Natural Intelligence moved the channel from test to growth budget after sending conversion events back to the algorithm; Adthena's 30-brand Q3 pull says 52% still cannot recoup spend, and Kaiser/Schulze put US revenue per session at $0.89 vs $2.45 on Google Search. Twilio is dabbling, Intuit is experimenting, and Qualcomm's CMO says targeting barely exists unless the ask is super broad.
Why it matters: Keep ChatGPT Ads in a tagged test cell until conversion events are firing; do not copy a growth-budget move on click metrics.
{searchtermmatchtype} now returns e/p/b/a, and a is the AI Max expansion and keywordless match. Pair it with {matchtype} so a click that looks exact in the keyword column but a in search-term match is expansion traffic you can score in analytics, not a blended number inside Ads. Spotted by Hana Kobzová.
Why it matters: Add the parameter to every Search and AI Max tracking template this week and split landing-page conversion by a vs not-a before the next AI Max budget argument.
Global CCO Neil Reynolds stood up One Demand squads (brand, insights, KAM, category, agency) and is buying LLM visibility data so M&M's and Dove show up when a shopping agent builds the weekly run. Publicis now owns connected commerce, and the next retailer meeting is data-team to Walmart data-team.
Why it matters: CPG and retail teams need an LLM-source audit and a named commerce owner before the agent starts substituting the impulse SKU.
The AI-Era Attribution Blueprint, out this morning, splits AI-influenced (human still buys) from AI-recommended (the model chooses), then asks three questions at conversion: what created intent, who decided, who executed. A claim is eligibility plus evidence (referrals, seller records, surveys), not a 30% default.
Why it matters: Stop taking platform-reported ChatGPT or AI Mode conversions at face value; score each as influenced vs recommended and demand the evidence trail.
Supermetrics Campaign Manager, live Nov 10, writes to Google, Meta, LinkedIn, Microsoft, TikTok, Snap, and ChatGPT Ads from Claude or ChatGPT with per-account write switches, budget caps, and a change log. Premier Cru's Matthew Thompson used it on Google and Meta and is not taking more clients with the time; he is using it for answers the account never had.
Why it matters: If the team already chats performance in Claude, this is the write path to test with paused creates and a hard budget cap, not a new dashboard.