Tokenomics Needs a Taxonomy

Measuring AI Spend Without a Taxonomy

The Tokenomics Foundation launched just under five months ago with the mission of creating a shared discipline for managing AI and related token costs.

Layered Attribution Model

The source outlines a three-layer approach. The first layer categorizes AI costs (LLM usage, cloud services, on-prem capital, labor, etc.). The second layer attributes those costs to defined AI capabilities to determine the total cost of each agent. The third layer allocates each agent’s costs to defined business outcomes, using agreed-upon formulas for items that are not easily taggable such as on-prem storage or indirect labor.

Lessons From Technology Business Management

Technology Business Management (TBM) translates technical resources into application-level and product-level cost insights rather than linking servers directly to revenue. The author notes that AI may need a similar connective tissue, observing that TBM models are powerful but difficult to implement and maintain because they require assumptions and governance across multiple layers.

Immediate Steps for Organizations

The source advises that organizations cannot wait for a universal taxonomy to emerge. It recommends creating an interim taxonomy by:

1. Defining the outcome and its measurement unit.
2. Mapping AI capabilities to those outcomes and establishing allocation drivers where capabilities are shared.
3. Building the total cost of each capability by categorizing AI spend and attributing costs to agents.
4. Governing the assumptions through documentation, ownership assignment, and stakeholder agreement.

Starting with a single use case can demonstrate the methodology and build broader buy-in before expanding the model.