eMarketer confirmed it this week. Meta hit $243.5 billion in ad revenue in 2026, overtaking Google at $239.5 billion. It is the first time in history that Meta has come out on top. Meta is growing at 24% annually. Google is at 12%. If your media plan still treats Meta as secondary, it is already behind.
Attention follows people, not platforms
Celeste Ntuli won Headliner of the Year last Saturday. She did not build that through media budgets. She showed up consistently, in her own voice, until the room had no choice but to notice. That is the same dynamic driving Meta's numbers. Personal profiles drive eight times more engagement than company pages. Individual voices beat institutional ones, every time.
The shift is not really about Meta versus Google. It is about where attention has migrated and why. People are spending time with creators, with communities, with content that feels like it comes from a person rather than a brand. Meta, through Reels, AI creative, and Advantage Plus, has built the infrastructure to monetise that attention at scale.
The media plan that worked in 2023 does not win in 2026
AI creative on Meta is now table stakes, not an edge. What separates winners is clean data and a sharp brief going in. Every year someone defends a Google-heavy media split in the planning meeting. In 2026 that conversation needs to be harder. The attention data is pointing somewhere else. The brands I respect are rethinking the split before their results force them to.
Cultural authority is built through consistency. Not campaigns. Not budgets. Celeste Ntuli is the proof. The media plan that won three years ago was built for a different attention landscape. Optimising the wrong split faster is not progress. It is just more expensive.
The question that matters
When last did you challenge your media mix rather than just optimise it? There is a difference between improving execution on a flawed strategy and questioning whether the strategy still fits the market. The numbers this week suggest it is time to ask the second question.
