A price file freeze was available on seven of ten private pricing agreements when asked for, and never offered by default

EDP and PPA are one instrument now, so the useful question is not which one you are on but which mechanism you are arguing. The discount line is the part everyone negotiates. The governance clauses are the part that was never in the draft and were granted most of the time when someone asked.

Prepared by Redress Compliance · August 16, 2026 · AWS advisory. 25 to 35 private pricing negotiations and benchmarks, 2024 to 2025.

The private pricing addendum is the master discount contract sitting under the AWS Customer Agreement, and the Service Order is the order form that executes against it. Knowing which document carries which term is what stops a negotiation arguing the right point against the wrong paper.

• The price file freeze locks AWS public list price at signature, so your negotiated percentage applies to a fixed base rather than to a list that can move underneath it. Available on roughly seven of ten agreements when requested explicitly, never offered by default, and worth 4 to 8 percent across three years.
• Change of control language decides what happens on acquisition or divestiture, and the default draft protects AWS rather than you. Read it before the corporate development team needs it.
• Audit reciprocity is asked for far less often than it is granted, and it costs nothing to include when the relationship is new.
• True up posture belongs in the addendum, not in a conversation. How a shortfall is measured and when it is assessed are separate terms and both are negotiable.
• Termination and exit clauses exist but rarely get used, which is precisely why they are conceded easily. Ask while the deal is open.
• None of these appear in the default draft. That is the pattern worth internalising: the discount line is contested and the governance clauses are simply absent until a buyer names them.

A private pricing negotiation has a shape that repeats across estates. The buyer prepares on the discount, the account team arrives with a band already sized against the commitment, and both sides spend the meeting on a number whose range was determined before anyone sat down. The discount steps by commitment threshold, the thresholds are known, and the movement available inside a band is small. Meanwhile the clauses that carry real multi year value are not being argued at all, because they are not in the document to argue about.

The price file freeze is the clearest example. A negotiated discount is a percentage applied to AWS public list, and public list is a document AWS controls and revises. Without a freeze, the percentage you won is a percentage of a moving base, which means part of your discount can be recovered by the publisher through ordinary list movement across a three year term. Freezing the price file at signature converts the discount into a fixed reduction against a fixed base. It was available roughly seven times in ten when a buyer asked for it explicitly, it was never in the default draft, and it was worth 4 to 8 percent across three years. That is comparable to an entire discount band, obtained by naming a clause rather than by winning an argument.

The same asymmetry runs through the other governance terms. Change of control language, audit reciprocity, true up measurement basis, and termination posture are all cheap for AWS to concede at signature and expensive for a buyer to need later without them. They are absent from the draft not because they are contentious but because nobody requests them, and a clause nobody requests is a clause the vendor has no reason to volunteer. The buyer side discipline is to arrive with a governance list alongside the commercial ask, and to treat the two as one negotiation rather than as a main event and a formality.

Timing then decides how much of any of this you can hold. Renewal lift ran 8 to 18 percent above the prior contract value across the cycles benchmarked. Buyers who arrived 270 days out with refreshed benchmark data closed that gap or improved on their prior tier, because at nine months there is time to build a credible alternative position and the account team's forecast is still open. Inside 90 days the lift held in AWS's favour, since the buyer has no runway and the vendor knows it. The commitment sizing mechanics sit in the EDP pillar, the discount bands in the discount benchmarks, and the wider library in the AWS practice.

Renewal lift averaged 8 to 18 percent above prior contract value. The variable that moved it was not negotiating skill but calendar position, because every lever that matters, the alternative, the threshold, and the governance list, needs months rather than meetings.

Across roughly 25 to 35 AWS private pricing negotiations and benchmarks:

Governance clauses, meaning price file freezes, change of control, and audit reciprocity, were rarely included in the default AWS draft. That is the finding that generalises: the contested terms get contested, and the uncontested ones are simply missing.

Service specific rates on compute, storage, networking, and managed services are negotiable within the overall ceiling, and term length is what AWS pays for. Three and five year commitments unlock deeper steps than one year terms, which is a structural preference rather than a concession.

Watch the briefing · 3:59Negotiating AWS 1: What You Are Actually SigningWhy EDP and PPA are one instrument, why the commitment is a floor rather than a budget, and the honest discount range.

Are EDP and PPA two different agreements?

They are best treated as one instrument. The private pricing addendum is the master discount contract under the AWS Customer Agreement and the Service Order is the order form. Arguing about which label applies wastes the meeting; name the mechanism you are negotiating instead.

What discount should we actually expect?

The honest range is 5 to 20 percent, stepped by commitment threshold: roughly 5 to 10 percent between $1m and $5m, 10 to 15 percent between $5m and $25m, and 15 to 20 percent above $25m. Any model that reaches above 20 percent by stacking instruments is modelling a deal that does not exist.

What is a price file freeze?

A clause locking AWS public list price at the point of signature, so your negotiated percentage applies to a fixed base rather than to a list AWS can revise underneath you. Without it, part of the discount you won can be recovered through ordinary list movement across the term.

How often is a price file freeze granted?

On roughly seven of ten agreements where a buyer requested it explicitly. It was never offered in a default AWS draft. Across a three year term it was worth 4 to 8 percent, which is comparable to an entire discount band.

Which governance clauses are missing by default?

Price file freezes, change of control language, audit reciprocity, true up measurement basis, and termination posture. They are absent not because they are contentious but because they are rarely requested, and a clause nobody asks for is one the vendor has no reason to volunteer.

When should renewal preparation start?

270 days before expiry. At nine months there is time to refresh benchmarks, build a credible alternative, and restructure the threshold, and the account team forecast is still open. Inside 90 days only the discount argument remains available.

How large is the renewal lift?

It ran 8 to 18 percent above prior contract value across the cycles benchmarked. Buyers arriving 270 days out with refreshed benchmark data closed that gap or improved on their prior tier. Those arriving at 90 days absorbed it.

Are service specific rates negotiable?

Yes, within the overall ceiling. Compute, storage, networking, and managed services can each carry custom rates, and pricing your heaviest lines individually is more productive than accepting a single blended figure across the estate.

Does term length change the discount?

Yes. AWS pays for term length, so three and five year commitments unlock deeper steps than one year terms. That is a structural preference in the pricing model rather than a concession extracted at the table.

Do exit clauses matter if we never use them?

They matter precisely because they are rarely exercised, which is why they are conceded relatively easily at signature. Termination and change of control language cost little to include while the deal is open and cannot be obtained when you actually need them.