Indeed's leadership is rejecting AI token tracking leaderboards, arguing that measuring activity over outcomes creates perverse incentives that destroy real productivity.
Companies are spending aggressively on AI, and naturally, executives want to know what they are getting for their money. The instinct is to track usage, build dashboards, and rank employees on how many AI tokens they consume. Indeed's chief information officer, Anthony Moisant, thinks that is exactly the wrong approach, and he is willing to explain why in detail.
As Business Insider recently reported, Moisant confirmed that Indeed actively tracks token consumption but keeps that data entirely in the background. The company will not build a leaderboard, and it will not turn AI adoption into an internal competition. The reasoning is straightforward: people optimize for whatever you measure. If you measure volume, you get volume. You do not necessarily get better work.
This is not an abstract concern. The tech industry has developed a peculiar fixation on quantifying AI engagement. Nvidia CEO Jensen Huang recently suggested that a well-compensated engineer should be consuming roughly half their salary in AI token value, a striking benchmark that reflects how deeply the hardware layer wants to normalize massive AI spending. Companies like Meta have reportedly implemented internal leaderboards to encourage employees to use more tokens, turning adoption into a visible, gamified pursuit. Moisant acknowledges there is nothing inherently wrong with this approach, but it does not align with how Indeed wants to operate.
Indeed has already run this experiment and watched it underperform. When generative AI coding tools like Claude and Cursor started gaining traction, Indeed began measuring the percentage of code written by AI. After three or four months, the metric revealed its own limitations. It was a proxy for adoption, not a measure of impact. Shipping speed did not improve proportionally. Customer satisfaction did not climb. The number looked promising on a slide, but the underlying outcomes remained flat.
That experience shaped Indeed's current philosophy. Moisant's team now focuses on two things: how quickly products ship and how customers respond. These are harder to measure than token counts, but they reflect actual business value rather than internal activity.
The Cost Question Looms Larger
This debate is unfolding against a stark financial backdrop. Indeed's AI bills are projected to quadruple compared to 2025, driven primarily by research and development usage. Moisant told his board that aggressive cost controls in this area could undermine the productivity gains the company is chasing. The board's response was nuanced: spend what is necessary, but maintain visibility into where the money goes and what it produces.
That balance is where most enterprise AI strategies will live or die. Companies across sectors are facing similar cost pressures as AI budgets scale faster than revenue. The temptation to justify spending with easily quantifiable metrics like token volume is understandable. It gives executives a number to present, a chart to show, a sense of control. But it risks creating what behavioral economists call a proxy trap, where the measurement becomes a substitute for the thing it was supposed to represent.
Indeed's position is particularly interesting given its broader mission. The company has spent the last year pushing a "Humanizing Hiring" initiative, explicitly moving away from black-box algorithms and gamified ranking systems for job seekers. Applying the same logic internally, refusing to rank engineers on token consumption, is consistent with that external message. If Indeed believes leaderboards create toxic competition for candidates, it follows that the company would avoid them for employees as well.
The practical takeaway for other companies navigating AI adoption is worth considering. Before building a dashboard that ranks teams on usage, ask what behavior that dashboard incentivizes. If the answer is more usage rather than better outcomes, the metric may be costing more than it reveals. The companies that figure out how to measure AI's impact on actual product quality and customer experience, rather than raw consumption, will be the ones that justify these ballooning budgets over the long term.
