Google Ads AI Max for Search in 2026: A Controlled Testing and Migration Playbook

The short answer: what AI Max actually is

AI Max is not a new campaign type. Google describes it as an optimization layer for existing Search campaigns. Its two central areas are search term matching and asset optimization. Search term matching can expand beyond an advertiser’s existing keywords by using broad-match and keywordless technology. Asset optimization includes text customization and final URL expansion, which can adapt ad copy and select a relevant page from the advertiser’s domain. These functions can be managed separately after AI Max is enabled, according to the current Google Ads explanation of how AI Max works.

This distinction matters. AI Max keeps the familiar structure of a Search campaign, while Performance Max is a separate, goal-based campaign type designed to access inventory across Search, YouTube, Display, Discover, Gmail and Maps. Google positions Performance Max as a complement to keyword-based Search, not simply another name for AI Max, as documented in its Performance Max overview.

Why the 2026–2027 transition timeline matters

There are two different transition dates, and combining them would create an inaccurate deadline. Google’s updated announcement says campaigns using automatically created assets or the campaign-level broad match setting remain scheduled to begin moving into AI Max in September 2026. Dynamic Search Ads follow a later timetable. On June 11, 2026, the Google Ads API team said DSA automatic migration had been postponed to February 2027, DSA creation was restored on June 15, 2026, and the extended period should be used for testing and voluntary migration. The same update says new DSA creation is expected to be removed in January 2027. See Google’s updated legacy-feature announcement and the June 2026 DSA migration update.

The operational lesson is straightforward: do not publish or act on the stale claim that every DSA will migrate in September 2026. Audit each account for DSA, automatically created assets and campaign-level broad match separately, then assign the correct transition window to each item.

What changes when AI Max is enabled?

1. Search term matching can move beyond the keyword list

Google says search term matching learns from existing keywords, creatives and URLs to find additional relevant queries through broad-match and keywordless technology. It is enabled at campaign level with AI Max and can be turned off for an individual ad group. Negative keywords are still respected, according to the official AI Max frequently asked questions.

2. Ad text and landing-page selection can become more dynamic

Text customization can use existing ads, landing-page copy, assets and generative AI to assemble more query-relevant messages. Final URL expansion can send a visitor to another relevant page on the same domain. Google states that final URL expansion requires text customization so the message can match the selected page. If final URL expansion or URL inclusions are used, pinned responsive-search-ad assets may not be honored. An advertiser that must preserve exact wording should therefore decide whether those settings are compatible before launch, rather than discovering the conflict after ads serve.

3. More automation also creates specific controls

Current controls include brand inclusions or exclusions, locations of interest, URL inclusions and exclusions, and controls over generated text. The reports can show an AI Max match type and source, plus combinations of search term, headline and landing page. These details make it possible to evaluate the customer journey rather than judging automation only from an account-wide conversion total. Google lists these controls and reports in its AI Max feature documentation.

AI Max, standard Search, Performance Max or DSA?

If AI Max and Performance Max are both eligible for the same query, Google says the candidate with the highest Ad Rank serves, as explained in its AI Max FAQ. Account design therefore needs to consider overlap, not treat the campaigns as isolated systems.

A practical decision matrix

The 97 Ads CONTROL-7 framework

CONTROL-7 is a pre-launch and review checklist. It is designed to prevent a common error: expanding reach before proving that the campaign is measuring the right outcome and sending people to the right pages.

C — Conversion truth

List every action included in bidding. Separate primary commercial outcomes from diagnostic events. A form submission, a qualified lead and a closed sale are not interchangeable. AI Max’s advanced search term matching relies on automated, conversion-based bidding; Google states that it does not function with manual CPC and recommends a conversion- or conversion-value-based strategy for full capabilities. This is a compatibility requirement, not proof that the underlying conversion data is trustworthy.

O — Offer and URL readiness

Crawl or manually review the pages AI Max could select. Confirm that offers are current, calls to action work, mobile layouts are usable and claims match the business. Create exclusions for pages that should never receive paid traffic, such as privacy notices, login areas, expired offers, recruitment pages or unrelated editorial content. If tracking templates are present, test dynamic landing pages. Google warns that incompatible templates and incorrect use of the {lpurl} parameter can produce incorrect destinations or 404 errors.

N — Negative and brand boundaries

Prepare negative keywords from known irrelevant intent. Define whether the campaign may match the advertiser’s own brand, competitor brands or ambiguous terms. Review geographic intent separately from the user’s physical presence where appropriate. Document URL rules and any words the generated message must avoid. Negative keywords remain valid with AI Max, but they still require active maintenance.

T — Test against a stable baseline

Use a campaign experiment where available and keep the evaluation objective consistent. Google’s DSA migration guidance explicitly recommends side-by-side testing rather than waiting for automatic migration. Avoid changing bidding goals, conversion definitions, geography, landing pages and AI Max at the same moment; otherwise the source of any performance movement becomes unclear.

R — Read the complete journey

Review the search-term report, AI Max source, headline and selected landing page together. The current reporting documentation also identifies keyword, asset and landing-page reports. Google advises waiting at least two weeks after activation before routine changes so the system can learn, but two weeks should not be misrepresented as sufficient for every business decision. The test also needs enough meaningful outcomes for the comparison being made. See the official AI Max reporting guide.

O — Outcome quality

Evaluate qualified leads, accepted opportunities, sales, revenue or another defensible business outcome. More conversions at the same reported CPA can still be worse if lead quality falls. Segment brand and non-brand demand where that distinction affects interpretation. Do not invent a universal minimum conversion threshold; the necessary evidence depends on variance, value and decision risk.

L — Learn, lock or leave

Exclude weak queries or landing pages, remove unsuitable generated assets and retain controls that protect message quality. Scale only when the business outcome improves within an acceptable range of uncertainty. If the test fails, record why and return to the baseline rather than hiding the result. If it succeeds, preserve the test definition and monitoring cadence before migrating additional campaigns.

AI Max launch checklist

1. Inventory Search campaigns using DSA, automatically created assets or campaign-level broad match.
2. Assign September 2026 or February 2027 to the correct legacy feature.
3. Verify primary conversions, values and bidding strategy.
4. Confirm the campaign is not constrained by a budget problem that invalidates the test.
5. Review all eligible landing pages and tracking templates.
6. Create brand, negative-keyword, location and URL guardrails.
7. Decide whether text customization, final URL expansion and search term matching may each be enabled.
8. Record the baseline and the business-level success criterion.
9. Launch a controlled experiment rather than an unmeasured account-wide switch.
10. After the learning period, inspect query–headline–URL combinations and outcome quality.
11. Document exclusions, asset removals and the final scale, revise or stop decision.

Frequently asked questions

Is AI Max a separate Google Ads campaign type?

No. Google defines it as an optimization layer activated inside an existing Search campaign in the official feature documentation.

How is AI Max different from Performance Max?

AI Max enhances Search campaigns. Performance Max is a separate goal-based campaign type designed to serve across Google’s advertising inventory. They can coexist and can be eligible for the same query.

Does AI Max eliminate keywords?

No. It can expand from existing keywords and also use keywordless matching based on ads, assets and landing pages. Standard Search structures remain relevant.

Are negative keywords respected?

Yes, according to Google’s current AI Max FAQ. Search-term review is still necessary because new relevant-looking but commercially weak intents can emerge.

Can AI Max change the landing page?

Yes, when final URL expansion is enabled. URL inclusions and exclusions should be prepared before testing, and tracking templates should be validated.

Will pinned responsive-search-ad assets always serve?

No. Google says pinning may not be respected when final URL expansion or URL inclusions are used. Businesses requiring fixed language should restrict those features or defer the test.

Can AI Max use manual CPC?

AI Max may be enabled with limited capability, but Google says search term matching does not work with manual CPC. Full functionality relies on conversion- or conversion-value-based automated bidding.

How long should a test run?

Google recommends at least two weeks before routine changes. That is a learning-period instruction, not a universal statistical rule. Continue until the test contains enough relevant business outcomes to support the decision.

When will Dynamic Search Ads migrate?

The current Google Ads API team timeline says automatic DSA migration begins in February 2027, with new DSA creation expected to end in January 2027. Automatically created assets and campaign-level broad match follow the separate September 2026 transition schedule.

Conclusion: automate reach, not accountability

AI Max can broaden query coverage, tailor messages and select more relevant landing pages, but those capabilities magnify the quality of the inputs and controls they receive. A sound decision begins with reliable conversion goals, a clean website, explicit guardrails and a test that measures commercial value. The approaching legacy transitions make preparation timely; they do not justify an uncontrolled switch.

For more planning context, visit the 97 Ads Google Ads services page, review additional articles in the marketing knowledge hub, or use the free ROI planning calculator. Businesses that want a cautious account-specific review can also request an initial consultation. No result is promised, and implementation should follow the needs and evidence of the individual account.

Primary sources

• Google Ads Help: How AI Max for Search campaigns works
• Google Ads Help: Set up AI Max for Search campaigns
• Google Ads Help: Frequently asked questions about AI Max
• Google Ads Help: Reporting in AI Max for Search campaigns
• Google: Updated transition from legacy Search features to AI Max
• Google Ads API Team: DSA automigration delayed to February 2027
• Google Ads API: AI Max for Search campaign documentation
• Google Ads Help: About Performance Max campaigns

Source review note: Google’s documentation is the primary source for product behavior, but it is also vendor-authored. Product claims have therefore been attributed to Google and are not presented as independent proof of performance. The page should be rechecked before future material revisions.