AI traffic gets its own channel — and search planning gets a zero

If any vendor is pitching an llms.txt or 'AEO schema' package, this is your out. Redirect that budget to fixing crawl, rendering, and the entity/Knowledge Graph work that actually shows up in AI Overview citations.

Decide your Google-Agent policy this month, especially if you operate gated content, e-commerce checkout, or accessibility-sensitive flows. Blocking it blindly will cut you out of agent-driven transactions; allowing it without rate limits will distort your analytics.

You don't have to agree with 'zero' to use it as a stress test. Run next year's revenue model with organic at –50% and –90% and see which programs survive. That's the conversation to have with finance now, not after AI Overviews coverage expands again.

If your team is still arguing whether AI traffic is worth optimizing for, this report makes the case. Tag the traffic correctly in GA4's new channel group and start reporting AI-assisted revenue as a separate line — it's small now but the trajectory makes it a board-deck metric by Q4.

Run your top 25 branded queries through AI Mode and AI Overviews this week and screenshot what surfaces. If negative review snippets are showing, the fix is review-platform hygiene and owned-content depth on the specific issues being cited, not a reputation-management retainer.

Keep shipping schema for the rich-result use cases it was built for, but stop selling it as an AI-citation lever. Reallocate that retainer time to entity coverage, internal linking, and getting cited by sources the LLMs already trust.

Pull a 90-day Search Terms export now as a baseline before the change fully rolls in, and re-audit your negative lists assuming you can no longer see the raw query. Brand-safety arguments to pause AI traffic just got a lot weaker since you can't fully inspect it.

Don't hand an agent your TikTok Ads Manager this quarter, but do have someone on the team stand up the MCP server in a sandbox account. The teams that understand the permission model and audit trail before clients ask will win the next round of RFPs.

Before adopting cross-campaign budget pacing, confirm your conversion imports and value rules are clean — these tools amplify whatever signal you give them, and PMax accounts with messy conversion setups are about to look much worse, much faster.

Lock in the new channel definition this week and annotate the change in GA4 — anything you don't annotate now will look like a referral collapse in your next QBR. Also rebuild any Looker Studio blends that hardcoded the old channel groupings.

MMM ownership is the right instinct, but the build cost is real — most marketers should start by demanding the model assumptions and code from their current vendor before greenlighting an internal version. Either way, MMM is replacing last-click in board reporting faster than most teams are ready for.

If you're a non-Publicis advertiser running clean-room or RampID-dependent activations, ask your account team in writing how data isolation will work post-close. Enterprise procurement teams should also revisit MSAs for data-use language.