Udam Dewaraja – Founder & CEO of StitcherAI. Founded FOCUS, a finance standard adopted by thousands of enterprises.

​Leaders often ask me: What does this product actually cost us, and is it earning its margin? It’s harder to answer than it should be.

Take one feature I looked at recently. It runs on cloud infrastructure, calls two AI providers, sits on a SaaS platform with its own contract and relies on a shared internal service half the company also uses. That’s five sources of cost behind one feature, and it’s a modest example. The answer leadership wants is spread across those five bills.

Five Bills, Five Languages

The cloud provider charges by the second and uses its own names for everything. The AI providers bill by tokens and API calls. The SaaS platform charges for seats and credits under its own contract, while the internal service lands in the general ledger (GL) as a lump sum with no record of what used it. Before anyone can say what the feature costs or calculate ROI, someone must translate it all and often rebuild much of it.

Almost every organization I’ve worked with does this same work, provider by provider, and none of them gains anything by doing it better than the company down the street. It’s undifferentiated work that eats the time meant for figuring out what the numbers mean.

It doesn’t matter much whether you buy a tool or build in-house. Build in-house, and your team translates every bill into an ad hoc internal model that only makes sense inside your company. Buy a tool, and the vendor translates it into its own predefined model, which you then have to learn and work around. Either way, each company ends up with its own private version of what cost means. The FinOps community built FOCUS (FinOps Open Cost and Usage Specification)—which I founded and chaired in 2023 and 2024—to solve exactly this problem.

Cost Data The Whole Company Can Read

FOCUS is a common language for consumption-based billing data. It gives data from any provider the same structure and definitions. An accrual-based cost column means the same thing whether it came from a cloud provider, an AI vendor or a SaaS platform. Convert all five of that feature’s bills to FOCUS, and they land in one table where cost means one thing. On cloud cost alone, the feature looked cheap to run. Add the AI providers, the SaaS platform and its share of the internal service on the same basis, and that healthy-looking margin may come under pressure.

It’s easy to assume the FinOps community built FOCUS to make its own job easier. That was part of it, but the bigger reason was everyone else. People make spending decisions all over the company: An engineer picks an instance type, and a product manager switches to a bigger model. If only a specialist team can read the cost data, accountability stays with that team, and finance ends up explaining numbers to the people who made them.

The point was cost data the rest of the organization could read. When the product team reads the same data as finance, in the same terms, the margin discussion stops being a report someone hands them and becomes something they own. FOCUS doesn’t get you there on its own, but it puts everyone on the same numbers, and that’s where ownership starts.

Agents Have The Same Problem

Agents and coding assistants now make some of the most important cost decisions. An agent that routes each request to a model or an assistant that helps an engineer design a new service needs cost intelligence to make that call well. Hand it a hodge-podge of billing exports, and it must guess what “cost” means in each one and how they fit together. You’ll get decisions that sound confident but are wrong.

Give it one format where every column has an agreed meaning, down to the difference between what the provider billed you and what something effectively cost you, and it can weigh those choices on the same numbers your finance team uses. If you want agents making these calls, the shape of the cost data they read matters as much as the model doing the reasoning.

Keeping FOCUS Focused

That readability is also what’s worth protecting as FOCUS evolves. Some of the largest providers in the world, including Amazon, Google, Microsoft and Oracle, adopted the early versions right away because the spec solved a problem the market wanted solved.

As more organizations use FOCUS, requests to extend it will keep growing, which is a good sign. Every change balances how broadly it helps against the complexity and dependencies it adds. Complexity and dependencies slow adoption. They also make the spec harder for people outside FinOps to read, and readability is the whole reason it’s useful to them. A change that helps most of the people using FOCUS is usually worth it. One that helps a few while making adoption harder for everyone else usually isn’t. Holding that balance together as a community is how FOCUS stays something the whole organization can use.

For anyone consuming it, the advice is simple: Don’t wait for the spec to be perfect. FOCUS doesn’t need to cover every corner of your spend before it’s useful. Start with your highest-priority providers or the ones whose data is already available, build your cost model and you’ll see a big part of your spend clearly for the first time. Then, add the rest and enrich the model as you go. Perfect is the enemy of good here.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?

Share.
Leave A Reply

Exit mobile version