The real discount is a band, and preparation decides which one
Every major software vendor publishes a list price and almost no enterprise pays it: realized discounts off list ran from about 5 percent to more than 70 across the eleven vendor panel in 2026, set entirely by the vendor, the product, and the deal in front of you. A single blended average describes no real deal, which is why the useful question is never what the average buyer gets, but what a comparable buyer with your scope and your leverage lands on the same paper.
Prepared by Redress Compliance · August 8, 2026 · Cross vendor advisory. Based on 160 to 200 enterprise negotiations and benchmarks supported 2024 to 2025.
Executive summary
Four variables set the band, and each moves double digits.
Scope, since a right sized estate and a padded one get very different effective rates on the same paper discount; term, with one year deals and five year commits priced on different curves; leverage, because a buyer with a live alternative negotiates against a different vendor than one without.
And timing, the same ask landing differently at the vendor's year end. A defensible band ties to a normalized unit and a stated scope, this vendor, this metric, this edition, this term, and anything looser is noise, including the published averages that compare deals sharing almost nothing.
The headline depth proves nothing, and the compression is real.
Oracle and IBM show the deepest headline discounts sitting on the most inflated list prices, so depth alone is meaningless without the normalized unit rate behind it.
And the room is shrinking: the median realized discount across the panel compressed from 34 percent in 2021 to 27 in 2026 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, which is why scope matters more each year the room narrows.
The prepared buyer's numbers repeat across every vendor. First quote discounts clustered 10 to 20 points below what a prepared buyer in the same band eventually realized, deals with a credible documented alternative landed 8 to 15 points deeper than otherwise identical deals without one.
And buyers who started the calendar nine to twelve months out beat late starters by 5 to 12 points on the same scope.
Across the panel, prepared buyers realized roughly 40 to 60 percent of the vendor opening ask, and a top quartile deal is separated from a median one by timing, a real alternative, and the clauses, not by negotiating harder on the day, a two times spread.
Right size before you chase the rate, because a deep discount on padding is not a win.
A deep discount on twice the licenses you need loses to a modest discount on the right count, which is why the estate right sizes before the quote and the comparison strips the deal to a clean rate per user, per core.
Or per consumption unit against the band for your vendor and metric: below the band midpoint you are doing well, above it the gap is your negotiation target, and the levers that close it are the calendar, the alternative, and the clause set rather than the table.
Scope normalization, or the number lies
| Alignment | What it corrects | The error it prevents |
|---|---|---|
| The same metric | Per user to per user, per core to per core | Comparing incomparable units and calling it a benchmark |
| The same term | One year against one year, commit against commit | Term curve differences read as negotiation skill |
| The same edition and scope | The tier and module set actually compared | A padded estate's paper discount flattering the rate |
| List against realized | The sticker against the signature | The published average that mixes both and means neither |
The deepest headline discounts sit on the most inflated lists.
Oracle and IBM head the depth table precisely because their list prices carry the most air, so a 70 percent discount there can price worse per unit than 20 percent elsewhere, and the only comparison that survives is the normalized unit rate, the clean per user, per core.
Or per consumption figure your deal strips down to.
Find the band your deal belongs in, then push to its top edge, and ignore every headline average a survey produced.
What moves the band, measured
- The early calendar, 5 to 12 points: nine to twelve months out against late starts on the same scope, the room every other lever needs.
- The credible alternative, 8 to 15 points: documented and costed, on otherwise identical deals, because captivity is what the account team prices.
- The preparation gap, 10 to 20 points: the distance between first quotes and what prepared buyers in the same band realized.
- Deal size, real but overrated: it moves the band, and the alternative and the calendar move it more.
- The clause set, compounding: the cap, the co terminus date, and the swap right decide what the discount is still worth in year three.
The renewal negotiation timing playbook
The band placement method, the normalization discipline, and the calendar that separates top quartile deals from median ones.
Get the white paper →Reading discount claims, and placing your own deal
The method for any claim, including this report's: ask what unit it is normalized to, what scope it assumes, and whether it reports list moves or realized signatures, because most published trackers report only the public list number while the buyer only ever pays the realized one.
Placing your own deal starts from the unit you actually consume rather than the unit on the quote, stripped to the clean rate and compared to the band for your vendor and metric, with the gap above the midpoint as the negotiation target.
The trajectory behind the bands, discounts falling across Oracle, SAP, Microsoft, and Salesforce since 2019, deepens the urgency, and the lever mechanics that close the gap run in the negotiation leverage report, the right sizing that must precede the rate chase in the shelfware report.
And the list movements the realized bands push against in the price increase index.
- Percentile standing for your exact deal size and industry, from real closed transactions
- Scenario simulation before the call: test alternative terms and see the financial impact of each
- A negotiation playbook, talking points, and a two page executive brief on day one
What we saw across negotiations, 2024 to 2025
Across roughly 160 to 200 enterprise negotiations and benchmarks our team supported between 2024 and 2025, the discount a buyer was first offered and the discount they finally signed were rarely the same number:
Deeper on deals with a credible documented alternative than on otherwise identical deals without.
By prepared buyers across the panel, the number every report in this family converges on.
The contrarian note earns the closing: chasing the discount percentage is the wrong game entirely when the estate is padded, because a deep discount on twice the needed licenses costs more than a modest discount on the right count, and the sequence that wins runs right size first, normalize second.
Band third, and negotiate last.
The compression from 34 to 27 percent means the discretionary room keeps shrinking as subscription unit rates firm, which shifts the recoverable money steadily from the rate conversation to the scope conversation, exactly where the buyer holds every card and uses none of them.
Your first five moves
- Strip your deal to the normalized unit rate, the clean per user or per core figure the bands actually compare.
- Right size before chasing any percentage, because the deep discount on padding is not a win.
- Place the deal in its band and target the top edge, never a headline average from a survey.
- Build the alternative and the calendar, the 8 to 15 and 5 to 12 points that beat every table tactic.
- Ignore headline discount depth without the list behind it, where Oracle and IBM flatter most. The cost optimization practice runs the benchmark with you.
Frequently asked questions
What discount do enterprises actually get off software list prices?
A band, not a number: realized discounts ran from about 5 percent to more than 70 across the eleven vendor panel in 2026, set by scope, term, leverage, and timing, each able to move the rate by double digits.
Prepared buyers realized roughly 40 to 60 percent of the vendor opening ask, and first quotes clustered 10 to 20 points below what prepared buyers in the same band eventually signed.
Why are published discount averages misleading?
Because they are not normalized: a blended average mixes a deep Oracle license discount with a shallow Broadcom transition quote and calls the midpoint a benchmark, comparing deals that share neither metric, term, edition, nor scope.
A defensible band ties to a normalized unit and stated scope, and the deepest headline discounts, Oracle and IBM, sit on the most inflated lists, so depth alone proves nothing.
Are software discounts getting bigger or smaller?
Smaller: the median realized discount across the panel compressed from 34 percent in 2021 to 27 in 2026, as vendors moved to subscription and held unit rates firmer.
The discretionary room shrinks each year, which shifts the recoverable money from the rate conversation toward the scope conversation, the right sizing lever entirely in the buyer's control.
What separates a top quartile software deal from a median one?
Timing, a real alternative, and the clauses, not harder negotiating on the day: the spread between median and top quartile ran about two times, built from the early calendar worth 5 to 12 points, the credible documented alternative worth 8 to 15, and the cap, co terminus.
And swap clauses that decide what the discount is still worth in year three.
How do you benchmark your own software deal?
Strip it to the unit you actually consume, the clean rate per user, per core, or per consumption unit rather than the unit on the quote, and compare that rate to the band for your vendor, metric, edition, and term.
Below the band midpoint you are doing well; above it, the gap is the negotiation target, closed by the calendar, the alternative, and the clause set.
Is a bigger discount always a better deal?
No: a deep discount on twice the licenses you need costs more than a modest discount on the right count, which is why the estate right sizes before any rate chase, and why the effective unit rate on the normalized scope is the only number that matters.
The sequence is right size, normalize, band, then negotiate, and the discount percentage is the last conversation rather than the first.
