Our Take
Token dashboards measure consumption, not value, which is why Disney's leaderboard immediately became a target for gaming rather than a signal of productivity.
Why it matters
As major enterprises (Meta, JPMorgan, Visa) deploy AI usage tracking, they're discovering that visible metrics drive behavior—often the wrong behavior. HR and ops teams need frameworks to distinguish real adoption from noise before incentives lock in the wrong habits.
Do this week
Finance and HR: Define what 'productive AI use' means for your business (time saved, output quality, cost avoided) before any dashboard launches, so managers can distinguish signal from gaming when usage spikes.
Disney's dashboard became a leaderboard, employees started gaming it
Disney gave nearly 5,000 product and tech employees access to an internal "AI Adoption Dashboard" tracking Claude and Cursor usage by request counts and token consumption, displayed in leaderboard format (per Business Insider). One employee invoked Claude about 460,000 times over nine work days in mid-April, amounting to roughly 51,000 calls per day. Automated agents tied to the same user racked up 234.2 million tokens in consumption. Disney leadership, including Andre Rohe (EVP of Product Engineering), publicly acknowledged the problem, urging faster AI adoption while cautioning against "tokenmaxxing"—gaming the metrics for leaderboard rank rather than business outcome.
Disney is not isolated. Meta ran an internal token-tracking dashboard showing over 60 trillion tokens consumed in 30 days before it was shut down over external sharing concerns (per reports). Visa reported 1.9 trillion tokens consumed monthly as of March and began rewarding high-usage teams. JPMorgan deployed dashboards to monitor adoption across the firm. The Linux Foundation announced the Tokenomics Foundation last week, a vendor-neutral governance body intended to support benchmarks and best practices for measuring token-based AI spending across enterprises.
Tokens measure consumption, not value
A dashboard worker using tokens to automate a time-consuming process appears identical to one running pointless queries that consume the same token budget. The metric is frictionless to track but decoupled from business outcome. Once usage becomes visible and ranked, employees optimize for the metric itself rather than the problem it was meant to solve.
This is the core tension enterprises face: token dashboards do encourage adoption and make spending visible. They do not prove productivity, quality, or return on investment. A leaderboard signals nothing about whether the work matters. HR leaders and operations teams can push adoption, but adoption without ROI is just spending acceleration.
Start with outcomes, not dashboards
Before launching any AI adoption program tied to usage metrics, clarify three things: what business outcomes the AI investment is meant to drive, how those outcomes will be measured independently of tokens, and who owns the connection between usage data and results.
Treat AI adoption programs like operating-model changes. Set clear guardrails for what counts as meaningful use. Train managers to distinguish signal from noise. Once visibility exists, the incentive to hit leaderboards exists too. Without explicit definitions of value, dashboards become scorecards for consumption theater, not transformation.