Contents
Key takeawaysHow the EDP worksWhat we have seenThe six provisionsSizing the commitBigger commit, bigger discount?Getting better termsWhat to do nextFAQAn AWS EDP discount is fixed on signing day, but your usage is not. The ramp, shortfall, carry forward, Marketplace, eligibility and reopener clauses decide what a missed forecast costs you.
- Commit for discount. You promise a minimum annual AWS spend over one to five years, and AWS discounts eligible usage across the bill.
- The risk is one sided. Overshoot the commit and you pay for the extra at your discounted rate; undershoot it and you still owe the shortfall.
- Six clauses hold the flexibility. Ramp schedules, shortfall treatment, carry forward, Marketplace counting, service eligibility and renegotiation triggers.
- Marketplace spend can fill a gap. Third party purchases through AWS Marketplace can count toward the commit when first party usage lags, if the amendment says so.
- Growth forecasts sink commits. A commit built on a hockey stick forecast tends to become a shortfall invoice in year two.
- Every term is negotiable. EDPs are individually negotiated contracts, and AWS changes the provisions when you ask with usage evidence in hand.
How does the AWS Enterprise Discount Program work?
The AWS Enterprise Discount Program (EDP) exchanges a multi year annual spend commitment for a discount across the bill. Terms run from one to five years, and three years is the most common term in our engagements.
AWS publishes its standard rates and pricing models on its AWS pricing page, but every EDP is a privately negotiated agreement. Newer AWS paperwork often calls the same commitment a Private Pricing Agreement (PPA), which our PPA buyer guide covers.
- Commit. A minimum annual spend, usually growing year over year.
- Discount. A percentage off eligible usage across the bill, fixed at signature.
- Term. One to five years.
- Shortfall. Spend below the commit is still owed. This clause creates almost all of the risk.
Why the risk sits almost entirely with you
The standard terms favor AWS. If your usage runs above the commit, you pay for the extra at your discounted rate and nothing else happens. If it runs below, you pay the gap in cash and receive no usage for it.
Your usage keeps changing with migrations, optimization work and acquisitions, while the discount stays fixed. That is why the flexibility provisions are worth more than the headline percentage.
Negotiating AWS 8: Size and Structure the Commit
What have we seen in recent AWS EDP negotiations?
The flexibility provisions decided more value than the discount rate. Morten Andersen advised on roughly 25 to 30 AWS EDP negotiations and renewals in 2024 to 2025, and three patterns came up again and again.
- Finance forecasts overshot. Commits sized on finance forecasts, rather than engineering forecasts, came in above actual usage in roughly a third of cases. The discount turned into shortfall exposure.
- Year one ramps paid off. Buyers who negotiated a ramp into year one avoided 10 to 20 percent of effective overcommitment compared with a flat annual commit.
- Early renewals scored higher. Renewals opened 12 months before expiry, with usage data in hand, landed discounts 3 to 8 points higher than renewals run in the final quarter.
AWS EDP Flexibility Checklist
The six provisions with sample wording, plus the worksheet for sizing a commit you can hit.
Get the white paper →Which flexibility provisions should you negotiate into an AWS EDP?
Six provisions carry almost all of the flexibility: a year one ramp, defined shortfall treatment, carry forward of overage, Marketplace counting, broad service eligibility and renegotiation triggers. Each one is negotiable, and none of them appears in the first draft AWS sends.
| Provision | What it does | What to ask for |
|---|---|---|
| Ramp schedule | Lowers the commit in year one | A commit that follows your real migration timeline |
| Shortfall treatment | Defines what a miss costs | Misses roll into carry forward instead of a cash penalty |
| Carry forward | Overage counts toward next year | Surpluses roll forward on the same terms as misses |
| Marketplace counting | Third party spend fills the commit | Written counting rules under the AWS Marketplace program |
| Service eligibility | Sets which services earn the discount | Broad scope, including newer AI services |
| Renegotiation trigger | Reopens terms on a major change | M&A, divestiture or a workload exit reopens the commit |
Use the table as your agenda, and send it before AWS sends a term sheet. Conceding a discount point in exchange for two or three of these provisions is usually a good trade in net present terms.
Ramp schedule and shortfall treatment
These two work as a pair. The ramp keeps the year one commit close to the usage you can prove, and the shortfall clause decides what happens when you still fall short.
- Ramp wording. Ask for a year by year commit schedule in the amendment, with year one set at or below your sized baseline. A written schedule stops the account team from reading the total as a flat annual figure.
- Shortfall wording. Ask that any shortfall below the annual commit rolls into the next contract year instead of being invoiced. If AWS refuses, ask for the invoice to be deferred to the end of the term, when later overage can offset it.
- Measurement date. Ask for the date and the spend definition (net of discount, before or after credits) to be written down. Both change how much usage counts.
Carry forward that runs in both directions
If AWS allows a miss to roll forward, overage should roll forward too. Ask that spend above the commit in any year counts toward the next year, and ideally across the rest of the term. Without that symmetry, a strong year three does nothing to repair a slow year one.
Marketplace counting
Third party software bought through AWS Marketplace can help fill the commit when first party usage lags. The counting rules differ from agreement to agreement, so get the percentage that counts and any cap written into the amendment. Our Marketplace and EDP guide covers the detail.
Check each listing as well. Since May 1, 2025, AWS Marketplace accepts SaaS products hosted outside AWS, and products verified as running on AWS carry a "Deployed on AWS" badge. AWS says only that badged products may qualify for additional customer benefits, so get written confirmation that a vendor you plan to buy through Marketplace counts toward your commit.
Service eligibility
Ask for the discount to apply to all generally available services, including services launched during the term. New AI services such as Amazon Bedrock are where many buyers expect their spend to grow, and a narrow service list leaves that growth undiscounted.
Renegotiation triggers
A three year commit assumes your business stays the same shape for three years. Ask for named events that reopen the commit: an acquisition, a divestiture, or the exit of a defined workload. Tie each trigger to a measurable threshold, such as a change in annual spend, so the clause cannot be argued away.
How should you size an AWS EDP commit you can actually hit?
Size the commit from engineering's trailing 12 month usage plus committed projects only. Leave finance growth targets out of it. The commit is a floor you must pay in cash, so every speculative workload in the number is risk you carry on AWS's behalf.
A sizing method that holds up in year two
- Take trailing 12 month spend from Cost Explorer as the baseline.
- Add only projects with signed budgets and named owners.
- Subtract workloads with exit plans, including anything moving to another cloud.
- Apply a haircut of 10 to 15 percent for execution slippage. In our engagements, this haircut is what kept buyers clear of shortfalls when projects slipped.
- Set the year one commit at or below the result, and let the ramp carry growth in later years.
A worked example
Say your optimized trailing 12 month net spend is $4,000,000. You have $800,000 of new projects with signed budgets, and $500,000 of workloads scheduled to leave AWS.
| Step | Adjustment | Running total |
|---|---|---|
| Trailing 12 month net spend | Baseline | $4,000,000 |
| Projects with signed budgets | +$800,000 | $4,800,000 |
| Workloads with exit plans | minus $500,000 | $4,300,000 |
| Option A: 10 percent haircut on $4,300,000 | minus $430,000 | $3,870,000 |
| Option B: 15 percent haircut on $4,300,000 | minus $645,000 | $3,655,000 |
A year one commit between $3,655,000 and $3,870,000 is one you can prove. If the account team proposes something nearer $4,800,000, ask which exits and which slippage their figure assumes.
Why optimization has to come first
Savings Plans, reserved capacity and rightsizing all reduce billed spend, and billed spend is what fills the commit. Run the optimization program first and size on the optimized run rate. Otherwise the EDP penalizes your own efficiency work with a shortfall invoice.
Savings Plans cut compute rates by up to 72 percent against On Demand. A large purchase made after signing can drop billed spend below a commit sized on the old run rate.
How to check your own numbers
- Cost Explorer. Shows up to the last 13 months of cost data by default. Switch the view to net unblended or net amortized cost and group by service and linked account for the baseline. Its 18 month forecast is a trend line and should not become the commit.
- Savings Plans recommendations. The recommendations page in the Billing and Cost Management console shows the hourly commitment AWS suggests for your usage. Deduct the saving from that purchase before you size the EDP.
- AWS Compute Optimizer. Lists rightsizing candidates, which tells you how much of today's run rate will not exist next year.
- Marketplace invoices. Pull current third party spend bought through Marketplace, so you know what could count toward the commit and what could not.
Should you commit more to get a bigger AWS discount?
Usually you should not. A larger commit earns more discount points, but the extra points rarely cover the shortfall and forced spend the larger commit creates.
Why we advise against maximizing the commit
AWS, and most consultancies, advise maximizing the commit to maximize the discount. We disagree. In roughly 8 of the more than 25 EDP engagements Morten Andersen benchmarked in 2024 to 2025, the larger commit produced a worse net position. Shortfall payments and year end spend to close the gap cost more than the extra points returned.
The better course is to commit to the floor you can prove and accept a point or two less. Spend the negotiating time on ramp, carry forward and Marketplace counting, because those provisions save cash in the years when usage misses the plan.
Go back to the sizing example. Say AWS offers one extra discount point for a $4,800,000 year one commit instead of $3,870,000. On about $4,000,000 of real annual spend, that point is worth roughly $40,000 a year, or $120,000 over three years.
Now let one migration slip two quarters so year one spend lands at $4,000,000. The larger commit leaves an $800,000 shortfall, more than six times what the extra point returns over the whole term.
The discount is fixed on the day you sign. The flexibility provisions decide whether the commit still fits your business in year two.
What gets AWS to improve EDP terms?
Timing, usage evidence and credible workload portability change EDP terms more than procurement pressure. AWS account teams forecast renewals quarters ahead, and a buyer who starts 12 months early negotiates against that forecast instead of against a deadline.
- Start early. Open the renewal 12 months out with your usage analysis finished, so AWS has to respond to your numbers.
- Show portability. Containerized workloads with tested multi cloud deployment change the conversation, even if you never move them.
- Trade precision for points. A tighter commit with a ramp and carry forward is worth more than a loose commit with a bigger headline discount.
Write every agreed provision into the EDP amendment itself. Side letter assurances and emails from account teams do not survive team changes, and AWS rotates enterprise sellers often.
What the account team will say, and what to say back
- "A bigger commit puts you in a better discount tier." Ask for the discount at the commit you sized, with a ramp. Then compare the difference in points with your modeled shortfall.
- "Carry forward is not standard." Every EDP term is negotiated, the discount included. Ask them to take the request to their deal approval team and come back with a written answer.
- "Your Marketplace spend will easily cover any gap." Then the counting rule and cap can go into the amendment, and you should ask for exactly that.
- "We can fix a shortfall at renewal." A fix at renewal usually means a larger new commit. Ask for the shortfall treatment now, while AWS still wants your signature.
A renewal timeline
| Months before expiry | What to do |
|---|---|
| 12 | Pull the Cost Explorer baseline, start the Savings Plans and rightsizing program, and tell AWS you are opening the renewal |
| 9 | Finish the sizing model with engineering and list the committed projects and exits |
| 6 | Send your six provision agenda and your proposed ramp before AWS issues a term sheet |
| 3 | Negotiate the amendment text line by line, including measurement date and Marketplace counting |
| 1 | Check that every agreed point is in the signed amendment and not only in email |
How a first EDP differs from a renewal
On a first EDP you have little history on AWS, and migration delay is the usual reason for a year one miss. A low year one commit and a ramp tied to the migration plan matter most here.
At renewal AWS starts from your current run rate plus growth, and may offer to fold an existing shortfall into a larger new commit. Keep those two discussions separate. Our EDP renewal guide and shortfall risk guide go further on both.
What to do next
- Build the baseline. Pull trailing 12 month usage from Cost Explorer and size the commit from engineering data.
- Add only funded projects. List committed projects with signed budgets and named owners, and add only those.
- Optimize first. Run the Savings Plans and rightsizing program before you size the commit.
- Draft your agenda. Write up the six flexibility provisions before AWS sends a term sheet.
- Start early. Open the renewal conversation 12 months before expiry.
- Put it in the contract. Get every agreed provision written into the EDP amendment. Our clause redlines guide shows sample wording.
Frequently asked questions
What is an AWS EDP?
An AWS Enterprise Discount Program is a privately negotiated agreement that exchanges a multi year annual spend commitment for a percentage discount on eligible AWS usage. Terms typically run one to five years, and every element, from the ramp to the service list, is negotiated for that customer.
What happens if you miss an EDP commitment?
You pay the difference. Spend below the committed level is still payable to AWS unless the contract says otherwise, and the payment buys no usage. Shortfall treatment and carry forward terms agreed before signing are the only reliable protection, since AWS has little reason to concede them once the gap exists.
Does AWS Marketplace spend count toward an EDP commit?
It can, when the agreement provides for it. That makes third party software a practical way to fill a commit when first party usage lags. Treatment varies by agreement and by listing, so confirm the counting percentage, any cap and the eligible sellers in your own amendment.
How big does AWS spend need to be for an EDP?
EDPs generally start around one million dollars in annual AWS spend, and the strongest terms appear well above that. Below that level, Savings Plans and reserved capacity remain the main discount instruments, and they carry no cross bill shortfall risk.
Is a bigger EDP commit always better?
No. In our 2024 to 2025 engagements roughly a third of oversized commits cost more than the extra discount points returned. Commit to the spend you can prove from usage data, then negotiate the ramp and carry forward terms that protect you if a project slips.
When should you start an EDP renewal?
Start twelve months before expiry. An early start means you negotiate against AWS's own renewal forecast rather than your deadline, and in our benchmarks it landed 3 to 8 discount points more than a renewal left to the final quarter.