Every vendor publishes a list price and almost no enterprise pays it. This benchmark reads what buyers actually negotiate off list across eleven major software vendors in 2026, in defensible bands rather than false averages, and names the levers that decide which band you land in.
Every major software vendor publishes a list price, and almost no enterprise pays it. This benchmark reads what buyers actually negotiate off list in 2026, in defensible bands rather than false averages, and names the levers that decide which band you land in.
About this report
This benchmark is a directional reference, not a price list. It reports defensible bands, never false precision, and draws on three inputs.
Where a single number appears, treat it as the middle of a range. Individual outcomes vary widely with estate size, metric, timing, and leverage.
The honest answer is a band, not a figure. Across the eleven vendor panel, realized discounts off published list ran from about 5 percent to more than 70 percent in 2026, set entirely by the vendor, the product, and the deal in front of you.
That spread is the whole point. A single blended average, say 25 or 30 percent, describes no real deal. It mixes a deep Oracle license discount with a shallow Broadcom transition quote and calls the midpoint a benchmark.
So the useful question is not what the average buyer gets. It is what a comparable buyer, with your scope and your leverage, lands on the same paper. That is the band this report tries to draw.
Four variables decide where your deal sits. Each one can move the realized rate by double digits, which is why a clean average is a fiction.
Read any benchmark, including this one, as a map of bands. Your job is to find the band your deal belongs in, then push to its top edge, not to hit a headline average you saw in a survey.
A defensible band is tied to a normalized unit and a stated scope. It says, for this vendor, this metric, this edition, and this term, comparable buyers land between X and Y. Anything looser is noise.
A band also carries a direction. Discounts on legacy perpetual licenses behave differently from discounts on new subscription seats, and the two are drifting apart as vendors migrate revenue.
The trend matters as much as the level. Discretionary discount room has compressed slowly since 2021 as vendors moved to subscription and held unit rates firmer. The discount you could win three years ago is not the discount on the table now.
Start with the unit you actually consume, not the unit on the quote. Strip the deal back to a clean rate per user, per core, or per consumption unit, then compare that rate to the band for your vendor and metric.
If your rate sits below the band midpoint, you are doing well. If it sits above, the gap is your negotiation target, and the levers later in this report are how you close it.
Because a discount percentage is only meaningful against a defined scope, and most published averages compare deals that share almost nothing. Normalize first, or the number lies to you.
Normalization adjusts deals to a common basis before you compare them. Four alignments do most of the work.
Miss any one of these and two deals that look identical on the headline can differ by twenty points on the rate that actually hits your budget.
Vendors know buyers chase the headline discount, so the easiest concession is a deep discount on a base that was inflated to begin with. A 50 percent discount on 1,000 licenses you will never deploy costs more than a 20 percent discount on the 600 you actually use.
Normalize to the unit you consume, not the unit on the quote. The table below shows two deals with an identical 40 percent headline and very different real costs.
Same 40 percent discount, two very different bills
| Line item | Padded deal | Right sized deal |
|---|---|---|
| Licenses on paper | 1,000 | 620 |
| List per license | $1,000 | $1,000 |
| Headline discount | 40% | 40% |
| Annual cost | $600,000 | $372,000 |
| Cost per license you actually use | $968 | $600 |
The right sized deal carries the same percentage off and costs 38 percent less in cash. The discount did not change. The scope did. That is the lever the headline hides.
The effective rate divides what you pay by what you use. It cuts through every packaging trick a vendor can build into a quote, from free units that expire to bundled modules you did not ask for.
Ask for the effective rate on every option the vendor offers. The cheapest headline discount and the lowest effective rate are often two different lines on the same quote.
Oracle and IBM post the deepest headline discounts off list, often 40 to 70 percent on licenses, but that depth sits on the most aggressively set list prices in the market. Broadcom VMware and AWS sit at the shallow end.
Depth is a function of how the list was built and how much the vendor must defend the account. Four forces explain most of it.
Broadcom reset VMware pricing after the acquisition and removed most discretionary discounting. AWS commit discounts are tiered and largely formulaic. ServiceNow and Workday hold firm on subscription unit rates and concede on term and ramp instead.
Even the firm vendors move on something. Salesforce raised list prices and concedes through bundling and consumption credits, while Microsoft 365 enterprise list pricing anchors a discount that arrives through enrollment level and term, not a simple percentage.
Oracle offers some of the deepest paper discounts in the market, routinely 40 to 70 percent on database and middleware licenses. The list is set high enough that the rate after a deep cut still protects Oracle revenue.
The trap is the support stream, not the license. A deep one time discount on the license still locks a 22 percent annual support fee on the discounted figure, so the real lever is the support base, not the headline.
Microsoft rarely moves on a simple percentage. The discount arrives through the enrollment level, the term, and the mix of products you commit to across the estate.
The realized band sits around 15 to 30 percent off list on a large enterprise agreement, with the deepest rates reserved for buyers who commit broadly and start the renewal early.
SAP discounting in 2026 runs through the move to its subscription and cloud bundles. The deepest concessions appear when SAP is trying to convert a perpetual estate, and they shrink once you are inside the subscription.
Realized discounts of 30 to 55 percent are common on a conversion deal, but the renewal that follows is where the rate quietly resets if the cap was not negotiated up front.
Salesforce holds firm on the per user list and concedes through bundles, free product, and consumption credits. The headline discount looks modest, often 20 to 40 percent, but the bundled value can be larger than the percentage suggests.
Read the bundle on the effective rate. Free units that expire after year one are a discount that disappears exactly when the renewal uplift arrives.
IBM enterprise license agreements open with generous discounts and broad use rights, which is what makes them attractive and dangerous. The giveaway is real, and so is the true up at the end of the term.
Realized discounts of 30 to 60 percent are common, but the value of the deal is decided by the true up and renewal terms, not the opening rate. Negotiate the exit before you sign the entry.
These vendors hold the subscription unit rate firmly and concede on term length, ramp, and added modules. The realized discount band sits around 10 to 25 percent, with most of the value coming from ramp and timing rather than rate.
The buyer side move is to stop pushing the rate after a point and trade for a renewal cap, a price hold, and the flexibility to reallocate seats.
Hyperscaler discounts are tied to commitment, not negotiation in the traditional sense. AWS private pricing and commit programs run on tiers, and the discount follows the size and length of the commit.
The realized band is shallow, often 5 to 20 percent, but the larger lever is the commit level itself. Overcommit and the discount is irrelevant against the unused spend you still owe.
Rank vendors by realized rate on a normalized unit, never by headline discount. The vendor with the biggest percentage off is often the one with the most room built into list to give away.
Bigger deals do earn deeper discounts, but the effect is smaller and less reliable than buyers assume, and it reverses once the extra volume becomes shelfware you will not use.
The first dollars of scale buy the most rate. After a point, more volume buys terms, not a better unit price.
Notice the gold bars in the chart. In every tier, a prepared buyer realizes more than the vendor opens with, and the size of that gap is decided by preparation, not by the tier.
The volume discount is a trap when it tips you into licenses you will not deploy. Vendors offer a better unit rate at a higher tier precisely because it grows the contract. Pay for what you will use, then negotiate the rate on that.
We see this most on seat based subscriptions and on Oracle and IBM enterprise agreements, where a tier upgrade looks cheap per unit but raises the total bill.
Buyers overbuy because they fear a true up penalty if they grow. The fix is a swap or expansion right at the negotiated rate, not a bigger commit on day one.
A right to add at the same rate is worth more than a deeper discount on volume you may never use. Negotiate the right to grow, then buy for today.
Four beliefs cost buyers more than any vendor tactic. Each one feels intuitive and each one is wrong in the deals we benchmark.
A bigger percentage off an inflated, padded base loses to a smaller percentage off a right sized one. The percentage is a vanity metric. The cash you pay for what you use is the only score that counts.
The first quote is an opening position, set with room to move. In our engagement file, first quote discounts ran 10 to 20 points below what a prepared buyer in the same band eventually realized.
Mid market buyers with a credible alternative often beat larger buyers who negotiated late and alone. Leverage is not the same as size. A small, well prepared buyer outperforms a large, unprepared one.
A multi year commit deepens the rate by roughly 3 to 10 points, but it locks scope you may outgrow or abandon. On a volatile estate, the flexibility you give up costs more than the points you win.
Three things separate a top quartile deal from a median one, and none of them is negotiating harder in the final call. They are timing, a credible alternative, and the clauses.
The buyer who starts 9 to 12 months out controls the calendar. The buyer who starts six weeks out negotiates against their own renewal date. Vendors read both positions instantly and price accordingly.
Time is the cheapest leverage there is. It costs nothing to start early, and it is the single factor most often missing from a median deal.
An alternative does not have to be a migration you intend to make. It has to be a documented, costed option the vendor believes you could take. That belief is what moves the rate, often by 8 to 15 points.
Build the alternative early and quietly. A competitor proof of concept, a costed exit plan, or a consolidation option all qualify, as long as the vendor sees it as real.
A discount you cannot hold is not a discount. Four clauses protect the number after the ink dries.
Win the rate and lose the cap and you have won nothing. The renewal uplift will erase a hard fought discount inside two cycles if it is left open.
What moves the discount band, by lever
| Lever | Typical effect on the realized band | Where it works best |
|---|---|---|
| A credible documented alternative | Adds 8 to 15 points | Any vendor facing a real competitor or exit |
| Calendar started 9 to 12 months out | Adds 5 to 12 points | Renewals, true ups, transitions |
| Scope right sized first | Cuts the bill more than any rate | Every deal, seat based most of all |
| Multi year commit | Adds 3 to 10 points, costs flexibility | Stable, predictable estates only |
| Quarter end and year end timing | Adds 2 to 8 points | Vendors with public quarterly targets |
No single lever wins a top quartile deal. The pattern we see is buyers who stack three or four of them, early, against a vendor who knows the alternative is real.
The band shifts with who is buying and how. Sector, region, and the purchasing vehicle each move the realized rate, sometimes more than deal size does.
Public sector and education buyers often access dedicated pricing well below commercial list, but with rigid terms. Financial services and healthcare buyers pay closer to commercial rates and win instead on compliance and audit protections.
High growth technology buyers tend to get aggressive new logo discounts that snap back hard at the first renewal. Read the renewal, not the opening rate, in any fast growing account.
Discount bands vary across regions with competition and currency. North American deals often carry the deepest discretionary discounts, while some regions trade rate for local terms, support, and data residency commitments.
For a global estate, benchmark each region against its own band. A single global discount target hides where you are overpaying and where you already lead.
The vehicle changes the math. Direct deals, reseller deals, and marketplace purchases each carry different economics and different leverage.
Pick the vehicle that maximizes leverage for the specific deal, not the one that is easiest to process through procurement.
Treat every vendor discount claim as an opening position measured against a list price the vendor controls. The percentage is only as honest as the base it is taken from.
If the answer to any of these is vague, the discount is not yet real. Pin each one down before you treat the percentage as a number you can compare.
Industry surveys report blended averages that mix incomparable deals. Use them for direction, never as a target. Gartner data on rising software spend is useful context, not a number to quote back to a vendor.
A vendor will happily accept a survey average as your target if it sits above what you could actually win. Bring a normalized band, not a headline you read in a report.
Watch for free units that expire, modules bundled in that you did not request, and a deep one time concession with no renewal cap. Each one signals a discount built to shrink.
The cleanest discounts are simple. A lower unit rate, a capped renewal, and the right to add at the same price beat any stack of expiring credits.
A benchmark is only useful if it changes the number you sign. The path from band to outcome runs through three steps, and most buyers skip the first two.
Reconcile entitlements to actual deployment and use. Strip out shelfware and double counted licenses, then express what remains as a clean unit you can benchmark.
Most estates carry 10 to 30 percent more licenses than they deploy. The baseline is where that gap surfaces, before a vendor uses it against you.
Place your normalized rate against the band for your vendor, metric, and term. Set a target at the top edge of what comparable buyers achieve, not at the survey average.
A target band gives your team a defensible position and a stopping point. It tells you when to keep pushing and, just as important, when you have won.
Open with the band, the alternative, and the calendar already in place. Trade rate for the protective clauses once the rate reaches the top of the band, and stop chasing points that cost you flexibility.
Document the final effective rate and the clauses, then carry them into the next renewal as your new baseline. A benchmark is a cycle, not a one time event.
Almost nothing, in marginal terms, which is why the floor is strategic rather than cost based. Understanding the vendor economics changes how hard and where you push.
The cost to deliver one more software license is close to zero. The price floor a vendor will accept is set by revenue targets, account strategy, and competition, not by what the product costs to make.
That means the discount is a negotiation about value capture, not cost recovery. The vendor can always go lower. Whether it chooses to depends on what it stands to lose if it does not.
Sales teams carry quarterly and annual targets, and a deal that slips a quarter can miss a quota. That calendar pressure is real leverage for a buyer who controls timing.
The same ask made in the vendor first quarter and at its year end can land points apart. Know the vendor fiscal calendar and place your decision where it helps you most.
Your leverage is the revenue the vendor fears losing, multiplied by the calendar pressure it is under. Build both, and the floor moves toward you without a single extra dollar of cost to the vendor.
AI add ons are the biggest change to the discount math in years. They are priced as premiums on top of the base, and they are discounted far less than the seats they sit on.
Generative add ons arrive at a firm premium with little discretionary discount in the early cycles. Vendors protect the new revenue line, so the discount you won on the base rarely extends to the AI layer.
Treat the AI add on as a separate negotiation with its own band. Bundling it into the base discount is how buyers end up paying full premium on a layer they could have piloted first.
Where AI is priced by consumption, the headline rate tells you almost nothing. The real cost depends on usage you cannot fully predict before deployment.
Model a high and low usage case before you commit. A discount on an unmetered consumption line is a discount on a number you do not yet control.
Pilot before you commit, cap the consumption exposure, and negotiate a separate exit on the AI layer. Keep the base license discount clean and do not trade it away to fund the premium.
A discount is not won at signature. It is held or lost across the term. Three forces quietly erase a hard fought rate if the contract does not block them.
The single biggest eroder is an open renewal uplift. A discount won on day one disappears inside two cycles if the renewal can rise at the vendor discretion.
A fixed cap is worth more than a few extra points off the opening rate. Win the cap, even at the cost of a slightly shallower headline discount.
Free units and ramp credits that expire after year one are a discount designed to vanish. They flatter the first year effective rate and leave you exposed at renewal.
Strip expiring value out of your effective rate calculation. Benchmark the steady state cost, not the year one promotional one.
Unmanaged growth in users or usage triggers true ups at rates you did not negotiate. Without a swap or add right at the locked rate, expansion is billed at a worse number than the original deal.
Govern deployment against entitlement through the term. A quarterly reconciliation catches creep before it becomes a true up bill.
The two run on different leverage and belong against different bands. Benchmarking them the same way is a common and expensive mistake.
At renewal, the vendor knows your usage, your switching cost, and your deadline. Your leverage is lower, so the band is tighter and the work moves to the clauses and the calendar.
The renewal benchmark asks whether your rate held against inflation and uplift, not whether you beat a new logo discount you will never see again.
A new purchase carries the deepest discounts because the vendor is competing for the logo against real alternatives. The new logo band is wide and the rate is at its most generous.
The trap is reading that new logo rate as your forever rate. It is the opening of a relationship that the renewal will reprice unless you locked it.
Never benchmark a renewal against new logo discounts. Compare a renewal to comparable renewals and a new purchase to comparable new deals. Mixing the two sets a target you cannot reach or one too easy to matter.
A short, anonymized composite from the engagement file shows how the band and the levers combine. The numbers are illustrative ranges, not a single account.
A buyer faces a renewal with a vendor opening uplift in the mid teens on a base padded with roughly 20 percent unused licenses. Left alone, the deal renews above the band.
The buyer right sizes first, then stacks the levers across a calendar started early.
The realized result lands near the top of the band rather than above it. The headline discount is not the deepest on paper, but the effective rate and the capped renewal beat a deeper discount on the padded base.
That is the pattern the whole report points to. The win was the scope and the clauses, not the percentage.
A credible benchmark rests on three data sets, and most buyers hold only the first. Without all three, the band is a guess dressed as a number.
Start with your current paper and your actual deployment. The gap between what you own and what you use is the first and largest source of savings, and it is data only you hold.
Pull the entitlements, the metric definitions, the renewal terms, and the real usage telemetry. A benchmark built on a clean internal baseline beats one built on a vendor summary every time.
The band itself comes from comparable deals, normalized for vendor, metric, edition, and term. This is the data buyers rarely have alone, which is why an independent benchmark adds the most value here.
One peer anecdote is not a band. A defensible band needs a set of comparable deals wide enough to show a range, not a single data point someone shared at a conference.
Public list prices and dated price actions anchor the top of the range. They tell you what the vendor asks before any concession, which is the reference point every discount is measured against.
Combine the three. Your baseline sets the scope, the comparable deals set the band, and the public list sets the anchor. Miss one and the benchmark loses its footing.
Most enterprises do not buy from one vendor. They run a portfolio, and the portfolio view changes both the leverage and the priorities.
Map every major contract by annual spend and renewal date. The largest spend with the nearest renewal is where benchmarking effort returns the most, and it sets the calendar for the year.
A portfolio calendar stops renewals from arriving as surprises. It turns a reactive scramble into a planned sequence where each negotiation starts early enough to matter.
Bands travel across the portfolio. What you learn about realized rates and clause language on one vendor strengthens your position on the next, even when the products are unrelated.
Consolidation is also a lever. Where two vendors overlap, the credible option to shift spend from one to the other deepens the discount on both.
Treat benchmarking as an ongoing program, not a series of one off events. A standing view of bands, renewals, and clauses across the portfolio compounds in value each cycle.
That is the logic behind a managed renewal and benchmark program. The portfolio view catches the next overpriced renewal before it lands, not after you have signed it.
The standard advice is to push for the biggest possible discount percentage and treat that number as the scorecard. We disagree, and the engagement file is clear on why. In the deals we benchmark, a deep discount on an inflated, badly scoped base routinely costs more in absolute dollars than a smaller discount on a right sized one. A 40 percent cut on twice the licenses you need is a loss dressed as a win. The buyer side move is to right size the estate first, normalize the scope to what you actually consume, benchmark the unit rate against comparable deals, and only then negotiate the percentage. Chase the bill, not the headline.
Source: Redress Compliance advisory engagement file, 2024 to 2025.
The vendor sets the list. Your alternative sets the floor. The buyer who right sizes first and benchmarks the unit rate pays the floor, not the headline discount.
Expect a band, not a number. Across the panel realized discounts ran from about 5 percent to more than 70 percent in 2026, set mostly by the vendor, the product, and your preparation. Find the band your deal belongs in, then push to its top edge.
Because averages are not normalized for scope, term, or deal size. A blended figure mixes a deep Oracle license discount with a shallow Broadcom quote and reports the midpoint. Compare like for like or the number misleads you.
Oracle and IBM post the deepest headline discounts, often 40 to 70 percent on licenses. That depth sits on the most aggressively set list prices, so a big percentage off proves little on its own. Broadcom VMware and AWS sit at the shallow end.
No. Larger deals earn deeper discounts up to a point, then the curve flattens and extra volume buys terms more than rate. Buying licenses you will not deploy raises the total bill even at a better unit rate.
Benchmark the unit rate on a normalized scope against comparable deals, not the headline percentage. A competitive discount is one that beats the band for your vendor, metric, edition, and term. The percentage alone tells you almost nothing.
Scope normalization adjusts deals to a common basis before comparing them. It aligns the metric, edition, term, ramp, and support terms so the discount reflects the same thing. Without it, two deals with the same percentage off can carry very different real costs.
Often yes, by roughly 3 to 10 points, in exchange for flexibility you give up. A multi year commit deepens the rate but locks scope you may outgrow or stop using. Take it only on a stable, predictable estate.
Before the first vendor response, not after the deal stalls. An independent benchmark sets your target band and exposes a padded base early, while you still have time and leverage. Brought in late, it only confirms what you already signed.
No. A deep discount on an inflated, badly scoped base costs more than a smaller discount on a right sized one. Right size first, normalize the scope, benchmark the unit rate, then negotiate the percentage. Chase the bill, not the headline.
The vendor by vendor discount bands, the deal size tiers, the scope normalization worksheet, and the lever scorecard that decides which band you land in.
Used across more than five hundred enterprise engagements. Independent. Buyer side. Built for procurement and finance leaders running the next negotiation.
A deep discount on the wrong scope is the most expensive line in the contract. Right size first, then negotiate the rate.
