OpenAI Adds Granular Usage Analytics to ChatGPT Admin Console

Analytics in the ChatGPT Admin Console integrate usage and cost data, outcome metrics, and task insights across both ChatGPT Work and Codex. [1] The Usage view provides visibility into active users, token usage, and credits to help administrators review costs and evaluate capacity requests. [1] Administrators can filter these metrics by user or group to identify areas of low adoption. [1] This visibility allows admins to review training needs and starting workflows directly with team owners. [1]

The Insights view features a task classifier that categorizes a sample of messages into specific use cases and tasks to reveal what work AI is supporting. [1] The Overview tab displays a summary of the work mix, while the Use cases tab provides a detailed breakdown of tasks. [1] For instance, software engineering tasks include code maintenance and feature development, whereas sales activities encompass account research and planning. [1] Administrators can apply group filters to examine team-specific AI usage and collaborate with business owners to assess workflows. [1]

The introduction of analytics in the ChatGPT Admin Console addresses a growing enterprise need to justify generative AI investments beyond basic seat counts. By categorizing prompts into specific functional areas like feature development or account research, the update shifts the focus from raw token consumption to concrete business tasks. This level of visibility within ChatGPT Work and Codex empowers organizations to directly target low-adoption groups and optimize training initiatives. As administrators transition from merely provisioning access to actively managing workflows, these granular insights will likely drive more strategic deployments across specialized teams. What the sources don't address: How the task classifier handles privacy or redacts sensitive information when sampling user messages to determine use cases.