Calculating the return on investment (ROI) of artificial intelligence (AI) requires allocating costs and aligning them to business outcomes, producing a Cost per Outcome that is compared against delivered value. AI deployments are classified as either external or internal. External AI maps directly to revenue-producing activities, with costs becoming part of the cost of goods sold. Internal AI supports developer productivity or broader employee efficiency, and attributing ROI is more difficult because these deployments often do not tie directly to revenue-generating activities. Within these use cases, structured interactions involve agents with defined goals, specific KPIs, and predictable execution patterns, making them easier to allocate, while unstructured interactions cover ad-hoc, general-purpose uses like coding assistants or chat interfaces, which may require more granular allocation. Increasing AI costs are an important catalyst for determining ROI, and cost can sometimes be misattributed and understated. The first factor in determining cost is to verify that the Total Cost of Ownership (TCO) has been calculated.