How a German online services company cut AWS spend by thirty one percent.

A German online services group reached the end of a three year AWS Enterprise Discount Program commitment of 40 million dollars. Its consumer platform runs on AWS, so this was not a negotiation the business could walk away from, and both sides knew it.

AWS opened the renewal at 55 million dollars over three years. The customer engaged Redress Compliance fourteen weeks out. The agreement closed at 38 million, a 31 percent reduction against the opening position and 12.4 million dollars saved across the term.

What follows is the whole engagement, including the analysis that set the counter and the nine moves that held it. For the wider practice, see the AWS advisory practice, the AWS EDP negotiation, the AWS vendor management playbook, and the AWS RI and Savings Plan optimization.

The situation

The group sells online services across the European Union and the wider European market. Its platform sits on AWS, spanning compute, storage, data transfer, database, analytics and security. There was no realistic scenario in which AWS left the estate.

The expiring commitment was 40 million dollars across three years. Actual consumption had grown to roughly 15 million dollars a year, which put the customer close to the original commitment by the end of the term. Growth had been real, and AWS had the data.

Two details changed the shape of the negotiation. The customer also ran workloads on Azure and Google Cloud, so a genuine multi cloud position already existed. And nobody internally had modelled what the next three years of consumption would actually look like.

What AWS proposed

AWS put 55 million dollars over three years on the table. The number was built by extending recent growth forward at full confidence, which is a reasonable thing for a vendor to do and a poor thing for a customer to accept.

The proposal bundled several things that looked like concessions. A Savings Plan sized to broad compute coverage. A data transfer position that left egress at standard rates. Marketplace terms set at the vendor default rather than tuned to the customer's third party software spend.

The discount tier was the sweetener. A larger commitment unlocks a better rate, which is exactly the mechanic that persuades buyers to commit past their real consumption. A better rate on volume you never use is not a saving.

What the numbers actually showed

We rebuilt the consumption forecast from the customer's own billing data rather than from the growth curve AWS had drawn. The exercise ran across four commercial dimensions, and each one moved the number.

• Commitment sizing. Modelled against an 85 percent confidence consumption trajectory instead of a 100 percent one. This single change accounted for most of the gap.
• Savings Plan coverage. Sized to the compute footprint that was genuinely stable, not to the whole estate.
• Data egress. Priced separately and negotiated explicitly, rather than absorbed into the headline commitment.
• Contract terms. Tested against what the business actually needed rather than the vendor's standard paper.

The reconstructed position came out roughly 30 percent below the AWS proposal. That figure was defensible line by line, which mattered later when AWS pushed back on the counter.

We also costed the alternative. With Azure and Google Cloud already carrying workloads, moving a defined slice was a real option with a real price attached. That turned the renewal from a captive conversation into a competitive one. See the AWS Azure GCP competitive analysis and the midwestern US bank AWS case study.

The nine moves

Nine moves carried the negotiation, and they compound. The early ones set the number and the later ones protect it.

1. Size the commitment honestly. Anchor to an 85 percent confidence consumption forecast, not the vendor's full confidence extrapolation.
2. Separate the discount tier from the commitment. Negotiate the rate on its own merits rather than buying it with volume you do not need.
3. Right size the Savings Plan. Cover the stable compute base and leave the variable layer flexible.
4. Negotiate egress explicitly. Data transfer relief is a separate line and should be argued as one.
5. Layer Reserved Instances deliberately. Target the genuinely steady workloads above the Savings Plan baseline.
6. Route third party software through Marketplace. Existing software spend can count toward the commitment, which lowers the commitment you need.
7. Make the alternative real. Cost the Azure and Google Cloud option properly so the competitive position survives scrutiny.
8. Negotiate the terms, not just the price. Risk allocation and flexibility outlast the discount.
9. Set up the next renewal now. Build the reporting that means you arrive at the next cycle with your own numbers.

The counter proposal

The counter landed at 38 million dollars over three years against the 55 million opening. It was not a haggling position. Every element traced back to the customer's own consumption data.

Alongside the number, the counter carried explicit data egress relief, a resized Savings Plan, a targeted Reserved Instance layer, Marketplace routing for third party software, and a rewritten set of commercial terms.

Presenting it as one coherent package mattered. A list of separate asks invites a vendor to concede the cheap ones and refuse the rest. A single position built from evidence is much harder to unpick.

The close

Contract finalisation ran four weeks. The commitment held at 38 million dollars, and the discount tier came in 35 percent better than the AWS proposal had offered at that commitment level.

Egress relief was written in explicitly. The Savings Plan and Reserved Instance layers were sized to the actual compute profile. Marketplace routing was agreed so third party software spend counts toward the commitment.

The renewal mechanics were negotiated alongside the customer's Azure MACC and Google Cloud committed use positions, so the three cloud commitments are now managed as one portfolio rather than three unrelated renewals. See the Azure MACC negotiation and the GCP negotiation leverage notes.

The outcome

• Commitment. 38 million dollars over three years against a 55 million dollar opening. A 31 percent reduction.
• Discount tier. 35 percent better than the tier AWS offered at the equivalent commitment level.
• Data egress. Explicit relief written into the agreement, cutting the egress run rate materially.
• Savings Plan. Sized to the stable compute base rather than to broad estate coverage.
• Reserved Instances. Targeted at steady workloads and layered above the Compute Savings Plan baseline.
• Contract terms. Rewritten risk allocation and improved flexibility across the agreement.
• Total saving. 12.4 million dollars across the three year term.

What to do next

If an EDP renewal is on your horizon, these are the steps that matter, in order.

1. Pull twenty four months of billing data and build your own consumption forecast before AWS builds one for you.
2. Model the forecast at 85 percent confidence and treat that number, not the optimistic curve, as your commitment ceiling.
3. Separate egress, Savings Plan coverage and Marketplace routing into their own negotiating lines.
4. Cost a real alternative on Azure or Google Cloud for a defined slice of the estate.
5. Check whether third party software spend can route through Marketplace and count toward the commitment.
6. Open the conversation at least fourteen weeks out, and put contract terms on the table alongside price.

Lessons learned

Three things travel from this engagement to any AWS renewal.

1. The commitment is the negotiation. Discount rate is where attention goes, but commitment size is where the money is. An excellent rate on a commitment you cannot consume is an expensive mistake in a smart suit.
2. A competitive position has to be costed. Saying you have alternatives changes nothing. Showing a priced migration plan for a defined workload changes the room.
3. Terms outlive prices. The discount applies for three years. Egress treatment, flexibility and risk allocation shape every renewal after this one.

The full method is set out in our AWS EDP Negotiation Guide, the AWS vendor management playbook, and the AWS RI and Savings Plan optimization notes.

Related reading: the AWS data transfer and egress negotiation notes, the global technology AWS case study, the midwestern US bank AWS case study, and the Azure cost optimization playbook.