NewsAgentic Marketing

Agentic Marketing | August 22, 2026

AI Max experiments no longer require dropping brand and location rails

You can now run AI Max tests with those controls left on, and from September a single A/B test can span multiple Search campaigns on budget and ROI targets. Performance Planner will forecast a bid or budget change and apply it in one click.

Why it matters: The excuse for not testing AI Max was the guardrail trade. That excuse is gone. Baseline the account before September's multi-campaign tests land.

Microsoft deletes the bid ceiling on new Max campaigns October 1

New standalone Max Conversions, Max Conversion Value and Max Clicks campaigns lose Max CPC. Existing campaigns and portfolio strategies keep it. Navah Hopkins is telling buyers to experiment the ceiling off before the date.

Why it matters: If an account needs a hard CPC cap after October 1, new work has to sit in a portfolio. Decide that now, not on the first October build.

ChatGPT Ads added three surfaces and left reporting at two

iOS App, Android App, and Web (desktop plus mobile browser) are now campaign-level inclusions. Insights still returns Mobile and Desktop, with mobile web folded into Mobile, so you cannot read the split you just bought.

Why it matters: Use the control to match the measurement stack you actually have, pixel versus AppsFlyer, not to report a platform mix the UI cannot show.

IAB killed the complementary-layers story on AAMP and AdCP

Shailley Singh listed 13 overlapping functions from briefs through serve-time decisioning and said major portions are substitutable. Shared MCP support is a pipe, not interoperability.

Why it matters: Dual-stack is now an explicit choice, not a hedge. Pick one protocol per client before the next agentic pilot.

LT.agency moved ad-ops hours into creative after a year on Fluency agents

A 40-hour local-ad refresh dropped 98%, and the agency hired three creatives in a quarter while the media team stayed flat. The 456% form-submission claim is one unnamed client and unaudited.

Why it matters: The usable receipt is the reallocation, not the lead-gen multiple. Deterministic execution is how agencies are answering the objection that the model should not spend.

Blanking meta descriptions did not move AI crawl or citations

Seer replaced descriptions on 21 high-traffic pages for six weeks and saw a flat -1.3% on AI crawl, with no citation penalty. Writesonic's six-assistant test scored meta descriptions zero. Title tags were the only tag the crawlers read.

Why it matters: Stop spending AEO cycles on meta tags. Put the extractable answer in the first 200 characters after the H1.

← All Agentic Marketing briefs RSS