The AI Max Field Guide: What Google Gets to Decide

The September migration mechanics live in the companion dossier, AI Max by Default: The Verified Dossier on Google's September Auto-Upgrade. It covers cohorts, time line, opt-outs, and six theories of Google's motivation. We verified these against 66 dated sources. This piece explains how to run AI Max.

Evidence legend

• ● Verified fact. A documented product mechanic, checked against the cited Google page on the access date.
• ■ Company claim. Google's own performance number, published without explaining its method.
• ◆ Independent measurement. Third-party data with a named method and sample.
• ◇ First-party data. Measured on accounts we manage. We used Google's application programming interface (API), the Google Ads API, on August 25, 2026.
• △ Inference. Our reading of the evidence. We could be wrong.

One toggle, three grants of decision-making

● AI Max is an optimization layer within existing Search campaigns. Google's setup page states: "AI Max isn't a new campaign type but an optimization layer within existing Search campaigns." The layer includes search term matching and asset optimization. Asset optimization includes text customization and final URL expansion. (Google AI Max setup, accessed August 25, 2026)

Each component grants Google a separate kind of control:

1. Search term matching: the query grant. ● With it on, "all keywords are treated as broad match by default unless search term matching is disabled at the ad group level," according to Google. Google also adds keywordless expansion from your landing pages. (Google Ads API AI Max guide, updated August 19, 2026)
2. Text customization: the copy grant. Google generates headlines and descriptions from your site, existing ads, and keywords. ● Generated assets carry the label "Google AI" in the Added by column. You can remove them, but the interface doesn't support bulk removal. (Google AI Max reporting, accessed August 25, 2026)
3. Final URL expansion: the destination grant. Google selects landing pages beyond your final URLs. This turns your website into a targeting feed. That is why we include the site audit gate below.

● You can adjust each grant, but they are linked. "In order to use the Final URL expansion setting, Text customization must be selected. Deactivating text customization automatically disables Final URL expansion." (Google AI Max setup, accessed August 25, 2026)

● Google's migration defaults separate the components, supporting separate decisions about each. Campaign-level broad match campaigns convert with search term matching on and text customization off. Legacy ACA campaigns convert with both on. Final URL expansion stays off for both. (Google Ads Developer Blog, published August 12, 2026)

The controls map

Each control works at a set level. That level decides who can contain a problem when it appears. ● Every row comes from Google's How AI Max works page and the API guide, accessed August 25, 2026.

● Almost nobody discusses the bundling_required flag. The API says it helps "determine if AI Max for Search campaigns must be enabled to respect or modify text customization and brand list controls." A REQUIRED value means those component settings need AI Max enabled.

◇ We see REQUIRED on 13 of our 18 enabled Search campaigns. Most of our campaigns therefore need AI Max as a package at the campaign level. The ad-group and list controls above provide the real limits.

September makes the first grant automatic

From September 1 to 30, Google converts every Search campaign still using either legacy setting. The triggers are campaign-level broad match and standalone automatically created assets. These campaigns become AI Max in place, with the different defaults described above. Google already blocked new legacy configurations on August 3.

September 1 is the deadline to prevent automatic mapping. Current documentation says you can still change AI Max settings after conversion. The full mechanics and opt-out paths are in the dossier.

We measured eligibility on real accounts the week before the window. The results look nothing like the headlines suggest. ◇ Our check covered all 19 Google Ads accounts under our manager account. The snapshot from August 25, 2026, shows the following:

• Zero of 18 enabled Search campaigns use campaign-level broad match. The setting exists on 2 paused campaigns across our accounts.
• The live trigger is legacy automatically created assets. We found 25 campaigns still opted in to ACA. Most are paused historical campaigns nobody looks at.
• The AI Max flag already appears on 14 campaigns, with 2 enabled. Those are our own agency campaign and our own wraps-shop campaign. Both are deliberate tests using our own money.
• The remaining auto-conversion path holds 3 campaigns, and exactly 1 is enabled and spending. That client location campaign recorded $1,199.20, 284 clicks, and 22 conversions in the last 30 days. It is the entire September decision for this My Client Center (MCC) account. We are making that decision deliberately before the window opens.

△ The migration language does not exempt paused campaigns. A paused campaign could convert silently in September. Months later, it could restart with AI Max settings nobody chose. Review everything, including campaigns that aren't spending.

Update, November 13, 2026. We ran the check again after the window closed. All three campaigns on the path now carry a migration date from Google, and so do 10 others. Read what the migration did.

The spectrum: from expansion engine to precision defence

The strongest AI Max campaign gives Google the widest safe room to decide. Advertisers must be able to measure, govern, and fund those decisions. Our spectrum measures fit for expansion, not enthusiasm for automation. We place every Search campaign at one of six positions:

Campaigns earn their way toward A through controlled tests of each component. No campaign starts at A simply because the toggle exists. Campaigns can leave B when they pass the gates below.

The seven gates

A campaign must pass gates to move up the spectrum. Record evidence at each gate, rather than just a yes.

Gate 1: Measurement integrity. AI Max optimizes to the conversion signals it receives. Your account must tell Google which form fills represent valuable customers. Are qualified outcomes imported? Does value reach Google before most optimization decisions? Do you exclude or devalue spam, duplicate, and unserviceable leads?

Green means qualified value flows reliably, so the campaign may move toward A. Yellow means partial or delayed feedback, so stay in the middle positions. Red means bidding runs on raw form counts. Stay near B until you repair measurement.

Gate 2: Economic room. Expansion only pays when the business can handle more qualified demand at similar margins. Hard capacity limits support B-side positions. So do mixed-margin catalogues and defence-only objectives, regardless of data quality.

Gate 3: Query opportunity. Expansion needs valuable demand that your keywords miss. That demand may hide in long-tail, conversational, or emerging queries. Search term matching has a job when those gaps exist. If your keywords already cover profitable demand, the grant buys risk without upside.

Gate 4: Creative tolerance. Generated headlines can misstate price, eligibility, location, or offer terms. Could that happen here? Who removes an inaccurate asset, and how fast? The hard-stop question concerns legal, patient-safety, or severe brand risk. If generated ads could create those risks, why is text customization on?

Gate 5: Site fitness. Final URL expansion treats your website as a feed. Audit the site as a feed before enabling it. Google can pick old offers, thin pages, careers pages, staging content, and unserviceable location pages. Exclusions must cover those destinations.

Gate 6: Brand, geography, and policy. Set business rules for your own brand, competitors, and jurisdiction limits. Apply them through brand lists, negatives, and location controls with a named owner.

Gate 7: Governance. Someone must check settings after launch and migration. That person must regularly review queries, assets, and landing pages, and hold rollback authority. Grant extra freedom only when someone can monitor it and act.

The scorecard

Score every gate 0, 1, or 2: fails, partial, or passes with recorded evidence. The maximum is 14. The total maps to a starting position on the spectrum:

Two overrides take priority over any total. Gate 1 at 0 caps the campaign at B, regardless of other results. Otherwise, every grant would optimize to the wrong signals. A Gate 4 hard-stop keeps text customization off at any score. No conversion lift properly accounts for the cost of a compliance incident.

What the independent record shows

The public evidence is mixed. Its differences follow the lines our spectrum predicts.

• ■ Google's April 2026 material reports a gain from the full AI Max suite. It produced 7% more conversions or conversion value than search term matching alone. Cost per action (CPA) or return on ad spend (ROAS) stayed similar. Google used internal non-Retail data. It disclosed no sample size, test length, or significance method. (Google, published April 15, 2026)
• ◆ Smarter Ecommerce measured more than 250 retail Search campaigns. The median increase was 13% for conversion value and 16% for CPA. The median ROAS difference was 0%. Scale rose, but efficiency did not. (Smarter Ecommerce, published March 23, 2026)
• ◆ Location3 ran AI Max through Google Experiments. The test covered five locations of one home-services franchise brand. It ran from December 1, 2025, to January 31, 2026. Reported conversion lifts were between +97% and +720%. CPA ranged from flat to a 63% improvement, depending on location. (Location3, published March 18, 2026)
• ◆ Brad Geddes of Adalysis found weaker results in highly optimized campaigns. "AI text customization wasn't as effective as human management of assets for highly optimized campaigns," he reported. He also found "AI performed quite well for the long-tail campaign" where human optimization had been limited. His business-to-business (B2B) test failed badly at qualifying the audience. The company "stopped the tests and went back to pinning its ads" after three weeks. (Search Engine Land, published July 28, 2026)
• ◆ Lunio's classifier measured invalid-traffic rates across about 49.5 million AI Max clicks. It found 2.46% in Q4 2025, 4.01% in Q1 2026, and 5.28% in Q2 2026. Standard Search ended Q2 2026 at 3.07% in the same analysis. Traffic quality needs its own monitoring line, separate from CPA. (Lunio, updated August 14, 2026)

△ Expansion helps where coverage was weak, including franchise locations and neglected long-tail ad groups. It struggles where humans already did extensive optimization. Examples include mature retail portfolios and pinned B2B copy. These findings make component fit more useful than a yes-or-no adoption decision.

What AI Max served on our own money

We deliberately run AI Max on two first-party campaigns. This lets us watch the grants in action without risking a client's budget. ◇ Both readings come from the Google Ads API's ai_max_search_term_ad_combination_view report. It shows exactly which search term, headline, and landing page AI Max combined. The period covers the last 30 days as of August 25, 2026.

Our agency account's Search campaign has AI Max and text customization on. It recorded 1,346 impressions, 36 clicks, and $115.32 in spend. The combination report recorded zero AI Max combinations. △ This tightly themed campaign had full keyword coverage, leaving little room for expansion. Low-volume rows can also fall below search-terms reporting thresholds, so we read this as minimal expansion.

Our vehicle-wraps campaign recorded 4 combinations totalling 5 impressions and 0 clicks. All four teach us something:

• The query "billboard sign" showed the headline "Serving Edmonton & Area" and led to the fleet decals page. This matched billboard intent to a decals destination outside the catalogue. A keyword plan would never have made that match.
• The query "nd graphics edmonton", a local competitor's brand, matched twice. It used "Edmonton Vinyl Lettering | Edmonton Pros Since 2010" and "Vehicle Wraps & Vinyl Graphics | Edmonton Truck Wraps & Decals" as headlines. Competitor conquesting happened by default. Brand exclusions therefore need your decision, rather than arriving as a default.
• The query "work van decals" shows the grant working as sold. It matched cleanly to the truck-wraps page with "Free Wrap Consultation | Professional Vehicle Branding" as the headline.

Five impressions tell us nothing about performance. They show what you can inspect: each expanded term, headline, and URL. Small campaigns take minutes to review. Larger ones take hours to sort through. Build this review habit before traffic grows.

◇ Keyword-attributed Search spend shows where our accounts earn results today. Across all 19 accounts, the split covers the last 30 days. Phrase accounted for 55.3%, exact for 32.5%, and broad for 12.3%. △ Expansion must beat a precision baseline that already converts. That is why we grant it by campaign and component, based on evidence.

The operating rhythm

Four reports cover the whole monitoring workload. All four exist today.

1. Search terms, filtered to match type "AI Max", with the Source column added. ● Google documents this as the way to inspect expanded traffic. Log spend share as well as term counts. Low-volume queries stay hidden under the standard privacy threshold. (Google AI Max reporting, accessed August 25, 2026)
2. Assets, filtered to "Google AI" in the Added by column. Check generated text against your list of claims. Remove anything that misstates the offer.
3. Landing pages, with the Selected by column. ● The column shows when final URL expansion picked the page. Compare chosen pages with your approved set.
4. Lead quality, from your own customer relationship management (CRM) system. The platform cannot see poor lead quality from volume alone. It needs you to send qualification data back.

Check settings the day after launch or migration. Review weekly through the first month. After that, choose a review schedule that traffic justifies.

Set rollback triggers before you start. A generated claim error or unsafe landing page triggers removal. So does brand leakage above an agreed share. Invalid-lead rates above an agreed line also trigger removal. Revoke the relevant grant when a trigger fires, without debating it.

Failure modes worth naming

• Irrelevant expansion: AI Max matches demand you cannot serve or do not want. The Source-labelled terms report catches it.
• Generated claim errors: Copy misstates price, eligibility, or geography. The Added by report catches it, and a named owner removes it.
• Unsafe destinations: Google sends traffic to pages never meant to sell. URL exclusions prevent this, and Selected by catches it.
• Brand leakage: Brand and competitor traffic blur nonbrand results. Brand lists and separate brand reporting prevent this.
• Invalid traffic: Track it as its own line, using the Lunio numbers above.
• False attribution: A campaign claims conversions another campaign would have won anyway. Check results at account level. Treat reattribution as noise rather than growth.

Graduation rules

• A campaign moves toward A only with controlled-test evidence of added qualified value across the account. Test one component at a time.
• ● Google is improving testing tools in time for the migration. It announced multi-campaign A/B experiments launching in September 2026. It also announced Performance Planner support to preview bidding and budget changes. (Google, published August 20, 2026)
• A campaign moves back toward B as soon as a rollback trigger fires. Restore the grant only after fixing the gate, rather than relying on memory.
• When a test is inconclusive, the component stays off. Automation carries the burden of proof because the precision baseline already pays the bills.

The preflight: four queries that audit any account

You can query every field name in this guide. We ran these four requests for the account audit above. We verified each against Google Ads API v22 on real accounts on August 25, 2026.

All four are read-only. Run them through the API, a script, or the agent helping manage your account. The interface shows the same facts. API field names are exact, and you can compare outputs cleanly each week.

◇ 1. September eligibility and current settings, one row per Search campaign. BROAD under keyword_match_type marks the campaign-level broad match cohort. TEXT_ASSET_AUTOMATION at OPTED_IN marks the legacy ACA cohort. enable_ai_max shows what already converted or opted in. A bundling_required value of REQUIRED means the component settings need AI Max enabled.

SELECT campaign.id, campaign.name, campaign.status,
  campaign.keyword_match_type,
  campaign.asset_automation_settings,
  campaign.ai_max_setting.enable_ai_max,
  campaign.ai_max_setting.bundling_required
FROM campaign
WHERE campaign.status != 'REMOVED'
  AND campaign.advertising_channel_type = 'SEARCH'

2. Ad-group opt-outs on campaigns already running AI Max. An ad group without the flag in its row has search term matching active. The API leaves out unset fields rather than returning false.

SELECT ad_group.id, ad_group.name,
  ad_group.ai_max_ad_group_setting.disable_search_term_matching
FROM ad_group
WHERE campaign.ai_max_setting.enable_ai_max = true

3. Every combination AI Max served: the term, the headline, and the landing page. The competitor-conquesting example above came from this report. This is AI Max's most useful monitoring report, yet almost nobody pulls it.

SELECT ai_max_search_term_ad_combination_view.search_term,
  ai_max_search_term_ad_combination_view.headline,
  ai_max_search_term_ad_combination_view.landing_page,
  metrics.impressions, metrics.clicks
FROM ai_max_search_term_ad_combination_view
WHERE segments.date DURING LAST_30_DAYS

4. Who changed what during the window. Use client_type and user_email to identify who made each change. They separate your team's changes from rule-based, API, and Google-initiated ones. This confirms what the migration wrote to your account.

We hit three traps during live use. change_event rejects relative ranges such as DURING LAST_30_DAYS. It requires a LIMIT. It also rejects start dates more than 30 days back, returning START_DATE_TOO_OLD. Schedule this audit inside the month it matters.

SELECT change_event.change_date_time, change_event.change_resource_type,
  change_event.changed_fields, change_event.client_type,
  change_event.user_email
FROM change_event
WHERE change_event.change_date_time >= '2026-09-01'
  AND change_event.change_date_time <= '2026-09-30'
ORDER BY change_event.change_date_time DESC
LIMIT 200

The September checklist

1. Inventory eligibility across every account you touch. Preflight query 1 returns both triggers and current settings in one pass per account. Keep the output as your before picture. Compare your post-migration check against it.
2. Decide per campaign before September 1. Choose among three honest options. Adopt deliberately with your chosen settings, or remove the legacy setting to avoid mapping. Or allow migration with a calendar entry to verify the result. Doing nothing chooses the third option without that calendar entry.
3. Sweep paused campaigns too. They carry eligible settings into the window. They convert without anyone noticing.
4. After conversion, verify the defaults landed as documented for your cohort. Check ad-group search term matching flags. Run the first weekly review immediately.
5. Keep the change history. Preflight query 4 separates changes made by your team, rules, the API, and Google. The migration writes to your account. Your audit trail should record those changes.

The bottom line

Ask which decisions Google may make for this campaign, rather than whether AI Max works. Define the limits and the evidence needed before granting more freedom.

A-side campaigns turn trustworthy signals, deep sites, and scalable economics into added value. B-side campaigns protect precision, message control, and capacity limits that expansion would damage. Middle positions use ordinary account management based on evidence. They test individual components with named owners and rollback triggers.

The appendix your agent can read

A human-only guide is half a product when more Google Ads accounts have AI agents watching them. The block below gives your agent this guide's operating facts in a reusable form. Paste it with the four preflight queries into whatever tool helps run your account.

The appendix makes no claims beyond those already sourced above. It separates verified product mechanics from our operating framework. Your agent should be free to disagree with that framework.

# AI Max operating facts
# Source: https://choice.marketing/blog/google-ai-max-field-guide/
# Mechanics verified against Google documentation and Google Ads API v22, 2026-08-25.

# --- verified product mechanics ---
migration:
  window: 2026-09-01 to 2026-09-30
  triggers: [campaign_level_broad_match, legacy_automatically_created_assets]
  legacy_creation_blocked_since: 2026-08-03
  converted_defaults:
    broad_match_cohort: {search_term_matching: on, text_customization: off, final_url_expansion: off}
    aca_cohort:         {search_term_matching: on, text_customization: on,  final_url_expansion: off}
  paused_campaigns: no exemption documented (our inference), sweep them too
api_fields:
  umbrella: campaign.ai_max_setting.enable_ai_max            # campaign level
  bundling: campaign.ai_max_setting.bundling_required        # REQUIRED = components need the umbrella
  broad_trigger: campaign.keyword_match_type                 # BROAD = campaign-level broad match
  aca_trigger: campaign.asset_automation_settings            # TEXT_ASSET_AUTOMATION + OPTED_IN
  stm_opt_out: ad_group.ai_max_ad_group_setting.disable_search_term_matching
  served_combinations: ai_max_search_term_ad_combination_view  # search_term, headline, landing_page
  change_audit: change_event   # explicit date range + LIMIT required, 30-day lookback max
component_rules:
  final_url_expansion_requires: text_customization
  search_term_matching_default: on when the umbrella enables, disable per ad group
  negative_keywords: honored unchanged
monitoring_reports:
  - search terms filtered to match type "AI Max", with the Source column
  - assets filtered to "Google AI" in the Added by column
  - landing pages with the Selected by column
  - CRM lead quality compared against platform-reported conversions

# --- Choice OMG operating framework ---
placement_spectrum:
  A:  {stm: on,  tc: on,  fue: on,  use: full expansion engine}
  A2: {stm: on,  tc: on,  fue: off, use: controlled expansion}
  M1: {stm: on,  tc: off, fue: off, use: matching-led expansion}
  M2: {stm: off, tc: on,  fue: off, use: creative assist}
  B2: {stm: one ad group, tc: off, fue: off, use: contained experiment}
  B:  {stm: off, tc: off, fue: off, use: precision defense}
readiness_gates: [measurement_integrity, economic_room, query_opportunity,
  creative_tolerance, site_fitness, brand_geo_policy, governance]
scoring:
  scale: 0-2 per gate, 14 max
  bands: {12-14: trial A2, 8-11: M1 or M2, 4-7: B2, 0-3: B}
  overrides:
    - gate 1 (measurement) at 0 caps placement at B regardless of total
    - gate 4 (creative risk) hard-stop keeps text_customization off at any score
rollback_triggers: [generated_claim_error, unsafe_landing_page,
  brand_leakage_past_agreed_share, invalid_lead_rate_past_agreed_line]

An appendix field that stops resolving signals an API version change. Check the source date above against Google's current release notes before trusting the rest.

Sources checked for this guide

All Google Help Center pages accessed August 25, 2026.

• Google, AI Max setup
• Google, How AI Max works
• Google, AI Max reporting
• Google Ads API, AI Max for Search campaigns, updated August 19, 2026
• Google Ads Developer Blog, migration announcement, published August 12, 2026
• Google, AI Max testing and planning tools, published August 20, 2026
• Google, AI Max upgrade announcement, published April 15, 2026
• Smarter Ecommerce, AI Max guide, published March 23, 2026
• Location3, AI Max franchise results, published March 18, 2026
• Search Engine Land, Geddes AI Max copy test, published July 28, 2026
• Lunio, AI Max invalid traffic rates, updated August 14, 2026
• Choice OMG, AI Max auto-upgrade dossier, published August 25, 2026
• First-party account data: Google Ads API v22, MCC sweep of 19 accounts, August 25, 2026