AI Max Is Replacing Legacy Search Settings. Treat the Upgrade as a Governance Change

Google is moving more Search campaign decisions into AI Max. That does not make paid search strategy obsolete. It changes where the important decisions sit.

The immediate timetable deserves care. Google’s updated announcement says campaigns using Automatically Created Assets and the campaign-level broad-match setting begin automatic upgrades from September 2026. The Dynamic Search Ads sunset and automatic upgrade, however, were extended to February 2027.

That distinction matters. Teams that assume every legacy feature changes at once may rush the wrong work. Teams that treat the transition as a harmless interface update may miss changes to targeting, generated copy and landing-page selection.

The better response is a controlled migration: understand which decisions the platform is taking on, define the boundaries it must respect, and verify that measurement still reflects commercial value.

What AI Max changes

Google describes AI Max as an optimisation layer within existing Search campaigns, not a new campaign type. Its feature set can include:

• search-term matching beyond the advertiser’s existing keyword set;
• text customisation based on ads, keywords, landing pages and other business inputs;
• final URL expansion, which can select a different relevant page on the advertiser’s domain;
• controls for brands, locations, URLs and messaging.

This expands the platform’s decision space. Instead of only optimising bids against a tightly defined set of keywords, ads and destinations, the system can have more influence over who is matched, what is said and where the visitor lands.

That can create useful reach. It also makes the quality of the inputs and controls more important.

Google reports that campaigns using the full AI Max feature suite averaged 7% more conversions or conversion value at a similar CPA or ROAS than campaigns using search-term matching alone. That is Google’s aggregate result, not a promise for an individual advertiser. The commercial question is not whether the feature can work. It is whether it improves outcomes that matter to your business without creating unacceptable message, destination or measurement risk.

The timetable is split

The headline “DSA is upgrading to AI Max” can obscure the current sequence.

Google’s updated product announcement says the Dynamic Search Ads sunset and auto-upgrade begin in February 2027. It also says the auto-upgrade of campaigns using Automatically Created Assets and the campaign-level broad-match setting continues from September 2026.

In practice, marketers should not work from a single calendar date. They should identify which legacy settings each Search campaign uses and read the notices inside their own Google Ads account.

This is also a reason to avoid waiting passively for an automatic change. A voluntary, measured test gives the team more opportunity to inspect settings, compare behaviour and correct problems before a migration becomes unavoidable.

Five things to review before enabling or accepting the upgrade

1. Conversion signals

Automation will optimise towards the outcomes it can see. A form submission, phone click or online purchase may be useful, but it may not represent margin, lead quality, cancellations or repeat value.

Before expanding matching, check that the primary conversion actions reflect the business outcome you actually want. Remove duplicated or low-value actions from bidding where appropriate. If offline outcomes matter, decide how qualified leads or completed sales will flow back into measurement.

More reach is not necessarily better if the feedback loop rewards the wrong behaviour.

2. Search and brand boundaries

AI Max includes controls for search-term matching and brand inclusions or exclusions. Those controls need an explicit policy.

Define the categories, competitor contexts and ambiguous searches that are acceptable. Identify regulated, reputational or commercially irrelevant areas that should be excluded. Review search-term reporting after launch rather than assuming the initial setup will remain safe as demand changes.

The goal is not to constrain the system until it can do nothing. It is to give it a commercially sensible field to explore.

3. Generated messaging

Text customisation can generate assets using content from the domain, landing pages, existing ads and keywords. That makes the website part of the advertising input.

Audit the pages the platform may learn from. Remove stale offers, unsupported claims and language that no longer reflects the brand. Use available text or messaging guidelines for words, promises and tones the system should avoid.

Google has also announced richer AI Brief controls for messaging, matching and audiences, but rollout language is gradual. Verify what is available in the account rather than designing a process around a control that has not appeared yet.

4. Landing-page routing

Final URL expansion can choose a relevant page beyond the advertiser-provided destination. That can help match a detailed query to a more useful page, but it changes the customer journey.

Create URL exclusions for pages that should never receive paid traffic: expired offers, recruitment pages, policy pages, weak content and products that are unavailable or unsuitable. Check analytics tagging and conversion behaviour across eligible destinations.

There is also a creative implication. Google notes that pinned Responsive Search Ad assets are not used when final URL expansion selects another page, because the pinned message may not match that destination. Teams relying on tightly controlled message sequencing should examine this before switching the feature on.

5. The migration record

When Dynamic Search Ads are upgraded, dynamic ad groups become standard ad groups and existing DSAs become Responsive Search Ads. Google says legacy ads may run temporarily while new assets are generated and pass policy checks. Some old DSA URL-rule types can become read-only.

Capture the pre-migration setup: campaign settings, URL rules, exclusions, brand controls, key queries, ad assets and performance baselines. Record what changes during the upgrade and who owns the post-change review.

Without that record, a performance movement can be difficult to separate from seasonality, budget changes or the migration itself.

A practical test plan

Choose campaigns where conversion quality is reasonably observable and where the commercial risk is manageable.

Run a controlled experiment where the account supports it. Agree in advance on the evaluation window, budget limits and success measures. Review not only CPA or ROAS, but also:

• search-term relevance;
• lead or order quality;
• generated-message accuracy;
• landing pages selected;
• brand and URL exclusions triggered;
• conversion volume and value outside the platform where possible.

Do not change every variable at once. If bidding, budgets, conversion definitions, creative and matching all move together, the team will struggle to explain the result.

Automation increases the value of judgement

AI Max continues the shift from manually specifying every search interaction towards setting objectives, supplying evidence and governing a broader optimisation system.

That is not a reason to defend legacy campaign mechanics for their own sake. Nor is it a reason to accept every default.

For New Zealand marketing teams, the useful question is straightforward: if Google can make more targeting, copy and destination decisions, have we improved the inputs, boundaries and feedback needed to make those decisions commercially sound?

Treat the upgrade as a governance change and the work becomes clearer. Audit the signals. Define the boundaries. Test deliberately. Keep a human owner accountable for what the system learns to do.