Paid media / Practical guide

Google Ads AI Max migration: your account checklist

Check Google Ads AI Max migration settings, generated copy and lead quality. Understand which campaigns change in September and which follow later.

A control console routes search paths through a review gate towards a website panel. Conceptual editorial illustration.
A control console routes search paths through a review gate towards a website panel. Conceptual editorial illustration.

The immediate job is an account check, not a wholesale rebuild. A familiar campaign name can now sit above a different settings bundle. Conversely, a performance change during September is not proof that AI Max caused it.

Reviewed 20 September 2026. This guide covers the September rollout documented in Google's 12 August announcement. It is a source-based operational checklist, not a D4J performance experiment.

Which campaigns are affected this month?

Google's migration announcement identifies Search campaigns using campaign-level broad match or standalone Automatically Created Assets, often shortened to ACA. Their automatic upgrade runs progressively from 1 to 30 September.

Campaign-level broad match is a particular setting. Merely finding a broad-match keyword in an account does not establish that this migration applies. Check the campaign settings and migration information rather than inferring eligibility from keyword match types.

Dynamic Search Ads follow a separate timetable. Google says active legacy DSA campaigns are excluded from September's automatic migrations, with their automigration scheduled for February 2027. September notices may encourage voluntary upgrades. A notice about a future change is not evidence that the change has already happened.

What settings does Google switch on?

The documented defaults differ according to the legacy configuration:

Previous configuration Search term matching Text customisation Final URL expansion
Campaign-level broad match On Off Off
Standalone Automatically Created Assets On On Off

Google also says existing brand inclusions and exclusions are preserved for the campaign-level broad-match migration. These are migration defaults, not a substitute for inspecting the current account. An advertiser or scheduled integration may subsequently change them.

Search term matching affects the searches the campaign can reach. Text customisation affects automatically produced messaging. Final URL expansion affects destination selection. Treating all three as one vague “AI setting” makes investigation unnecessarily difficult.

Record a baseline before touching the budget

Capture the campaign ID, settings, migration date where available, conversion goals and recent change history. Keep campaign IDs alongside names because names can change. Record which person or integration is authorised to edit the account.

Export comparable periods for cost, enquiries and qualified outcomes. Account for conversion lag, weekdays, seasonality and changes to the offer. A quiet weekend should not be compared with a busy working week and labelled an automation failure.

Then write down the commercial boundary. A Pretoria business serving commercial premises, for example, may not want residential enquiries or work outside its service area. This is an illustrative scenario, not a claimed client result. The boundary gives the reviewer something more useful than “traffic looks different” to test.

If the account counts every contact-button click as a valuable conversion, resolve that measurement weakness before drawing confident conclusions from a lower reported cost per conversion. Our qualified-enquiry guide explains that distinction.

Inspect searches, copy and destinations separately

Searches: review search terms against the services you actually sell. Look for irrelevant needs, unsupported locations and requests your team cannot fulfil. Consider negative keywords and available brand controls in context; do not automatically block every unfamiliar phrase. An unfamiliar query may still express a useful need.

Copy: where text customisation is active, examine generated assets for unsupported prices, service promises, superlatives and exclusions that have disappeared. An ad being generated by Google does not make its claims accurate for your business. Keep evidence for claims and identify who approves corrective action.

Destinations: confirm whether Final URL expansion is active. It is off in the migration defaults above, so do not blame every landing-page change on automatic expansion. If it has been enabled, review the selected pages, applicable exclusions and whether each destination supports the advertised action.

A relevant page still needs a usable enquiry route. Check phone links, form validation and the handover to the team without creating real test enquiries in a live customer pipeline.

Decide whether to retain, adjust or investigate

Use a short decision record rather than a hurried reaction:

  • Retain and monitor: settings are understood, searches fit the offer and there is no established deterioration in meaningful outcomes.
  • Adjust a specific control: an identifiable mismatch exists, such as inaccurate generated copy or a clearly irrelevant query group.
  • Investigate before scaling: tracking, lead qualification or simultaneous changes prevent a reliable interpretation.

Do not change bids, budget, landing pages and conversion definitions together unless a genuine operational risk requires it. Multiple simultaneous changes make the next review harder to interpret. Record any necessary emergency intervention and its reason.

Choose a review interval that allows enough outcomes for the account's volume and sales cycle. There is no universal number of days that makes a low-volume campaign statistically decisive.

Questions to take to your account manager

Ask which campaigns migrated, what evidence identifies them, which features are enabled now and what the team has checked since migration. Ask whether campaign-management tools or imports can overwrite those settings. Finally, ask how qualified leads, rejected enquiries and sales are fed back into the evaluation.

A useful answer should identify the account evidence and the next decision. “Google recommends it” and “AI never works” are equally poor substitutes.

For help defining that review, see our Google Ads management and account-review scope. If you are deciding who should own the work, the management, training or review guide explains the different responsibilities. Bring the campaign context and the business outcome you need to the existing Discuss project form.

Sources & review boundaries

Source links checked 20 September 2026. Platform settings and policies can change. Verify the current documentation before implementation.

Examples and checklists are practical guidance, not guarantees, legal advice or a replacement for account-specific testing. Linked D4J case studies carry their own evidence cut-offs and limitations.

Worth passing on?

Send the guide to the person making the decision.

Email

A useful conversation

Have a question or a correction?

We keep discussion focused and reviewed. Send us the article link and what you would like to clarify.

Talk to Digital 4 Jesus