Amazon shifts AI metrics focus to tangible results
In brief
- Amazon drops indicators based solely on token consumption for AI.
- An internal memo advises prioritizing customer impact and quality over quantity.
- Teams are urged to document cases where AI provides no value.
Amazon acknowledges the flaws of its internal AI usage metrics, often reduced to token consumption. According to Next, the tech giant now prioritizes concrete outcomes over quantitative indicators.
An internal memo from September 2025, signed by Terk Terkowitz, engineering vice president, outlines new practices. Amazon retains some metrics like token consumption or code modifications but now treats them as measurement tools rather than targets. The company emphasizes AI adoption, workload volume, and customer impact while recommending analysis of both the top and bottom 10% of teams to identify barriers or best practices.
The text highlights that AI must serve real efficiency rather than internal rankings. Amazon encourages engineers to document cases where AI adds no value, while stressing long-term impact on customers and business operations.
What it changes for you
- Human ResourcesHR must rethink evaluations of technical teams by incorporating qualitative criteria tied to tangible outcomes rather than AI usage metrics.
This summary gives you the essentials. The full article, its figures and quotes are on Next.
Read the original articleAI-generated summary of the source, checked by rules. How we write · Report an error