

It is Microsoft 365 Copilot Business at $21.00, matched to the cent, which tells you exactly how much room sits behind it: Google set that number to win a comparison, and it will move on it when you show a live Copilot quote.

The editions above ($30 and up through the $60+ Plus tier), plus Gemini Code Assist at $19 per user per month and the AI Expanded Access add-on at roughly $20, are all seat lines you can benchmark, discount, and true down.

What you cannot benchmark, and what will actually blow the budget, is the metered layer sitting underneath every one of those seats: token-rate API calls on top of Code Assist, consumption charges beyond the Gemini Enterprise seat quota.

And a set of meters that switch on by calendar date whether or not you have modeled them.

Look at the dates. Agent Gateway billing became effective July 13, 2026. Memory Bank and Sessions billing commenced September 1, 2026. Neither of those was in the run rate of a deal signed in early 2026, and neither required your consent.

That is the real exposure: a consumption schedule that lets Google introduce or reprice a meter mid-term converts your negotiated seat discount into a rounding error. Grounding makes the point most sharply. Web grounding runs $45 per 1,000 grounded prompts.

Grounding against your own data runs $2.50 per 1,000 requests. Same feature, eighteen times the price, and the routing decision is made by an agent configuration your platform team controls loosely at best.

Ten thousand users at fifty grounded web prompts a month is $270,000 a year that never appeared in the business case.

Ask for three things and treat them as one package, because Google's account team will happily concede the softest and call it done. First, capped unit rates in a named consumption schedule attached to the order form, not incorporated by URL reference.

Second, 90 days written notice before any new meter starts billing or any listed rate changes, with the notice period running from receipt, not from a page update.

Third, an express right to disable any metered feature at any time without penalty, without losing the seat discount, and without triggering a shortfall recalculation.

Google's counter will be a cap on the rate but not on the volume, or a notice provision that excludes "new services." Refuse both: a new meter attached to a feature you already licensed is a repricing, and the drafting should say so.

The same discipline applies across the AI stack, and the broader logic is set out in our work on short, capped, portable cloud AI commitments and the specific levers in Google Cloud AI contract negotiation.

A strong outcome looks like this: seat rates 25 to 45% below list at enterprise scale, every meter in the table fixed for the term, and a written right to switch a meter off in thirty days. That last right is worth more than another two points on the seat price.

Treat these three as one negotiation, because they are three answers to a single question: what happens to a $10M three-year commit when the business stops looking like the business that signed it. Divestiture, acquisition, a workload migration, a platform decision reversed by a new CIO.

Google's paper answers that question with silence, which means the commit survives the change and the buyer eats it. Negotiate them separately and you will win the cheapest one and be told the other two are non-standard.

Termination for convenience is the one you will not get outright, and you should not spend capital chasing it. In twenty-five years across this table, unrestricted TFC on a discounted multi-year commit has been granted perhaps twice, both times to customers who were paying near list.

What is genuinely achievable is partial termination tied to a defined event: a divestiture of a business unit above a stated revenue or headcount threshold, with the commit reduced pro rata to the divested consumption over the trailing twelve months.

Google will counter with a "reduction subject to mutual agreement," which is not a right. Push for a formula and a ninety day notice window. Our sub-article on termination for convenience walks the drafting.

Swap and substitution rights are the practical substitute, and they are where Google is most flexible because nothing leaves the building.

A right to reallocate committed dollars across service families (compute to BigQuery, storage to the AI platform, GCP to Workspace where the paper allows) preserves Google's revenue while removing your stranded-spend risk.

Ask for unrestricted swap within GCP, then push for cross-family movement into Workspace and Gemini seats. Google concedes the first readily and the second at scale. The sub-article on swap rights covers the mechanics; the wider exit architecture sits in our Google Cloud exit rights guidance.

Change of control is the clause everyone forgets and the acquirer inherits. Without an express carve-out, a buyer of your business acquires a commit sized to your consumption and priced to your discount, usable only if it adopts your platform. That is a real diligence deduction.

Ask for assignment on notice to an affiliate or acquirer, plus a right to terminate the unconsumed commit within ninety days of closing on payment of a capped percentage, twenty five percent of remaining is a defensible landing point. See the assignment and change of control sub-article.

Route selection is a clause decision before it is a price decision, and most buyers get the order backwards. Buying through a Google partner does not give you Google's contract with a discount on top.

It gives you the reseller's contract, into which some Google terms flow down verbatim (acceptable use, service levels.

Data processing) while others are replaced by the reseller's own commercial terms, which are almost always thinner on the three things that matter here: price protection past the Discount Period, true-down rights, and remedies when the vendor changes packaging mid-term.

On direct paper you are arguing with the party that owns the pricing catalogue.

On reseller paper you are arguing with a party that cannot rewrite Google's Post Discount Period language even if it wants to, because it inherits the reversion itself and would be selling you protection it does not hold.

Quantify the trade before you pick the route. On a $10M three-year GCP commitment, the reseller's margin is typically in the low single digits of the deal value, and a good chunk of that is rebate-funded rather than sacrificed at your expense.

A Post Discount Period continuation clause on the same commitment, holding your negotiated rate for a stated wind-down while renewal is finalized, is worth the full delta between your rate and list on whatever spend runs through the gap.

At a 28 to 35% effective discount, a single quarter of reversion on a $10M annual run rate costs roughly $1.0M to $1.3M. That is not a margin conversation. It is an order of magnitude.

Run the same nine-clause list against both paper sets and score which clauses survive, rather than assuming the channel is neutral. Our sub-article on reseller versus direct paper works through where flow-down holds and where it silently drops.

Where the partner brings genuine services value, take it, but insist the reseller's Google-facing agreement backstops your price hold in writing, and mirror the exit and assignment rights you would demand direct so a partner change does not reset your commercial terms.

Google will respond by pushing you toward whichever route it currently rewards internally. That preference is information, not instruction.

The anchor case worth keeping on the desk: a 10,000-seat Enterprise Plus estate with Gemini across all seats lists at $6.6M per year before a single conversation.

Reach the top of the 10,000+ band with a three-year term and you are near $4.4M, and the entire $2.2M gap sits on paper that reverts to list at the end of the Discount Period unless somebody drafts around it.

That is the whole argument for treating clause construction as the primary work rather than the cleanup after price.

The pattern that keeps repeating across 2024 to 2026 engagements is counterintuitive to most procurement teams. Buyers who tabled the structural clauses in week one, before naming a target price, closed at rates equal to or better than buyers who led with a discount ask.

The mechanism is simple: once you have anchored on percentage off list, the rep manages the conversation entirely inside that frame, and clause redlines arrive in the last fortnight as inconvenient obstacles to a deal both sides have already emotionally closed.

Table reversion, true-down, and drawdown definition first, and the discount conversation happens against a structure that holds. Google's counter-move is predictable: concede a point or two of rate to buy the reversion clause back.

Refuse that swap, because the terms agreed at your moment of least information are used at your moment of least leverage, and two points of rate is a rounding error against a full reversion to list.

What is Google's Post Discount Period clause and can it be removed?

It is the provision stating that if you keep using the services after the Discount Period ends, you automatically move to then-current list prices. Outright deletion is rare because Google, like AWS, has shown limited flexibility here.

The achievable position is a continuation clause: discounted rates persist month to month while both parties negotiate in good faith, and if no agreement is reached within a defined window, the discounts hold through a transition period regardless of monthly spend level.

Can you negotiate a true-down on a Google Cloud commitment?

Yes, more readily than with AWS or Azure. Google handles private-agreement shortfall on a negotiated basis, and existing agreements variously contain hard shortfall payment obligations or automatic commitment reduction rights depending on what the buyer asked for at signature.

The commercial team will not raise it proactively. A realistic ask is a year-three reduction right of 20 to 30% of the original commit, tied to defined triggers such as divestiture or workload migration.

What discount should we expect on a $10M three-year Google Cloud commitment?

A three-year $10M GCP commitment routinely closes at a 28 to 35% effective discount, which runs deeper than comparable AWS EDP or Azure MACC outcomes because Google is buying market share.

The EDP layer itself typically sits at 9 to 25%, with committed use discounts stacking underneath to add another 20 to 55% on covered usage. Published CUD rates (up to 55% on eligible Compute Engine hardware, up to 70% on memory-optimised series) are the floor, not the negotiated price.

How much did Google Workspace prices increase in 2025 and can we avoid it?

Base Workspace list rose 17 to 22% to absorb Gemini: Starter $6 to $8, Standard $12 to $18, Plus $22 to $28 per user per month, effective March 17, 2025 for existing customers or at next renewal, whichever came first. There is no opt-out and no way to buy Standard or above without Gemini.

The only protections are contractual: a fixed per-seat rate for the full term, a renewal uplift cap, and a repackaging clause that ties your rate to functional entitlement rather than SKU name.

What counts toward a Google Cloud EDP drawdown?

Marketplace spend on third-party software and data counts at 100% toward the commitment, which is one of Google's genuinely favourable positions.

CUD-covered consumption and each service family's usage should also draw down, but the specifics live in your schedule rather than in a standard definition.

Get the inclusion list enumerated in writing with a catch-all for future Google-billed SKUs, and confirm explicitly whether CUD-covered spend draws down at retail value or net of the discount, because the difference can be several points of commitment coverage.

What does Gemini Enterprise cost and what should we cap in the contract?

Gemini Enterprise runs $21 per seat per month for Business edition and $30 or more for Standard and Plus, with the full range spanning $21 to $60+ across four editions, plus separate token and compute billing for custom agents.

The Business price point matches Microsoft 365 Copilot Business at $21 exactly, which is a usable price-match lever.

Cap the metered layer in a consumption schedule (Agent Storage at $0.30 per GiB-month, compute at $0.085 per vCPU-hour, grounding at $45 per 1,000 web prompts versus $2.50 per 1,000 on your own data) and require 90 days' written notice before any new or repriced meter takes effect.

Do you lose contract protections by buying Google Cloud through a reseller?

Some, and they tend to be the ones that matter most. Reseller paper flows down parts of the Google agreement but substitutes the reseller's own terms elsewhere, commonly around price protection, true-down rights, and remedies.

Run your full redline list against both direct and reseller paper before choosing a route, because on a $10M commitment the value of a Post Discount Period continuation clause typically exceeds the reseller margin you are being offered.