Gabriel Espinheira
Google moved Dynamic Search Ads auto-upgrades to February 2027, but not every September migration moved. Automatically created assets and campaign-level broad match still begin moving into AI Max for Search in September 2026. One switch can influence the search, the message and the landing page before your CRM decides whether the enquiry was worth having.
That is the part most AI Max guides skip. Google can find a new query, write a headline and choose a destination. None of those actions defines a sale for your business. If your agency cannot connect the whole route to a qualified enquiry, a better Google Ads chart proves very little.
TL;DR: AI Max for Search can expand queries, customise ad text and choose landing pages. Before enabling it or accepting a legacy-setting migration, audit those three controls against one qualified CRM outcome. Test one campaign, inspect the full route weekly and scale only when the extra reach produces enquiries worth calling.
Which AI Max changes still start in September 2026?
Two legacy settings remain on the September schedule: automatically created assets and campaign-level broad match. Google delayed Dynamic Search Ads auto-upgrades until February 2027 in a June update to its migration announcement.
That split is easy to miss. An owner sees the revised DSA date, closes the alert and assumes the account can wait. Meanwhile, another Search campaign uses automatically created assets or campaign-level broad match and still enters the September upgrade path.
Google says the default mapping depends on the setting already in use:
- Dynamic Search Ads will move later with search-term matching, text customisation and final URL expansion enabled, while preserving legacy URL controls.
- Automatically created assets will move with search-term matching and text customisation enabled.
- Campaign-level broad match will move with search-term matching enabled.
Open the campaign settings before you discuss performance. Record which legacy feature is active, which AI Max controls will switch on and which pages the campaign may reach. A migration banner is an implementation detail. The owner still needs to approve the operating boundary.
What does AI Max for Search actually get to choose?
The feature suite can change three connected decisions inside an existing Search campaign: who triggers the ad, what the ad says and where the click lands. Google's current AI Max documentation describes search-term matching, text customisation and final URL expansion as separate controls for good reason.
| Decision | AI Max influence | Evidence to inspect |
|---|---|---|
| Search | Expands beyond the keyword list through broad and keywordless matching | Search term, match source, brand and location controls |
| Message | Generates or customises headlines and descriptions from ads, assets and website copy | The exact headline served and the source page or asset |
| Destination | Selects a relevant page when final URL expansion is enabled | The landing page used, its offer and its conversion path |
| Qualified outcome | No platform setting knows your commercial definition on its own | CRM stage, fit, sales response and eventual value |
The fourth row is where senior judgement starts. A search can look relevant while the generated headline makes a broader promise than the business can keep. The headline can be accurate while final URL expansion sends the visitor to an old article, a login page or a service page built for another market.
Responsive search ad pinning does not settle this by itself. Google notes that pinned assets may not be respected when text customisation and final URL expansion, or URL inclusions, are active. Treat the generated search, headline and URL as one route. Inspecting only the keyword leaves two-thirds of the decision untouched.
Why Google's 7% is not your verdict
Google reports that the full AI Max suite produced an average of 7% more conversions or conversion value at a similar CPA or ROAS than search-term matching alone. The footnote identifies Google internal 2026 data for non-retail advertisers. It is useful evidence that the suite can add value. It is not a forecast for your account.
Picture the Friday review. Google Ads shows another conversion and the cost per lead still looks acceptable. The CRM record belongs to a job applicant, a student researching the service or an enquiry outside the area the business can serve. The platform found volume. The sales inbox found work it should never have received.
AI Max may uncover demand your keywords missed. It also gives a plausible-looking match three places to go wrong: the query, the claim and the page. More reach earns its place only when those extra routes end in qualified outcomes.
Do not judge the experiment from total conversions alone. Compare AI Max expansion with the advertiser-provided keyword traffic, then read the actual enquiries. A campaign can improve inside Google Ads and deteriorate inside the business. The CRM gets the deciding vote because the CRM knows whether anybody should call back.
How do you audit the route before enabling AI Max?
A useful AI Max audit needs four artefacts: the campaign controls, the source copy, the eligible landing pages and the CRM definition of a qualified enquiry. If one is missing, the switch is early.
Set the search boundary
Start with one established non-brand campaign rather than changing the whole account. Read the current search terms, negative keywords, brand controls and location settings. Google confirms that negative keywords remain respected, so use them to document traffic the business has already decided against.
Write the boundary in plain English. A Lisbon consultancy might accept searches from decision-makers across Europe while rejecting job searches, free templates and markets it cannot support. The keyword list is only one input. The boundary is the commercial decision behind it.
Inspect every claim the system can borrow
Text customisation learns from the website and existing assets. Open the pages the campaign uses and mark any sentence you would refuse to put in an ad: expired offers, stale timeframes, unsupported superlatives, old service names or language aimed at a different country.
Then read the responsive search ad assets. If a pinned line protects a legal or commercial claim, verify whether the chosen AI Max settings preserve it. A pin you assumed was fixed can lose that role when final URL expansion enters the route.
Control the destination set
Final URL expansion should not inherit the whole domain without inspection. Google's AI Max reporting guidance for developers uses careers, privacy, login and outdated blog pages as examples of destinations that may need URL exclusions.
Run the eligible pages through a simple test. Does the page name the same service as the search? Does the first screen make the same promise as the ad? Can the visitor take the intended next step? Is that step measured? One failed answer is enough to exclude or repair the page before paid traffic reaches it.
Define the qualified outcome
Choose the CRM stage that counts before the test starts. “Form submitted” is too early for most service businesses. A useful definition might require the right service, a market the business serves and genuine buying intent. Keep it simple enough that sales can apply it consistently.
Name the person who owns the classification and the time by which it happens. AI Max can only learn from the feedback the account receives. An unreviewed inbox turns every form fill into approval by default.
What should the first two weeks prove?
Google's own reporting guide tells advertisers to “wait at least 2 weeks after enabling AI Max” before making changes such as adding negative keywords. That asks for test discipline. It does not excuse a false claim, an unsafe destination or obvious traffic outside the agreed boundary.
The work trail should show one row for every route worth discussing:
| Search term | Match source | Headline served | Landing page | Google Ads result | CRM result | Decision |
|---|---|---|---|---|---|---|
| Exact query | Keywordless or broad expansion | Exact generated or selected headline | Final URL used | Click and conversion status | Qualified, rejected or pending | Keep, exclude, repair or investigate |
Google's API documentation calls the search-term, headline and landing-page combination view its most granular AI Max report. Use it for diagnosis. Do not add its totals to the standard search-term view because the views can describe the same traffic and double-count it.
Put that row inside SharpOS and link it to the CRM outcome. A weekly update should not say “AI Max is learning”. It should show which route appeared, whether the enquiry was qualified and what decision follows. The owner can inspect the judgement without booking a status call.
AI Max can write the route. It cannot decide what a qualified enquiry is.
When is AI Max for Search the wrong experiment?
Leave the switch alone when the account cannot answer any of these questions:
- Which current conversion represents a qualified business outcome?
- Who will review the search-term, headline and landing-page route each week?
- Which pages are safe to receive paid traffic?
- Which claims must stay fixed for brand, legal or commercial reasons?
- Can the campaign absorb a learning period without forcing a quick verdict?
Manual CPC is another fit issue. Google's FAQ says AI Max search-term matching relies on automated bidding signals and does not work fully with Manual CPC. A campaign that is also limited by budget may receive an alert because the system has less room to optimise. Those are constraints to resolve or accept before testing, not after the first poor week.
The private objection in PPC communities is blunt: “not yet, I like to be in control.” Blind adoption is reckless. Permanent refusal wastes a useful test. Control moves from hand-picking every keyword to defining boundaries, inspecting the route and making the business outcome visible. If nobody owns those jobs, AI Max is the wrong experiment for now.
Frequently asked questions
Is AI Max a new Google Ads campaign type?
No. AI Max for Search is a feature suite added to an existing Search campaign. It can enable search-term matching, text customisation and final URL expansion while retaining the campaign structure. Performance Max is a separate campaign type that can run across more Google inventory than Search.
Will Dynamic Search Ads move to AI Max in September 2026?
No. Google updated the timetable in June 2026 and now says Dynamic Search Ads auto-upgrades begin in February 2027. Automatically created assets and campaign-level broad match remain scheduled to start auto-upgrading in September 2026, so check the legacy setting used by each campaign.
Does AI Max respect pinned responsive search ad assets?
It depends on the controls enabled. Google says pinning may not be respected when text customisation and final URL expansion are enabled together, or when URL inclusions are used. If a pinned line protects a required claim, verify the live combination or restrict those settings before launch.
Should every Search campaign use AI Max?
No universal rule survives contact with the account. Test AI Max where the conversion signal is reliable, the landing pages are current and somebody can inspect the expanded routes. Keep it off when the business outcome is vague, the destination set is unsafe or nobody owns the review.
The switch is permission, not proof
AI Max for Search can widen demand, adapt the message and choose a better page. That is useful power. It deserves a visible boundary and a business scorecard before it receives more budget.
SharpHaw runs Google Ads as part of one search-to-sale system, with the work and decisions visible every week. Review the current Plans if you want the full operating model, or book a 30-minute fit check. Bring one Search campaign, the pages it can reach and the CRM stage you call qualified. You will leave knowing which controls to lock before the switch moves.
Digital work that compounds.
