Reported execution
dentsu says the partners tested AI Max for eight weeks on selected generic categories, then expanded after monitored performance and query coverage results.
Our reading
This is media optimisation, not generative creative. The useful practice is the staged test: human teams retained targets and budget controls while automation explored additional intent.
Evidence limits
Agency-reported ROAS and incremental-growth figures need independent measurement design to establish causality. The public case does not expose queries, controls or the full attribution method.
Sources & limits
dentsu India case study; ROAS and incrementality figures are agency/client-reported.
- dentsu India — Sokrati for Myntra — source publication date not stated.
Send a correction with the passage and supporting source.