Three commitment instruments, layered in the wrong order, cost more than no commitment at all. Getting the order right took twenty percent out of the run rate.
Negotiating AWS 10: Terms That Outlast the Discount
The base agreement changes on posting with no notice, service credits require you to prove AWS's outage, there is no durability SLA at all, AI data use is opt in by default across security tooling, and the regulator has closed the file.
A German online services company with around five thousand employees was renewing its AWS commitment. Its platform runs on AWS across compute, storage, database, networking and the AI services, so leaving was never the question.
The question was how much to commit and through which instrument. AWS offers three that overlap, and the order you apply them in changes the answer by a large margin.
The engagement took roughly twenty percent out of the AWS run rate across a three year term.
Around five thousand employees, operating across Europe, with a consumer facing platform carrying real traffic variation between weekday and weekend.
That variation matters more than the headline spend. An estate with a stable floor and a variable peak is exactly the shape where the choice between reservations and savings plans stops being academic.
The AI services line was the fastest growing part of the estate and the hardest to forecast, which is now true of most AWS customers we see.
AWS opened with an Enterprise Discount Program commitment sized on a growth curve extended forward from recent consumption at full confidence.
The proposal treated the commitment as the primary lever and the reservation layer as an afterthought, which is the wrong way round for an estate with a large stable base.
It also assumed the AI services growth would continue at its recent rate for three years. That may prove correct. It is not something to commit to contractually.
We rebuilt the position from the customer's own billing data and separated the estate into three layers before discussing any commitment.
The stable base, which never varies and should be reserved outright. The predictable variable layer, which suits a savings plan because it moves between instance families. And the genuinely unpredictable layer, mostly the AI services, which should carry no commitment at all.
Only after those three were sized did the Enterprise Discount Program number make sense, because the commitment should sit underneath what remains rather than on top of everything.
These are the moves we ran. The first four set the number.
The layering is the transferable lesson. Committing to everything at one rate looks simpler and consistently costs more than committing to each layer at the rate that suits it.
The eleven moves, splitting the estate into stable, variable and unpredictable layers, sizing the commitment from evidence, and the buyer side position at every step of an AWS renewal.
Used across more than five hundred enterprise clients. Independent. Buyer side. Built for IT procurement leaders running the next AWS renewal cycle.
AWS framed the AWS Enterprise Discount Program as the immediate AWS uplift across the broader AWS. Redress reframed the approach around the customer's actual AWS utilization and applied the AWS Reserved Instance. Twenty percent saved against the publisher's opening AWS renewal quote.
Vendor management, contract negotiation, audit defense, renewal strategy. One firm. Eleven practices.
AWS commit signals, AWS Enterprise Discount Program signals, AWS data transfer egress signals, and the broader AWS licensing leverage signals across the practice.
