Robot Active Use Metrics

Robot active use metrics are Xu Huazhe’s proposed discipline for judging embodied-AI companies by actual repeated use rather than by shipments, production volume, or performance videos. In 166: 许华哲再次具身创业：不想错过最大的西瓜, he argues that robot sales can be misleading if buyers do not keep the machines active in real tasks.

The concept extends AI Consumer Growth Metrics into physical products. A robot can be sold once, rented once, or shown in a demo, but the harder signal is whether it becomes part of a household or workplace routine often enough to justify cost, maintenance, safety risk, and data collection.

Key Claims

• Active rate and repeated daily use are stronger evidence than shipment count when demand is uncertain.
• A dance or stunt demo can prove hardware control without proving useful intelligence.
• Robot usage metrics also shape the data flywheel: inactive robots do not generate enough diverse, corrective, task-grounded data.
• For household robots, active use has to be interpreted with safety, trust, maintenance burden, and service value rather than raw screen-time-style engagement.

Connections

• AI Consumer Growth Metrics — consumer AI retention analogue.
• Product Led Willingness To Pay and Customer Pull — demand signals behind active use.
• Household Robot Data Flywheel and Physical World Data Flywheel — data loops that need real usage.
• Home Service Robots and Poke Robotics — category and company context.
• AI Commercialization Pressure — pressure to convert robot capability into durable adoption.