Reported execution
dentsu says it split an equal budget between auction buying and Reach and Frequency in high-spend locations. DMC flagged underperforming interests, recommended replacement segments and suggested bids across Facebook, Instagram and Messenger.
Our reading
The useful AI claim here is bounded: it is a decision aid inside media buying, not a claim that an algorithm made the campaign. A controlled comparison and named human choices make this more legible than a generic optimisation boast.
Evidence limits
The agency reports a 13% CPM improvement while maintaining its reach KPI, but does not publish the dates, raw delivery data, model features or statistical treatment. The public source therefore supports the test design and claimed direction, not an independent performance conclusion.
Sources & limits
dentsu Indonesia case study; its outcome figures and the optimiser’s internal method are agency-reported.
- dentsu Indonesia — AI Driven Optimisation for Nestlé Indonesia — source publication date not stated.
Send a correction with the passage and supporting source.