Migration sequencing for AI Max in Search campaigns: controls, experiments and safe cutover
Short answer: Migrate to AI Max as a controlled Search-campaign change, not a one-click account-wide upgrade. Freeze the current campaign baseline, map which AI Max controls will change targeting or assets, test a bounded cohort or AI Max experiment, review search terms, generated assets and landing pages, then expand only when performance and control behavior are acceptable. Keep a rollback state for each campaign.
AI Max is an optimization layer, not a separate campaign type
Google documents AI Max as a suite of targeting and creative features activated inside existing Search campaigns. The main components include search term matching and asset optimization, with controls for text customization, Final URL Expansion, brands and URLs.
That distinction matters during migration. The campaign keeps its Search-campaign identity, but several mechanisms that determine matching, creative and destination behavior can change together.
A safe cutover therefore starts by identifying exactly which AI Max features are being enabled.
Phase 0: capture the campaign baseline
Before changing settings, export or record:
- campaign and ad-group structure;
- keyword and match-type configuration;
- negative keywords;
- brand controls;
- current RSA assets and pinned assets;
- landing pages;
- bidding strategy;
- conversion actions used for optimization;
- recent search-term report;
- budget state;
- current tracking parameters.
The goal is not to preserve the old setup forever. It is to make the migration auditable and reversible.
Phase 1: check bidding and budget dependencies
Google's AI Max documentation ties several features to conversion-based Smart Bidding and notes that budget-limited campaigns can constrain effectiveness.
Before testing, confirm that:
- the campaign's conversion actions reflect the business outcome the bidding system should optimize;
- conversion tracking is stable;
- campaign budget is not unintentionally limiting the test;
- offline or qualified-lead data is imported where the business relies on it;
- major tracking migrations are not occurring in the same window.
Do not interpret an AI Max test if the optimization target itself is unstable.
Phase 2: map the controls that change reach
Search term matching can expand beyond the campaign's current keywords through broad-match and keywordless mechanisms. Google also provides brand, location and ad-group-level controls.
Document:
- whether search term matching is enabled;
- brand inclusions/exclusions;
- geographic intent controls where used;
- negative keywords;
- any ad-group restrictions.
Then write the allowed audience/search boundary in plain language. If the team cannot explain what should and should not match, the migration is not ready.
Phase 3: map the controls that change creative and destinations
Asset optimization can introduce two major changes:
- text customization can generate additional ad copy from existing ads, landing pages and assets;
- Final URL Expansion can choose a different relevant URL from the domain.
Google notes that Final URL Expansion requires text customization and that pinned RSA assets may not be used when a different landing page is selected.
This means creative governance and landing-page governance must be reviewed together.
Verify:
- which URLs are allowed;
- which URLs must be excluded;
- whether URL inclusions are needed at ad-group level;
- whether all eligible pages are conversion-ready;
- whether generated text can safely reflect the offer;
- whether mandatory wording can be preserved.
Phase 4: use an experiment before broad cutover
Google provides AI Max experiments that split traffic and budget within an existing Search campaign between control and trial states.
For a migration pilot, define before launch:
- campaign selected;
- control and trial configuration;
- AI Max features included in treatment;
- observation window;
- primary business metric;
- guardrail metrics;
- reasons to stop early;
- changes that are prohibited during the test.
Do not redesign the landing page, change conversion definitions and alter campaign targeting at the same time unless those changes are part of the declared intervention.
Phase 5: inspect transparency reports, not only aggregate performance
A migration review should look below campaign-level totals.
Inspect:
- search terms and match-source information;
- headlines and URLs associated with search terms where reported;
- optimized asset reports;
- landing pages selected through Final URL Expansion;
- spend and conversions by meaningful segment;
- lead quality or downstream business outcome where available.
An acceptable aggregate CPA can hide unacceptable search or landing-page behavior.
Phase 6: apply, modify or roll back deliberately
If the experiment passes, apply the configuration intentionally rather than treating the end of the test as automatic approval.
If it does not pass, identify whether the issue came from:
- search term matching;
- asset generation;
- Final URL Expansion;
- brand or geographic controls;
- conversion tracking;
- budget state;
- creative or landing-page quality.
AI Max settings can be modified or disabled. Keep a record of the last approved state and the reason for any rollback.
Migration risks to monitor
Search expansion without business relevance
More eligible queries can increase activity without improving qualified outcomes. Review downstream quality.
Landing-page mismatch
A query-relevant page is not automatically the best conversion page for a regulated or highly qualified offer.
Generated asset drift
Text customization can create assets from page and campaign context. Review material claims and remove unacceptable generated assets.
Pinned-asset assumptions
Pinned RSA assets may not behave as expected when Final URL Expansion selects another destination. Campaigns with mandatory wording need special review.
Unstable attribution
If conversion definitions change during the test, the migration cannot be interpreted cleanly.
A practical cutover checklist
Move from pilot to normal use only when:
- baseline is stored;
- conversion objective is trusted;
- control settings are documented;
- experiment outcome is reviewed;
- search-term behavior is acceptable;
- generated assets pass policy/brand review;
- eligible URLs are approved;
- downstream lead or revenue quality is not materially degraded;
- rollback is documented.
The migration rule
AI Max should be treated as a controlled change to matching, creative and landing-page behavior.
The safest migration does not ask whether automation is good or bad in general. It asks which feature changed which decision, whether the result is acceptable for this campaign, and whether the team can still explain and reverse the state.
Sources reviewed
- https://support.google.com/google-ads/answer/15909989?hl=en
- https://support.google.com/google-ads/answer/15910187?hl=en
- https://support.google.com/google-ads/answer/16450159?hl=en
- https://blog.google/products/ads-commerce/ai-max-new-features/