Google Ads Automation Is Removing Manual Controls (2026)

Google Ads automation stopped being a setting you opt into. Over roughly six weeks, Google and Microsoft have started removing the manual levers advertisers used to hold, and the timing is not a coincidence. On October 1, 2026, Microsoft drops the Max CPC cap from new automated campaigns. Google is replacing the report you build by hand with a Gemini dashboard. Performance Max is folding local store targeting into a single toggle. Read together, these are one move. Control is leaving the account and moving into the model.

Microsoft removes the Max CPC cap on October 1

Start with the clearest change. From October 1, 2026, advertisers creating new campaigns with standalone Maximize Conversions, Maximize Conversion Value or Maximize Clicks can no longer set a Max CPC (Search Engine Land, September 2026). Portfolio bid strategies keep the cap. Target Impression Share and Enhanced CPC are untouched. Existing campaigns hold their limits for now.

Microsoft’s stated reason is the tell. It says a manual Max CPC can override your performance goal and create spend pacing irregularities. The API deadline is later and firmer. Max CPCs set through the APIs hold through the 2026 holiday season, which Microsoft dates to January 12, 2027. From that date the removal reaches API users and third-party tools building new campaigns, and Google Import stops supporting Max CPC on new campaigns too (Search Engine Land, September 2026). If your bid management writes a Max CPC into a new Microsoft campaign after that date, it fails silently, which is worse than failing loudly.

Google Ads AI Dashboards replace the report you build

The second change arrives dressed as convenience. In early September 2026, Gemini-powered Google Ads AI Dashboards began appearing in advertiser accounts (Search Engine Land, September 8, 2026). You type a plain-text prompt. Google builds the chart and picks the breakdown, then writes the summary explaining what moved. The report you used to assemble by hand is now something you describe and receive.

That is useful, and worth watching closely. The dashboard decides which cut of the data answers your question, and its summary decides which cause is worth naming. When the explanation is generated rather than derived, you lose the audit trail that told you whether to trust it. We traced the same drift toward agentic reporting in Google’s AI Mode search features: the tool stops being a surface you operate and becomes one you converse with. Convenient until the number it explains is one you needed to defend.

Performance Max absorbs local store optimization

Local Customer Optimization is a single campaign-level toggle in Performance Max store-goals campaigns. Flip it on, and Google prioritizes budget toward in-market shoppers near your locations, spread across Google Maps and Waze as well as local Search (Search Engine Land, September 2026). A companion feature, Store Sales in Data Manager, will let eligible advertisers pipe CRM or Sheets transactions straight into the account, and it is expected in the coming weeks. What was a set of location and channel decisions you made explicitly is now one switch.

Removing manual controls is the tradeoff builders should watch

Here is the part that reaches past ad accounts. Every one of these changes swaps a control you understand for automation you have to trust, and that trade only pays when the automation genuinely beats the hand you were playing. Microsoft says its AI Max testers gained 8% more conversions (ppc.land, 2026). Plausible. But the cap you are losing was also your guardrail against a bad model day, and averages hide the accounts where automation quietly overspent.

Anyone betting on AI-run systems should read this cluster as a preview. When a platform abstracts away a lever, it abstracts away your ability to diagnose why a number moved. The failure mode is not that the system breaks. It is that it works, opaquely, and you cannot tell a good outcome from a lucky one. That is the design question underneath every automated interface:

• Does the system expose enough state to explain its own decisions?
• When it is wrong, can you tell, and can you override it in time?

We build our own agents with that constraint in front, because a system you cannot inspect is a system you cannot defend to a client. The advertisers losing Max CPC this month are early. The developers wiring these platforms into their own products are next, and the same Google Ads API deprecations we mapped for the AI Max migration show how fast a removed control becomes a broken integration. If you are deciding how much control to cede, start by asking what you will need to prove when the number is questioned. We take the same view of measurement in demand generation for the AI search era.