An AWS private pricing commitment trades certainty of spend for a discount. The discount is the visible half of the deal. The commitment — and the shortfall payment attached to it — is the half that shows up on the P&L when forecasts are wrong.
The rule of thumbYour tolerance for missing the commit is roughly equal to your discount rate. At 8% off, landing below ~92% of commit leaves you worse off than having no agreement.
The asymmetryOver-committing costs cash. Under-committing costs leverage and discount depth. Both are fixable — but far cheaper to fix early in the term.
The leversTen levers decide the outcome. Four are set at signature, three are pulled mid-term, three are renegotiation tools.
What an EDP / PPA actually is
AWS sells most of its services at public, on-demand rates. Customers that can commit to a minimum level of spend over a fixed term can negotiate a private agreement that discounts those rates. AWS historically called this the Enterprise Discount Program (EDP); the current contract vehicle is commonly referred to as a Private Pricing Agreement or Addendum (PPA). The mechanics are the same, and this paper uses the terms interchangeably.
At its core, the agreement has three numbers and one clause:
Number 1CommitmentThe minimum eligible spend you agree to over the term, usually expressed per year. Commonly starts around $1M per year; larger customers sign far more.
Number 2TermTypically one to five years. Longer terms and larger commitments generally earn deeper discounts.
Number 3DiscountA percentage off eligible usage, applied on the invoice. Can be flat, tiered by spend band, or include service-specific private rates.
The clauseShortfallIf eligible spend in a measurement period is below commitment, the difference is invoiced. This is what turns a forecast error into a cash cost.
What counts toward the commitment
Not every dollar on an AWS invoice counts. The contract defines “eligible spend,” and the definition is where most forecasting errors begin. Typical treatment, which should always be confirmed against your own agreement:
Compute, storage, database, network usageUsually countsThe bulk of eligible spend. Counted after the private discount in most agreements.
Savings Plans & Reserved InstancesUsually countsThe commitment fees count — but they are lower than the on-demand spend they replace.
AWS SupportNegotiableTreatment varies by agreement; confirm whether support fees are inside eligible spend.
AWS Marketplace purchasesCappedThird-party software bought through Marketplace often counts up to a negotiated share of the commitment.
Spend billed through a reseller / partnerNegotiableDepends on how the reseller relationship and payer accounts are structured.
Promotional & migration creditsUsually excludedCredits reduce what you pay but generally do not count as committed spend.
Taxes, refunds, some third-party feesUsually excludedNever forecast these as progress toward commitment.
The lifecycle of a commitment
T − 6 months
ForecastModel eligible spend for the full term from run-rate, roadmap, migrations, and planned optimization.
T − 3 months
NegotiateSet commitment, term, ramp, discount structure, Marketplace allowance, and shortfall mechanics.
T = 0
SignThe discount applies from the effective date. The measurement clock starts.
Each quarter
TrackCompare eligible spend to the commitment pro-rata. The earlier a gap is visible, the more levers remain.
Each year end
True-upShortfall, if any, is calculated and invoiced for the measurement period.
Term end
RenewYour consumption history becomes your negotiating position for the next agreement.
The ten levers that decide the outcome
A commitment is not a single decision. It is a set of dials, some fixed at signature and some adjustable throughout the term. Finance teams that treat it as a one-time procurement event tend to discover the mid-term dials only after a true-up invoice arrives.
Set at signatureHardest to change later
01
Commitment amountThe headline number. Determines discount depth and shortfall exposure at the same time.
Trade-offHigher commit, deeper discount, larger downside if spend underdelivers.
Pull it whenYou have a defensible forecast with known downside cases — not just a growth target.
02
Term lengthMore years usually means a better rate, but every extra year compounds forecast uncertainty.
Trade-offDiscount gain versus loss of flexibility in architecture and vendor choices.
Pull it whenYour platform roadmap is stable and multi-year cloud strategy is settled.
03
Ramp shapeAnnual commitments can step up over the term rather than stay flat, matching expected growth.
Trade-offBack-loaded ramps protect year one but concentrate risk in later years.
Pull it whenLarge migrations or product launches land mid-term.
04
Shortfall mechanicsWhether commitment is measured annually or cumulatively, and whether shortfall can roll forward into later years or a renewal.
Trade-offFlexibility is usually bought with a slightly lower discount.
Pull it whenSpend is lumpy or seasonal, or growth timing is uncertain.
Pulled mid-termOperational levers
05
Eligible-spend routingBuy third-party software through AWS Marketplace, and move partner-billed or acquired accounts under the agreement payer.
Trade-offProcurement and vendor-contract changes; Marketplace caps still apply.
Pull it whenTracking short and you already buy software that is listed on Marketplace.
06
Workload timingPull forward planned migrations, data-center exits, or new regions to land spend inside the measurement period.
Trade-offAccelerated projects carry delivery risk and may run in parallel with on-prem.
Pull it whenThe migration was going to happen anyway and the gap is timing, not volume.
07
Optimization pacingSequence rightsizing, Savings Plans purchases, and cleanup against the commitment curve.
Trade-offDelaying savings costs money; taking them blind can create shortfall.
Pull it whenAlways — optimization and commitment tracking should share one forecast.
Renegotiation toolsRequire AWS agreement
08
Early renewalReplace the current agreement before term end with a new one sized to actual consumption.
Trade-offLocks in a new, usually longer term in exchange for relief or a better rate.
Pull it whenYou are materially ahead (more discount available) or materially behind (shortfall relief needed).
09
Term extensionSpread the remaining commitment over more months so annual targets fall.
Trade-offTotal commitment unchanged; you trade time for relief.
Pull it whenThe full-term total is achievable, but not on the original schedule.
10
Service-specific pricingPrivate rates on individual high-volume services (for example data transfer or a specific storage class) alongside the blanket discount.
Trade-offMore complex invoices and tracking; requires concentrated spend.
Pull it whenA small number of services dominate your bill.
The optimization paradoxEngineering savings — rightsizing, Savings Plans, cleanup — reduce the dollars counted against the commitment. A successful cost program can push a well-sized commit into shortfall. Optimization and commitment sizing have to be planned together, not by separate teams.
Over-committed, under-committed, and the breakeven between
Two failure modes, with very different consequences. Over-committed means you signed for more than you consume: the gap is invoiced as a shortfall payment. Under-committed means you consume more than you signed for: nothing is invoiced, but you are paying a shallower discount than your spend could command.
Over-committed · spend < commitYou pay for spend you never used
- Shortfall is invoiced at the end of the measurement period — usually annually, sometimes at term end.
- It is an unbudgeted cash outflow that buys no capacity, and typically has to be recognized as an expense.
- A miss larger than your discount rate wipes out the agreement’s entire benefit.
- Leverage at renewal drops: AWS sees a customer who cannot hit plan.
Under-committed · spend > commitYou leave discount on the table
- The negotiated discount still applies to all eligible spend — there is no penalty.
- But discount depth is priced on the committed number, not the actual one.
- Every point of discount you did not negotiate is lost on 100% of your spend, every year of the term.
- Sustained over-delivery is your strongest case for an early renewal at a deeper rate.
The breakeven rule
Let d be the discount and u the share of commitment you actually consume (measured on post-discount spend). The discount saves you u · C · d / (1 − d); the shortfall costs (1 − u) · C. The two are equal when u = 1 − d. Below that line, the agreement is a net loss.
Worked example · $3.0M annual commit · 8% discount · 3 years
Scenario · annualDiscount earnedShortfall paidNet vs. no dealOn target · $3.00M eligible$261k$0+$261kShort 5% · $2.85M$248k$150k+$98kShort 8% · $2.76M (breakeven)$240k$240k$0Short 20% · $2.40M$209k$600k−$391kAhead 30% · $3.90M$339k$0−$85k forgone*
Illustrative. Assumes commitment measured on post-discount eligible spend, annual true-up, no carry-forward. The under-committed row shows the value of 2 extra discount points not negotiated.
Early-warning signals
ShortTwo consecutive quarters below pro-rata commitment.
ShortA large Savings Plan or rightsizing program approved without a commitment impact check.
ShortA planned migration or launch slips by more than a quarter.
AheadRun-rate exceeds the annual commitment before the year is out.
AheadAn acquisition brings significant AWS spend outside the agreement.
EitherForecast owned by procurement with no engineering input.
Remediation playbook
If you are tracking short
+Quantify the gap on eligible spend, not invoice totals — credits and excluded items distort the picture.
+Route eligible third-party software through AWS Marketplace, up to the allowed cap.
+Bring partner-billed or acquired AWS accounts under the agreement payer.
+Re-sequence optimization: pace commitment-reducing changes against the true-up date.
+Open the conversation with AWS early — extension, restructure, or roll-forward are negotiable well before the true-up, rarely after.
+Insure or reserve for the residual shortfall so it never lands as a surprise.
If you are running ahead
+Document the consumption trend — it is your negotiating evidence.
+Price the opportunity: each extra discount point is worth 1% of eligible spend for every remaining year.
+Propose an early renewal at a commitment sized to the new run-rate, with downside protection built in.
+Fold acquired or shadow accounts into the agreement before negotiating, to show the full number.
Plan. Protect. Mitigate.
01 · ForecastA 36-month eligible-spend model built from billing data, roadmap, and planned optimization — so the commitment is sized to a number you can defend.
02 · InsureShortfall insurance underwritten against measured usage, covering the gap between commitment and consumption up to the policy limit.
03 · MitigateFor agreements already off-track: restructuring, eligible-spend recovery, and direct advocacy with your AWS account team.
EDPEnterprise Discount Program — the historical name for AWS’s committed-spend discount agreement.
PPAPrivate Pricing Agreement / Addendum — the current contract vehicle for AWS private pricing.
Eligible spendThe subset of AWS charges that count toward the commitment, as defined in the contract.
ShortfallThe amount by which eligible spend falls below commitment in a measurement period; invoiced to the customer.
True-upThe periodic reconciliation of eligible spend against commitment.
RampA commitment schedule that increases year over year rather than staying flat.
Marketplace capThe maximum share of commitment that AWS Marketplace purchases may satisfy.
UtilizationEligible spend as a percentage of commitment. Below 100% means shortfall exposure.
This paper describes common structures observed across AWS private pricing agreements. Terms are negotiated individually and change over time; nothing here replaces a review of your own contract. Figures are illustrative.
Put your own numbers through the model.
The interactive explorer runs the same math as this playbook against your commit, discount, growth, and levers.