MarTech Featured · October 7, 2026
The platforms now want your goal, your creative and your data, and very little else. As AI Max, Advantage+ and ChatGPT Ads take over the levers, the performance marketers who win will be the ones who control inputs and measurement, not bids.
The last manual controls are leaving
For a decade, performance marketing meant steering: keywords, audiences, bid caps, placements. This autumn the steering wheel is being removed in plain sight.
Google began automatically upgrading Search campaigns that use campaign-level broad match or standalone automatically created assets to AI Max on September 1, with Dynamic Search Ads next. Microsoft stopped offering Max CPC on new non-portfolio campaigns using automated bid strategies on October 1. And since August 17, Google has been steering budget-limited Target CPA and Target ROAS campaigns toward the target you entered rather than the better number the system had been hitting. Google's own example: a campaign with a $10 target CPA converting at $5 will drift toward $10.
The implication is uncomfortable. The target you typed 18 months ago is now the instruction the platform actually follows.
"Agentic" is real, but uneven
Google's AI Max and Meta's Advantage+ now run targeting, bidding, creative assembly and budget pacing with steadily less human input. Meta's own figures put the average CPA improvement for Advantage+ sales campaigns at 20% over standard campaigns, and a 10% reduction in cost per qualified lead for lead generation. Google cites its own AI Max lifts too.
Treat those numbers as marketing, not audits. Industry analysts note that most headline lifts come from the vendors' own pages, and that eMarketer frames 2026 as the beginning of the end of manual programmatic buying, not the arrival of fully autonomous buying. A lot of what is sold as "agentic" is still an ad maker or an optimisation layer rather than a closed-loop agent.
Account structure has clearly changed. Practitioners report that many D2C accounts have collapsed from dozens of fragmented ad sets into a handful of consolidated Advantage+ campaigns, fed by strong first-party data and judged on blended, incremental results rather than Meta-reported ROAS.
Creative is the new targeting
When the algorithm picks the audience, the ad itself becomes the targeting signal. Generative AI is making it cheap to produce hundreds of variations of a single concept for different micro-segments, and delivery engines reward volume and diversity. Marketing Brew reported in April that one agency client needed 1,000 creative assets across personas and concepts just to feed Meta's delivery engine.
The discipline shifts from audience building to creative systems: briefs, brand guardrails, fast testing loops, and a way to learn which concepts actually drive incremental revenue rather than cheap clicks.
A new paid channel: ChatGPT
OpenAI says ChatGPT Ads reached a $1 billion annualised revenue run rate in under 200 days, opened self-serve buying across Europe, India, the Middle East and North Africa on August 31, and on September 16 announced sponsored agents, AI ad-building tools and HubSpot and Shopify integrations (available first in the US).
Advertiser results are mixed. MediaPost reports some advertisers see CPCs under $3 while others pay $10 to $13, and some campaigns generate little activity after the first click. OpenAI's run rate measures OpenAI's business, not yours. Test it like any new channel: a fixed budget, a fixed window, and a cost-per-qualified-lead range agreed in advance.
Elsewhere, community platforms are maturing as performance channels. iQuanti's Q2 2026 report highlights Reddit, where Reddit Max automates targeting and creative selection and performance advertising now makes up most of its ad revenue, and Snapchat, which is improving conversion efficiency through Dynamic Product Ads and AI-based intent matching.
The measurement gap is the real story
As automation grows, the platforms show less. Google's generative AI performance report reached Search Console worldwide on August 31, showing how often pages appear in AI Overviews, AI Mode and Discover's generative features, but with no click data. Search Engine Journal's September survey of search professionals found 39% saw traffic fall or stay flat while leads and conversions held or improved, and only 15% lost both. Sessions are no longer a safe proxy for business.
The trust question is also now legal. On August 31 the FTC and 22 states sued Amazon, alleging an "invented auction participant" inflated what advertisers paid across Sponsored Products, Sponsored Brands and Sponsored Display. Amazon calls the claims patently false, and these remain allegations. Whatever the outcome, the lesson stands: every auction you pay into reports its own results.
What to do this quarter
Audit every tCPA and tROAS target against your own economics. Customer value and margin should set targets, not history.
Reconcile platform conversions against closed revenue in your CRM. Know how far apart the two numbers really are.
Save your first Search Console AI impressions report now. It is your baseline for every later comparison.
Build a creative engine. Plan for far more concepts and variants than your team produces today, with clear guardrails.
Run a disciplined ChatGPT Ads test, with success and stop criteria written before spend begins.
Measure incrementally. Use holdouts, geo tests and blended metrics, and treat platform ROAS as supporting evidence.
The bottom line
The marketer's job is moving upstream: from pulling levers to supplying the machine with the right goals, first-party data and creative, and then checking its work independently. Automation is not the risk. Trusting only the automation's own scorecard is.