The vanishing discount. Why it is not coming back.

For two decades enterprise software pricing followed a stable ritual: high list prices, deep negotiated discounts, and a final number that rewarded customers who pushed. That ritual is ending, and the data across 800 contracts shows exactly how fast.

What does the 800 contract benchmark show?

Average achieved discounts have fallen at Oracle, SAP, Microsoft and Salesforce every year since 2019. The benchmark tracks the discount achieved against list price at signature, by vendor and year, across new agreements and renewals of comparable size. Four different shapes, one direction: down.

The numbers come from 200 contracts per vendor, drawn from buyer side engagements and benchmark contributions. Deals are normalized to comparable annual contract value bands and measured as achieved net price versus the applicable list at signature.

How should you read the four vendor curves?

SAP falls fastest because the product carrying the discount changed underneath the customer. Perpetual S/4HANA discounting has been all but withdrawn to make RISE subscriptions the only economically rational path.

Oracle erodes by policy, roughly two points a year from 57 to 44 percent, without any single announcement to point to. From June 2026 the list prices those discounts apply to have risen sharply as well.

Microsoft shows a step change. Discounting was stable through 2024, then the removal of programmatic Enterprise Agreement price level discounts from 2025 produced a roughly thirteen point fall in two years.

Salesforce looks gentlest, but its line understates the truth. Two rounds of list price increases mean a stable discount percentage is applied to a higher base, so net cost rises anyway.

What does the methodology cover, and what does it not?

Averages smooth wide individual variation. Well leveraged deals still beat these lines, which is the point of the buyer response section below. The benchmark excludes deals where scope changed so much between cycles that net price comparison would mislead.

• Panel: 800 contracts, 200 per vendor, new agreements and renewals, normalized to comparable annual contract value bands.
• Measure: achieved net price versus the applicable list price at signature, expressed as a percentage discount.
• Period: contracts signed 2019 through the first half of 2026.
• What it is not: a promise of what any single deal can achieve. Bands, leverage and timing move individual outcomes by double digits.

Why are enterprise software discounts disappearing?

The discount is disappearing because the subscription model changed what a discount costs the vendor, sales compensation was redesigned to defend list price, and Broadcom proved the market rewards repricing a locked in base. The obvious driver is the cloud transition, but the mechanism matters, because it explains why the decline will continue.

How did subscription economics change the cost of a discount?

In the perpetual era, a discount was a one time concession on a one time fee, and maintenance revenue arrived regardless. In the subscription era, a discount given once is given every year of the term, and usually every term after that.

Vendor finance teams have understood this. Discount discipline has become a board level revenue quality metric, because recurring revenue at protected prices is what the market values these vendors on. The discount you are asking for is no longer a sales concession. It is a permanent reduction in the asset their share price is built on.

How do sales compensation plans now work against your deal?

The second driver sits in sales compensation design, and it is the one buyers most underestimate. Across the four vendors, the patterns are consistent.

What did the Broadcom VMware playbook demonstrate to the market?

The third driver is demonstration. Broadcom's VMware repricing showed every software board that a locked in installed base can absorb dramatic price increases, that some customer loss is an acceptable cost, and that capital markets reward the result.

The four vendors in this report all watched that experiment succeed. Switching costs on ERP, CRM and productivity estates are even higher than on virtualization, and the lesson has not been missed.

How is each vendor executing the subscription pivot?

Each of the five vendors below runs the same strategy through a different mechanism: relocate or remove the discount, and reprice the installed base on the model the vendor prefers. The order tells a story, from the bluntest execution to the subtlest.

SAP: where did the perpetual discount go?

SAP has all but eliminated discounting on perpetual S/4HANA licenses, after two decades in which large ECC deals routinely closed 60 to 75 percent below list. The intent is explicit: make the perpetual route economically irrational so RISE becomes the only path.

RISE subscriptions arrive with conversion credits and year one discounts that look generous beside the inflated perpetual quote. The discount has not disappeared. It has been relocated to the model SAP wants you on, and it expires at the first renewal. Our guide to negotiating S/4HANA conversions and migration credits covers the counter moves.

Broadcom and VMware: what does the take it or leave it template look like?

Broadcom ended VMware perpetual licensing outright, collapsed a large standalone product catalogue into a handful of subscription bundles, attached minimum commitments, and presented renewal pricing on effectively take it or leave it terms, often at multiples of historic cost.

Customers pay for bundled components and packaged services they never deployed. It is the purest version of the strategy, and the template the rest of the market is studying.

Oracle: is the legacy estate next after Java?

Oracle moved Java SE to an employee count subscription metric in January 2023, repricing entire workforces regardless of actual usage. That model change was executed on the installed base rather than negotiated with it. On 1 June 2026 Oracle followed with broad list price increases across the portfolio, reaching up to 40 percent on some products.

Yet Oracle has so far left its large legacy perpetual licensing business, database and middleware, untouched. We do not expect that to last.

Our assessment is that Oracle will move the legacy perpetual business toward a subscription model within the next 12 to 24 months, with Java as the proven mechanism: change the metric, change the model, and reprice the base.

Customers should negotiate their next Oracle renewals as if that shift were already announced. The June 2026 Oracle price list changes are the immediate, documented half of the squeeze.

Microsoft: what ended the programmatic discount era?

Microsoft removed the built in Enterprise Agreement price level discounts for online services from 2025, aligning EA pricing with web direct. Smaller enterprises are being pushed off the EA entirely toward CSP buying, where negotiated discounting barely exists. The mechanics are documented on Microsoft's licensing news pages and in our EA discount removal impact analysis.

Combined with the price actions documented below, the result in our benchmark is a roughly thirteen point fall in average achieved discount in two years, after half a decade of stability.

Salesforce: how does the list price do the work?

Salesforce has paired gradual discount tightening with repeated list increases, the first in 2023 after seven years of stable pricing and another round in 2025, while attaching its AI portfolio to premium editions and consumption pricing.

When the list rises and the discount percentage merely holds, the net price increases without a single concession being withdrawn. It is discount reduction by other means. Our Salesforce discount benchmark shows where the bands sit today.

Which vendor price actions are documented since 2022?

Discount compression is only half of the squeeze. The list prices the discounts apply to have been rising in parallel, in a sequence of documented vendor actions.

These are published vendor price actions as documented at the time of writing. Individual customer impact varies by agreement, region and currency. The combination is what matters: a stable discount applied to a list price that has risen 20 percent is a 20 percent increase.

The compounding effect is brutal. A customer holding its 2019 discount percentage through every one of these actions has still taken a double digit net price increase. A customer who also lost discount points, which the benchmark shows is the average experience, has taken both at once. This is why renewal budgets built on last cycle's numbers keep failing.

How do you tell a real concession from a relocated discount?

A real concession reduces what you pay over the life of the relationship. A relocated discount moves the same money to a place where the vendor recovers it later, usually at the first renewal. Telling them apart is now the core skill in enterprise software negotiation.

The benchmark data makes the distinction visible. Deals that looked well discounted at signature, then renewed 25 to 60 percent higher two or three years later, almost always carried a relocated discount rather than a real one.

What does a relocated discount look like in practice?

The pattern repeats across all four vendors, with different labels on the same mechanism. The concession is loaded into year one of the model the vendor prefers, and the recovery is loaded into every year after it.

• SAP conversion credits: generous looking credits against the RISE subscription, priced beside a perpetual quote that was inflated by withdrawing its discount. The credit expires; the subscription does not.
• Microsoft transition bridges: funded ramps and one time concessions that ease the move into the new commerce motion, while the programmatic discount that used to recur is gone for good.
• Oracle migration incentives: support credits and cloud commitments offered against moves off perpetual estates, ahead of the broader repricing we expect on the legacy base.
• Salesforce growth pricing: rate protection offered only on net new volume, while the installed base absorbs the list increases and the AI SKU attach.

Which concessions still carry real, durable value?

The concessions worth trading for share one property: they are written into the contract and survive the renewal. Anything that lives in the order form for one term is a teaser, not a term.

• Indexation caps: a hard ceiling on annual uplift, applied to the whole estate, not just a SKU subset.
• Renewal price locks: the right to renew at the same net rate, for a defined volume corridor, for at least one further term.
• Price holds on growth: committed unit rates for additional volume, so expansion does not reopen the whole price.
• Downsize rights: the contractual ability to shed a defined percentage of volume at renewal without repricing the remainder.
• Exit and portability terms: data egress, transition assistance and short tail obligations that keep your alternative believable next cycle.

What three questions expose a teaser at the table?

Ask these in writing, before signature, and the relocated discount declares itself. Vendors structure teasers precisely so these questions are never asked directly.

1. What does this exact scope cost in year one of the next term, at today's list, if we change nothing?
2. Which of these concessions are written as contract terms that survive renewal, and which exist only in this order form?
3. If we decline the model change entirely, what is the price of standing still, and for how long is that price available?

The answers convert a discount conversation into a total cost conversation. That is the conversation the deal desk is not scripted for, and it is where prepared buyers recover the ground the headline percentage no longer gives them.

What can buyers do about falling discounts?

Six moves still work, and none of them is asking harder for a bigger percentage. The negotiation value has moved from the discount line to protection, alternatives and timing, and the buyers who internalize that first keep beating the benchmark averages.

Which six responses still move the number?

1. Benchmark against today's market, not your last deal. The 2019 discount is gone as a reference point. Negotiate from current, deal size matched benchmark data, and set internal expectations from it, so a strong outcome is recognized and a weak one is not signed out of anchoring. Our cross vendor discount benchmark shows the current bands.
2. Move the negotiation from discount to terms. In a subscription world the protections outlast the percentage: indexation caps, renewal price locks, price holds on growth, downsize rights and exit terms. Every term not written at signature is a future price increase the comp plan will defend.
3. Build alternatives early enough to be believed. Centralized deal desks concede to risk, not to requests. Third party support, competing platforms, workload portability and partial exits move pricing only when developed twelve months out and visibly real.
4. Exploit the compensation plan you are negotiating against. Quarter ends, year ends, cloud multipliers and quota gaps are your leverage. Time closes to the vendor's fiscal calendar, and trade the commitments their plans reward for the protections their plans resist.
5. Treat model changes as negotiation events, not announcements. SAP's RISE pivot, Broadcom's bundles, Microsoft's EA changes and the Oracle legacy shift we predict are moments when vendors need signatures to validate the model. That need is leverage, and it is largest for customers who move deliberately rather than first.
6. Right size before every renewal. When discounts cannot be won, volume can be cut. Usage based right sizing of users, editions, modules and shelfware before the renewal is now the most reliable source of double digit savings at all four vendors.

Where the common advice on discount negotiation is wrong

The standard procurement playbook says to escalate harder for a bigger percentage: push the rep, threaten the renewal, demand the discount your peers supposedly got. We disagree. Across the 800 contracts in this benchmark, and in the renewals Fredrik Filipsson advised through 2024 and 2025, percentage pressure alone moved deal desk floors by low single digits at best, because the floors are now set above the account team and defended by compensation design. The deals that beat the band by 8 to 15 points did something else: they built a credible alternative a year early, timed signature to the vendor's quarter, and spent their leverage on indexation caps and renewal locks rather than headline rate. The buyer side move is to stop negotiating the percentage and start negotiating the structure the percentage lives in.

800

Contracts in the benchmark panel

44%

Oracle average discount today, down from 57

8 to 15

Points gained by buyers with a real alternative

Source: Redress Compliance advisory engagement file, 2019 to 2026.

The discount you achieved in 2019 is not coming back. The question that decides your next renewal is what you trade for instead: caps, locks, rights and timing.

What does this mean for your next renewal budget?

Budget for net price, not for percentage. A renewal budget that assumes the last cycle's discount carries forward is the single most common planning failure we see, because it ignores both the discount erosion and the list increases underneath it.

Three planning rules follow from the data. First, rebase the budget on current list prices and current bands, not historic ones. Second, model the vendor's preferred motion, because the discount will be relocated there, as SAP did with RISE. Third, fund the alternative early, because the credibility that moves deal desks takes a year to build.

• Rebase: pull current list prices and apply the band for your deal size, not your historic percentage.
• Model the pivot: price the subscription motion the vendor will push, including the year one teaser discount and the first renewal uplift after it expires.
• Fund leverage: budget for the migration assessment, third party support evaluation or competing platform pilot that makes your alternative believable.
• Protect the downside: value indexation caps and renewal locks in money, so the negotiation team can trade them against rate with a clear conscience.

How should you present this to the CFO?

Present the renewal as a net cost forecast with a confidence range, not a discount target. A discount target invites the question of why the percentage fell. A net cost range, built from current list, current bands and the documented price actions, invites a decision about leverage instead.

Three numbers carry the conversation. The do nothing number: current scope at the vendor's new list and your realistic band. The protected number: the same scope with caps, locks and right sizing applied. The leverage cost: what building the credible alternative will take in time and money.

Framed this way, funding the alternative stops looking like an IT project and starts looking like what it is: the cheapest insurance on a multi million dollar recurring cost line. In our engagement file, the spread between the do nothing number and the protected number routinely exceeded ten times the cost of the leverage work.

What to do next

Work the sequence below before your next major renewal. Each step compounds the one before it, and the calendar is the part you cannot recover later.

1. Inventory every agreement expiring in the next 24 months across Oracle, SAP, Microsoft, Salesforce and VMware, with current net prices and metrics.
2. Rebase each renewal budget on current list prices and current discount bands, not the percentage you achieved last cycle.
3. Right size the estate now: users, editions, modules and shelfware, so the volume baseline is defensible before pricing talk starts.
4. Pick the one or two estates where an alternative is real, and fund the work that makes it visible twelve months before signature.
5. Draft the term sheet you want at signature: indexation caps, renewal price locks, price holds on growth, downsize rights and exit terms.
6. Map the vendor's fiscal calendar and compensation incentives, and time your close to their quarter, not yours.
7. Benchmark the final offer against deal size matched data before you sign, and escalate above the deal desk if the band says you should.

Frequently asked questions

Why are enterprise software discounts disappearing?

Because the subscription model made every discount a permanent recurring revenue loss, sales compensation now defends list price, and Broadcom proved a locked in base can be repriced. All three drivers are structural, so the decline continues rather than reversing with the cycle.

How much have Oracle discounts fallen since 2019?

Oracle average achieved discounts fell from 57 percent in 2019 to 44 percent in 2026 across our 200 contract Oracle panel. On top of that, June 2026 list price increases of up to 40 percent raised the base the remaining discount applies to.

What happened to Microsoft Enterprise Agreement discounts?

Microsoft removed the programmatic Enterprise Agreement price level discounts for online services from 2025. Our benchmark shows a roughly thirteen point fall in average achieved discount in two years, and smaller enterprises are being directed from the EA toward CSP, where negotiated discounting barely exists.

Does SAP still discount perpetual S/4HANA licenses?

Effectively no. SAP now concedes close to zero discount on perpetual S/4HANA, after decades when large ECC deals closed 60 to 75 percent below list. The discount moved into RISE subscriptions as conversion credits and year one pricing, and it expires at the first renewal.

How does Salesforce raise prices without cutting discounts?

Salesforce raises the list price and holds the percentage. Broad increases of around 9 percent in 2023 and 6 percent in 2025, plus premium priced AI SKUs, mean a stable discount applied to a higher base raises net cost without withdrawing a single concession.

Will Oracle move its legacy database licensing to subscription?

We expect Oracle to move its legacy perpetual business toward subscription within the next 12 to 24 months. Java is the proven mechanism: change the metric, change the model, reprice the base. Negotiate your next Oracle renewal as if that shift were already announced.

What is the best negotiation lever now that discounts are falling?

A credible alternative built twelve months before signature is the strongest lever. In our benchmark it was worth 8 to 15 points against otherwise identical deals. After that, trade for indexation caps, renewal locks and downsize rights, which outlast any percentage.

When should we start preparing a renewal under this pricing regime?

Twelve months before signature at minimum. Alternatives, right sizing and benchmark baselines all need that runway to be credible, and deal desks concede to visible risk, not to late requests. Starting at the quote stage forfeits most of the available leverage.

Is the discount decline cyclical or permanent?

Permanent, on the evidence. The drivers are structural: subscription economics price every discount as a recurring revenue loss, compensation plans defend list, and the Broadcom demonstration was rewarded by the market. None of these reverses with the economic cycle, so plan on the new baseline.