AI is creating a new version of a very old problem: measuring the wrong thing.

Lately, we’ve seen more organizations track tokens, prompts, and “AI usage” as proof that they’re making progress.

We’ve been here before. When you measure activity, you get more activity, not better results. 

At Indeed, we treat AI as a core business capability. We evaluate it the same way we evaluate any meaningful investment: by outcomes that deliver on our mission to help people get jobs.

When Activity Becomes the Goal

As AI accelerates, it’s compressing the execution layers of work. Code, content, and analysis are increasingly handled by machines, shifting where human value shows up.

Human contribution is moving up the stack: judgment, prioritization, identifying the signal through the noise, and integration that matters to customers and the business.

Usage alone doesn’t tell you whether work is getting better, faster, or more impactful.

That’s why the most important question isn’t whether people are using AI or whether token counts are going up.

What matters is whether teams are shipping faster, improving quality, and most importantly, delivering better results. For example, we measure how AI coding assistants save Indeed engineers an average of more than four hours per week, and how AI tools help increase overall engineering productivity by more than 50%. With more time and efficiency, teams can move faster and focus on strategic challenges.

The real test is whether AI is helping solve meaningful problems — creating better and quicker matches, and more effective outcomes for the business.

If AI isn’t moving a real metric, it’s not productivity. It’s just expensive electricity.

Focus on Outcomes, Not Proxies

We’ve seen firsthand why proxy metrics fall short. Like many companies, Indeed briefly tracked the percentage of code written with AI. After a few months, it became clear this wasn’t reshaping productivity or improving outcomes. So, we moved on – fail fast, learn faster.

This isn’t to say adoption metrics are useless. Adoption is often a necessary precursor to value. You have to track usage early on to see if a tool is gaining traction. But there is a fine line between using adoption as a diagnostic tool and using it as a performance target. Any time a metric like usage or token counts becomes part of an incentive system, you create a perverse incentive. People start chasing what’s easy to measure rather than what’s meaningful to improve. They stop optimizing for the mission and start optimizing for the metric.

We now focus on how quickly we can go from an idea to a tool in the hands of job seekers and employers. We anchor our AI investments on real workflows and high‑friction problems, tying each use case directly to clear business outcomes like faster time to hire.

One example comes from our customer service and sales teams. Our AI chat agents are resolving 28% of cases autonomously while escalating the rest to our teams. This clears the way for employers to spend less time waiting for support to ultimately get jobs posted and hire faster. Internally, we’re also implementing agentic AI solutions to help automate the ‘work about work’–meeting prep, account research, and planning–so sales teams can trade administrative toil for meaningful customer strategy. 

This shift not only helps our sales teams reclaim multiple hours every week, but it’s tightened our feedback loop with customers, leading to shorter hiring cycles and ultimately better hiring outcomes. 

This has also changed the role of leaders. Their job isn’t to maximize usage. It’s to decide what should be automated, what should be augmented, and what should remain human‑driven.

We still look at usage patterns to understand what’s happening beneath the surface. When a team’s token usage spikes, the question isn’t “Why are they using more?” It’s “What are they building and is it working?”

Transparency Matters. Outcomes Matter More.

We see the impact of these decisions at scale at Indeed. When teams measure the wrong thing, the cost isn’t abstract.

By focusing on outcomes, we ensure AI is making a real difference in how Indeed operates and serves our customers, partners and employees – staying true to our mission to help people get jobs.