Your experience with the CI/CD pipeline is a perfect example of the operational reality clashing with the sales model. We saw something similar, but with autoscaling container workloads. Each new pod identity was counted, turning a predictable cost into a variable one tied directly to our scale-up events.
The sales-to-ops disconnect you mention is the critical failure. You can have clauses in the contract, but if the engineering team's definition of "entity" in their collection logic drifts, you're stuck reconciling it after the fact. Our solution was to mandate a billing audit API as part of the deal, so we could see the raw count in real time. Even then, it was a constant source of friction.
—Anita
That final point about governance assessment is exactly where we landed. We started asking product managers for their quarterly roadmap and then cross-referencing each feature against the billing metrics it might affect. It's tedious work, but it's the only way to see the potential drift before it hits.
The committed directory object count model has its own problems, but at least it's a stable target. With consumption based pricing, you're not just managing the tool, you're auditing their product team's priorities. It turns procurement into an ongoing operational task.
Yeah, that sales-to-ops disconnect you mentioned is such a common pain point, and it's frustratingly hard to guard against. You can have all the contract language in the world, but if the engineering team pushing updates has a different internal dictionary for what an "entity" is, you're left holding the bag.
Your CI/CD example is brutal but really illuminating. It turns a tool meant to provide clarity and control into a source of constant financial uncertainty. That's the opposite of what anyone wants from an IAM platform.
It makes you wonder if the real due diligence is less about the feature list and more about getting a product manager to commit, in writing, to a change management process for anything that touches billing logic. Even then, as you said, good luck.
Let's keep it real.
That idea of making change management for billing logic part of the contract is key. We tried something similar, but found the real challenge is enforcement. Even with a written process, if a vendor's engineering team pushes a hotfix that tweaks a collector, you might not get flagged until the next quarterly business review when the usage report lands.
It shifts the relationship from a technical evaluation to a continuous governance one, which is a heavier lift than most teams anticipate.
Stay grounded, stay skeptical.