Your AI Max problem this week is not another on/off debate.
On September 23, 2026, Google’s Ads & Commerce blog said two things that should change how media teams staff Search: AI Brief’s closed beta expands beyond English into Dutch, French, German, Italian, Japanese, Portuguese, and Spanish, and Google is packaging a unified Search ads journey report that lines up the search term that triggered the ad, the creative assets the user saw, and the landing page they hit. Search Engine Roundtable’s Barry Schwartz, writing the same day from the DMEXCO framing, put Dan Taylor’s product pitch in plain English: steer messaging and audience in more languages, then evaluate AI Max through an end-to-end journey view, with fuller availability details later this year.
That is not a feature brochure for people who still think AI Max is a checkbox. It is Google telling you where control actually lives after the September settings migration.
If Monday’s Vault piece (The upgrade is not a toggle) was about whether campaign-level broad match and standalone Automatically Created Assets got stamped into AI Max without a click, today’s job is different on purpose. Migration audited settings. Delivery audits the brief and the journey. Confusing those two tickets is how agencies burn a week re-checking toggles while generative copy and Final URL expansion keep shipping combinations nobody briefed.
What landed on September 23
Three facts. Keep them tight.
Fact one: AI Brief is the natural-language control panel, not a slogan. Brandon Ervin’s April 30, 2026 product post defined the operator object when AI Brief launched for AI Max’s first birthday. Messaging Guidelines tell the system what ads should and should not say (example: never mention prices). Matching Guidelines set boundaries for searches you want or want to avoid (example: prioritize healthy pantry staples). Audience Guidelines steer who gets which message (example: for health-conscious people, highlight clean products). Gemini-powered previews of sample assets and searches exist so you can iterate before you commit. Existing text guidelines move into Messaging Guidelines when you start. The September 23 post does not reinvent those fields. It expands the closed beta into seven more languages so non-English operators can write the same kind of brief in their own words.
Fact two: the journey report is term plus creative plus URL, not a prettier query list. Google’s September 23 language is explicit: a single unified view of the Search ads journey so you can see what search terms triggered ads, which creative assets the user saw, and exactly where they landed. Help Center already documents the practical surface under “About reporting in AI Max for Search campaigns”: from the search terms report, open the Search terms dropdown and choose “Search terms and ad combinations” (formerly “Search terms and landing pages from AI Max”). That view lists search terms with the headlines, landing pages, campaigns, and ad groups that made up the customer ad journey. You can exclude underperforming terms or URLs from there. Google Ads API docs (last updated September 10, 2026) already expose ai_max_search_term_ad_combination_view with search_term, headline, and landing_page as the granular recipe for auditing AI-driven matches. Treat September 23 as Google elevating that journey narrative in the product story, not as permission to invent a report that Help and the API never described.
Fact three: availability is not the same as existence. The September 23 blog ends with “We’ll share additional details and availability later this year.” Schwartz’s Roundtable write-up repeats that cadence. So do not promise your CFO a brand-new UI tile on every MCC tomorrow. Do promise that the operator discipline Google is selling (brief in, combination out) is already partially instrumented in Help Center views and in the API combination view. Teams that wait for a conference slide to start writing briefs and mining combinations will be late twice.
Bright line against the migration piece
Monday’s thesis: if you never clicked AI Max, September may still have changed your Search settings through progressive auto-migration of campaign-level broad match and standalone ACA cohorts, with DSA parked until February 2027. The deliverable was a settings audit with migration timestamps.
Today’s thesis: once the umbrella is on (or once you turned it on yourselves), the control surface for delivery is the brief you write and the journey combinations you accept or kill. Toggles still matter. They are not the weekly craft.
PPC Land’s June 23, 2026 documentation rewrite already flagged the industry’s reporting friction: four search-terms views, a match-type filter that undercounts AI Max when “Other search terms” are ignored, aggregate keyword rows for expanded matches and landing-page matches, and a long runway of practitioner complaints about query-to-page routing. The September 23 story does not erase that history. It reframes the product answer: stop pretending a flat search-term export is enough, and start managing the combination.
What operators get wrong
Mistake one: mining search terms like it is still 2019. A query list without headlines and landing pages cannot tell you whether AI Max failed on intent, on creative, or on routing. Help Center is blunt that the combination view exists to assess headline and landing-page relevance to search terms. API docs go further: mismatch between search_term and headline is a signal to feed better assets, and bad landing_page rows are often a URL exclusion problem, not a negative-keyword spiral.
Mistake two: treating AI Brief like a brand book paste. Messaging Guidelines are not a PDF dump. Matching Guidelines are not a keyword spreadsheet rewritten as prose. Audience Guidelines are not a CRM segment name. Ervin’s examples are short, prohibitive, and testable. If your brief reads like a homepage hero paragraph, you gave Gemini mush and then blamed the model when previews look generic.
Mistake three: filtering Match type = AI Max and calling it the footprint. Help Center still warns that this filter can show lower numbers because it does not consider “Other search terms.” If your weekly readout uses that filter as the denominator for “how much AI Max traffic we have,” you are understating the system you claim to govern.
Mistake four: summing combination-view metrics with search_term_view totals. Google Ads API docs are explicit: do not reconcile those views against each other, do not sum them, and do not use the combination view as a financial reconciliation tool. Privacy thresholds hide low-volume terms from search-term reports while clicks still sit in campaign totals. Double counting and undercounting are both failure modes. One source of truth per question.
Mistake five: confusing exclusions. API guidance is clean enough to tattoo on an agency wiki. Bad query intent gets a negative keyword. Bad destination gets a URL exclusion. Brand cannibalization gets a brand list exclusion. Whack-a-mole negatives for every query that bounced on one outdated blog post is how you invent work while the real leak stays live.
Mistake six: staffing Brief in English only while spend is not. Language expansion is not a localization vanity project. If French, German, Japanese, or Spanish Search is material for the brand, and Brief beta is available in those languages, leaving Matching and Messaging Guidelines English-only while the account spends elsewhere is a control gap with a date on it: September 23, 2026.
Mistake seven: reopening Monday’s migration ticket under a new subject line. DSA February 2027, cohort defaults, and aca_migration_date_time / broad_match_migration_date_time still matter. They are not this essay’s job. If your Slack thread about “journey reporting” turns into another argument about whether AI Max should exist, you lost the week.
Second-order effects for media and agencies
Creative and SEM stop being sequential handoffs. Text customization marks Google AI assets in the Added by column. Final URL expansion picks pages. The combination view makes those choices visible as recipes. That means the person who owns RSA assets and the person who owns Search negatives need the same weekly artifact: top converting and top wasting term-headline-URL triples, plus Brief revisions that would have prevented the waste.
Measurement politics get sharper. Leadership still asks for “AI Max vs keyword” as if those are clean partitions. Source columns and search_term_match_source values (AI_MAX_KEYWORDLESS, AI_MAX_BROAD_MATCH) help, but privacy thresholds and view overlap mean your board slide needs methodology footnotes. Agencies that cannot explain why campaign clicks exceed visible search-term clicks will keep losing trust to whoever pretends the numbers reconcile.
International GTM changes the Brief backlog. Local market owners become Brief authors, not just translators of English negatives. That is a staffing model change for holding-company Search pods that still centralize copy in one language and treat markets as bid modifiers.
Tooling follows the API. If your BI layer only pulls classic search_term_view, you cannot operationalize the product Google is selling. Pull ai_max_search_term_ad_combination_view for recipe audits. Keep search_term_view for match-source breakdowns. Never glue them into one vanity ROAS.
Regulated categories get a compliance path that is still not a free pass. April’s text disclaimers story (mandatory text even with Final URL expansion) sits beside Brief Messaging Guidelines. Journey reporting makes noncompliant combinations easier to spot. It does not make them legal.
What to do this week
Run this as a delivery control sprint, not a philosophy seminar.
- Inventory Brief eligibility by language. List Search campaigns with meaningful spend in Dutch, French, German, Italian, Japanese, Portuguese, Spanish, and English. Note which already have AI Brief access, which still sit on legacy text guidelines, and who owns the rewrite. Do not wait for “additional details later this year” on reporting before writing Messaging / Matching / Audience Guidelines where beta is live.
- Stand up a journey scorecard. In Google Ads, open Search terms and ad combinations for every AI Max Search campaign that has at least two weeks of post-enable learning (Help Center’s own ramp guidance). Export or API-pull the top 50 spend and top 50 conversion recipes: search term, headline concatenation, landing page. Label each row Intent OK / Creative miss / URL miss. That label is the ticket type for exclusions and Brief edits.
- Separate the three exclusion queues. From the same scorecard, file negatives only for bad intent, URL exclusions for bad destinations (including directory rules like blog or careers), and brand list fixes for cannibalization. Kill the habit of solving page problems with keyword negatives.
- Rewrite one Brief per priority market, short. Three blocks, each under 80 words: what we never say, what searches we prioritize or refuse, who gets which proof. Use Ervin’s shape, not a brand manifesto. Preview, reject, iterate before apply. If English Briefs already exist, port the constraints into the new languages instead of inventing seven unrelated stories.
- Fix the readout. Stop presenting Match type = AI Max filter totals as the AI Max volume number. Add Source / match-source methodology. Document that combination-view metrics are for recipe QA, not ledger reconciliation. Put migration dates beside September performance only when someone tries to blame journey waste on the auto-upgrade window. Different tickets.
- Assign owners. Brief author (market or category), combination auditor (SEM), asset killer (creative/compliance), URL governor (web/analytics). One Slack channel that posts the weekly triple list beats four teams arguing from four exports.
The operator close
Google is not asking you to love AI Max this week. It is telling you how it wants to be managed: write the brief in the languages you spend in, then judge delivery as a journey from query through creative to page.
Search terms still matter. They are an input row, not the full story. Settings still matter. They are the floor, not the craft. The control surface is the brief plus the combination. Teams that keep arguing about the toggle while shipping unbriefed journeys will look busy and still lose the auction for the right reason.
Ship the scorecard. Write the Brief. Treat every bad term-headline-URL triple as a decision, not a surprise.
