It’s uncommon, but some companies find themselves in the position of missing their EDP/PPA commitments—sometimes by substantial amounts.
A classic scenario: You signed an agreement to spend $150 million over three years, fully expecting to blow past that with all your growth—but now your past spend and projections are showing you’ll miss that by a healthy enough margin to cause concern.
The Zen of AWS Cost Management
The first thing I tell everyone in this situation: chill, dude.
This is not an emergency. It’s not even necessarily a problem.
Engineers in particular (speaking as one myself!) have a visceral reaction to paying for something that isn’t getting used. After all, we’re trained to optimize systems and eliminate waste. Paying for nothing fundamentally conflicts with our values.
But consider an alternative perspective: you already decided to spend this money, years ago. It was within the budget and the forecasts supported it. That decision is in the past. You bought an option and didn’t exercise it.
Your focus now needs to be on the total financial picture—including all the discounts you’ve already received and will continue to receive—rather than the emotional reaction to “paying for nothing.”
Fortunately, you have options.
What You Can Do
If you’re facing a shortfall, there are a few courses of action you can take to close the gap.
Make Some Upfront SP/RI Purchases
Both the upfront payments and ongoing usage charges from Savings Plans and Reserved Instances purchases can count against your EDP spend commitment. If you have the cash, stock up on savings by prepaying for compute you know you’ll use. Of course, this option has gotten less attractive post-ZIRP with interest rates on the rise—it may be that the cash is more valuable in your bank account than the savings you’ll get from the upfront SP/RI purchase.
Consume More Services
Don’t do this. It’s true that one way to avoid paying for something you don’t use is to find a way to use up what you paid for, but this leads to terrible decisions like:
- Spinning up resources you don’t need
- Moving workloads to AWS that don’t belong there
- Building unnecessary projects
- Overprovisioning existing infrastructure
You’re just making the problem worse. You’re wasting resources, wasting engineering time, and setting yourself up for a worse negotiating position in the future—more on that last one in a bit.
Move Some of Your Vendors into Marketplace
With some restrictions, purchases you make on the AWS Marketplace count toward retiring your EDP spend commitment. For most agreements you’re been able to make a dollar-for-dollar match to retire up to 25% of your commitment. Professional services are excluded, and, earlier this year, it became a condition that the service you purchase must itself run entirely on AWS. Even so, it’s still possible to meaningfully chip away at your balance.
Renegotiate
You can go back to Amazon to renegotiate your agreement in some circumstances. If they’re up for it, here’s what that looks like:
- AWS will offer to create a new multi-year agreement for you, starting now
- This new contract (typically 3–5 years) supersedes your remaining commitment period
- New commitment levels are based on your current actual spend (your “run rate”)
- Discount percentages will be lower than your original contract for two reasons:
- They’d never admit to it, but there’s an implicit penalty for missing your original commitments
- Lower spending/usage tiers qualify for smaller discounts
Do Nothing
Believe it or not, this sometimes is a legitimate intentional strategy based on some simple arithmetic.
Suppose you purchased season tickets for your favorite baseball team at a 50% discount off the individual game prices. Even if you miss five games out of 20, it still leaves you with a net saving compared to buying 15 individual tickets. You don’t need to go to all the games to come out ahead.
Cloud contracts work similarly. You can still come out ahead when paying shortfall fees thanks to this bit of your agreement: shortfall fees for service discounts are calculated at the discount rate.
Instead of baseball tickets:
- DTAZ is $0.01/GB, but you commit to 500 GB at $0.005/GB, a 50% discount off retail
- You consume only 350 GB
- Your shortfall fee is 150GB × $0.005/GB = $0.75
All told, you’ve spent $2.50 (350 × $0.005 + $0.75) instead of $3.50, an effective discount of 29%. Yeah, it’s not the 50% discount you signed up for, but it’s still not bad.
Look at the entire commitment timeline and assess your complete outcome, not just the problem periods. If you’re in year 2 and missing the commitment, but project growth in years 3–5, the total value over the contract’s life might still be positive.
In fact, we have some clients who intentionally make aggressive commitments knowing they’ll miss them sometimes but the discount will more than make up for it. By committing to a higher usage than they actually expect, they can lock in deeper discounts on the actual usage. (Caveat: this is advanced mode–don’t even think about it unless you’e spending $100m+/yr with major growth ahead!)
Your Decision Tree
Here’s a simple decision tree we encourage our clients to follow:
10–15% Underwater: Do Nothing
Since your effective discount rate will be cheaper than the retail rate—even if you left a bit on the table—this is actually a win. Seriously, unless you have a crystal ball to do perfect forecasting, this is probably the best cost outcome. Focus on other stuff.
20–30% Underwater: Do Nothing (Probably)
If you go hat-in-hand to AWS for this level of overcommitment, the results will depend on how the tea leaves were read that morning. You might be able to kick off a re-negotiation, but you also could be told to pound sand.
Honestly, the shortfall is likely not enough to make renegotiation worth it. Suck it up, re-evaluate the inputs that led to your initial decision, and chalk up the moderate loss to the price of learning.
50% Underwater: Renegotiate
If you’re this far underwater, this is where I’d recommend renegotiating your contract. You’re now paying too much in shortfall fees.
The first step might feel counterintuitive: go deeper underwater by cutting costs further.
It’s time to get aggressive about optimizing costs. Aggressively optimizing costs before renegotiation lowers your baseline for the new contract. AWS will still offer worse discount terms, but your new lower baseline means you’ll come out ahead in total dollars saved.
Suppose your original commitment was $50 million/year and you’re spending only $25 million. A renegotiation would put your new baseline at $25 million/year plus annual growth. But since the best savings is the money you don’t spend, it’s worth your time to reduce your spend to $15 million through optimization efforts, first, to set an even lower baseline. It’s true that your new discounts will be way less interesting than your previous ones, but you’re spending less money overall.
Don’t Go It Alone
If you wouldn’t give yourself a haircut, you shouldn’t take this on by yourself. The world’s best procurement managers, CFOs, and finance teams rely on AWS experts to model their forecasts, understand how their architecture impacts their bill, and distinguish good contracts from bad—and so can you.
Take a breath, acknowledge that the universe continues to expand, and get in touch.
