AI Max Reporting in Google Ads 2026: The Unified Journey View for Search Terms, Creatives and Landing Pages

AI Max reporting is the thing most search advertisers have to relearn this autumn. Between 1 and 30 September 2026, Google auto-migrated a large share of eligible Search campaigns to AI Max. After that, your account contains traffic with no keyword behind it at all: the system matched the query itself, assembled the headline itself and picked the landing page itself. The classic search terms report still shows those queries, but it cannot explain them — you see the term and nothing about which creative ran or where the click actually landed.

Alongside the migration, Google shipped new AI Max reporting, including a unified journey view that puts the query, the creative and the landing page in a single row. Below: exactly what shipped, where to find it in the UI and the API, how to read the new match-source segment, which arithmetic mistake silently doubles your numbers, and the order in which to review an account that has just come out of auto-migration.

The 2026 timeline, briefly

The shape of today’s reports makes more sense when you have the sequence in mind.

Date	What happened
3 August 2026	Creation freeze for campaign-level broad match and legacy auto-created-asset campaigns
9 August 2026	AI Max became eligible for exact and phrase match keywords in standard Search campaigns
1–30 September 2026	Automatic migration of eligible Search campaigns
23 September 2026	Unified AI Max reporting: search term, creative and landing page in one view
February 2027	Separate wave: Dynamic Search Ads migration

Whether a given campaign was migrated is exposed as campaign.ai_max_setting.enable_ai_max. Check that first, before you start explaining a change in performance to yourself or to a client. The migration itself is covered in AI Max for Search: what the auto-migration actually changes, and the February wave in the DSA to AI Max migration playbook.

What the unified journey view actually contains

The headline addition is a view called ai_max_search_term_ad_combination_view. It joins three things that used to live in separate reports:

• the search term — what the person actually typed;
• the ad combination — which headline was assembled and served for that term;
• the landing page — the URL that was really delivered after final URL expansion.

Before September, stitching those three together was a manual join that most teams got wrong. Now a single row answers the whole question: “for query X we served headline Y and sent the click to page Z.” This is the first time Google exposes the query → creative → landing page chain natively inside Search reporting.

The practical value shows up fastest on large catalogues. A poor-converting query used to look like a query problem. Now you can see that the query was fine and the system routed the visitor to a category page instead of the product page — which means the fix is page structure, not negative keywords. That is exactly the diagnosis path described in the Landing Pages report in Google Ads 2026.

The match-source segment is the key to everything

The second change is a new segment, segments.search_term_match_source, with three values:

Value	What it means	How to read it
ADVERTISER_PROVIDED_KEYWORD	The query matched one of your keywords	Classic managed traffic. Your baseline for comparison
AI_MAX_BROAD_MATCH	AI Max expanded one of your keywords	Adjacent demand. Check relevance and margin
AI_MAX_KEYWORDLESS	Matched with no keyword involved at all	The newest and least predictable layer. Review it first

That third value is genuinely new. It is the first mainstream report that openly labels traffic where no keyword participated: the match came from your page content, your assets and demand signals. Until you split those three sources apart, any conclusion about post-migration performance is an average of things that are not comparable.

Where it sits in the interface

The changes are scattered across several reports, so here is the full list:

• Search terms report — a new “Search terms and landing pages from AI Max” view, an “AI Max” value in the match type filter, and a “Source” column.
• Keywords report — aggregate rows for “Total: AI Max expanded matches” and “Total: AI Max landing page matches”.
• Landing pages report — a “Selected by” column showing whether you or the system chose the page.
• Asset report — an “Expanded final URL assets” tab.

If you work from exports rather than the UI, you need four API objects: the combination view and the match-source segment above, plus expanded_landing_page_view with its expanded_final_url field and the campaign-level AI Max flag. Getting all of that into your own warehouse without weekly manual pulls is covered in Google Ads to BigQuery in 2026.

The trap: never sum the two reports

The most expensive mistake in the new reporting is arithmetic. The unified journey view overlaps the classic search terms report — many terms appear in both. Google says plainly that you must never sum metrics across the two views, because doing so inflates clicks, cost and conversions.

The unified journey view is not an extension of the search terms report. It is a different cut of the same events. Pick one source of truth per question and keep them apart.

A workable rule: the search terms report answers “where did the money go by query”; the journey view answers “why did it end up that way”. Use the first for a stakeholder dashboard and the second for diagnosis. If you are building a warehouse, keep them in separate tables and forbid yourself from joining on the term alone without an explicit combination key. The wider problem of contradictory numbers across reports is covered in media buyer dashboards and reporting.

expanded_final_url: the quiet break in your analytics

Final URL expansion can serve a page other than the one configured on the ad. For performance that is often a win — the system finds a better-matching page. For analytics it is a source of discrepancies that teams chase for weeks.

What breaks when you keep relying on the configured URL:

• UTM parsing. Your tags live on the configured URL; the delivered one may differ, especially when expansion routes to a page without your parameters.
• Page grouping. Rules like “anything under /collections/ is a category” stop matching reality.
• Joins to the CMS or product database. Product-level profitability reports silently drop rows or attribute revenue to the wrong page.

There is one fix: read expanded_final_url from the new view and build every join on it, keeping the configured URL as a reference field only. If expansion is routing traffic somewhere you do not want, that behaviour is controllable — see final URL expansion in Performance Max and AI Max.

A review order for a freshly migrated account

The sequence below takes roughly two hours on a mid-sized account and answers the only question that matters: did the migration help or hurt?

1. Record what migrated. Export the AI Max flag for every campaign and note each migration date. Without the date, period-over-period comparison is meaningless.
2. Split traffic by match source. Take 30 days post-migration and look at the share of cost across the three values. A keywordless share above a third, combined with rising CPA, means you need a targeted review rather than a blanket change.
3. Compare unit economics per layer. Do not compare CPA; compare each layer against your allowable CPA from margin. Expanded demand often produces a cheap click and an expensive sale. The method is in unit economics and allowable CPA from LTV.
4. Open the journey view on your worst terms. For each problem query, look at the assembled headline and the delivered page. In about half of cases the query is not the problem.
5. Audit pages via the “Selected by” column. If the system systematically routes from product pages to categories, constrain expansion rather than cutting queries.
6. Add negatives by theme, not by source. Blocking the entire keywordless layer is the common panic reaction and it cancels the point of the migration. The careful version is in negative keywords and search terms.
7. Pin the new fields into your reporting. Add match source and delivered URL as custom columns so you are not re-exporting every week — mechanics in custom columns and the Report Editor.

How to judge the keywordless layer honestly

The methodological problem is simple: keywordless traffic cannot be held to the thresholds you use for brand or exact-match search. It is broader, colder and slower by construction.

Three habits that produce an honest read:

• Compare the layer to itself over time, not to keyword traffic. The question is not “is keywordless worse than exact match” (it almost always is) but “is keywordless better than it was four weeks ago”.
• Look at the new-customer share inside the layer. If expansion mostly brings back people who already buy from you, you are paying for your own demand. Customer type labelling is covered in customer type labelling and lifecycle goals.
• Never evaluate over a window shorter than your conversion window. Judging a long-cycle layer on seven days guarantees you understate it.

As a reference point rather than a rule: on broad-catalogue accounts the keywordless share usually settles somewhere between 15% and 35% of spend. Much higher and your assets and page copy are probably too generic; much lower and expansion is being constrained by your settings — which may well be a deliberate choice rather than a fault.

AI Max reporting versus classic Search reporting: the deeper shift

Old reporting was built around the keyword. The keyword was the unit of control, the unit of analysis and the unit of reporting all at once. AI Max reporting is built around the match — the event in which the system paired a query, a creative and a page. The keyword is now just one possible source of that match, and not necessarily the dominant one.

Three consequences change the daily job:

• Control shifts from bids to inputs. The response to a bad query used to be a bid change or a negative. Now it is more often a change to what the system builds matches from: assets, page copy, catalogue structure, expansion exclusions.
• Reports stop being flat. One query row can carry several creative combinations and several landing pages. The total by query and the total by combination are different numbers.
• Page quality became a targeting factor. Your landing page content now directly influences which queries you appear for. Thin or outdated pages no longer just convert badly — they pull irrelevant traffic in.

That last point is the real shift of the year. Site quality used to affect Quality Score and conversion rate. It now affects reach as well. If your site still hosts forgotten pages with expired promotions or discontinued items, the system may well match them to a query, and you will pay for a click to a page you abandoned a year ago. Landing page housekeeping moves from “eventually” to standing maintenance, at the same cadence as speed and layout checks — see landing page speed and Core Web Vitals.

Three post-migration scenarios and what they mean

Almost every account lands in one of three patterns. Identifying yours saves days of guessing.

Scenario	What the numbers look like	Usual cause	First action
Volume up, CPA up more	Clicks +30–60%, conversions +10–20%, CPA +25% or worse	Expansion pulled in cold demand that does not clear your margin	Split by match source, constrain expansion on unprofitable themes
Volume flat, traffic mix changed	Same spend, but 20–30% now keywordless	The system redistributed the same budget across different queries	Compare lead or order quality per layer, not just CPA
Volume down	Impressions and clicks below baseline	Usually constraints: thin assets, aggressive exclusions, weak page content	Review assets and exclusion lists, then loosen one thing at a time

The classic error lives in the third scenario: teams respond to a volume drop by raising budgets, when the cause is that the system has nothing to build matches from. With three headlines and one landing page, no amount of budget creates demand. The correct order is material first — assets, pages, feed — money second. Asset logic is covered in ad extensions and assets in Google Ads.

Fitting the new data into a weekly routine

To stop this becoming a one-off curiosity, give it a rhythm. Forty minutes a week is enough:

1. 10 minutes. Cost and conversions by the three match sources, week over week. Note direction only, draw no conclusions yet.
2. 15 minutes. Top 20 new keywordless terms: open the journey view and read the headline-plus-page pair. Decide per row: keep, fix the page, or add a thematic negative.
3. 10 minutes. Check “Selected by” in the landing pages report for systematic routing away from product pages.
4. 5 minutes. Log what you changed. Without a log you will not be able to connect next month’s swing to this week’s edit — see change history and diagnosing performance drops.

What the reporting still does not give you

• It is not per-combination attribution. You get the pairing and its metrics, not an isolated measure of a creative’s contribution.
• Low-volume rows may be withheld. Reporting thresholds still apply, so part of the long tail stays invisible.
• There is no combination-level control. You can see what the system assembled but you cannot kill a specific query-plus-headline pair; control stays at the level of assets, rules and exclusions.
• History is shallow. The new views do not backfill deeply, so if you did not capture a baseline before migration, lean on aggregates from the classic reports.

Pre-Q4 checklist

1. Migrated campaigns exported with their migration dates.
2. Match source and delivered URL added to your reporting.
3. Journey view and search terms report separated in the warehouse, never summed.
4. Product-database joins moved to expanded_final_url.
5. A distinct evaluation threshold and window set for the keywordless layer.
6. Negatives collected by theme, not by match source.
7. Expansion settings checked against seasonal and sold-out pages.

If you run this review across dozens of campaigns, the access and structure layer matters as much as the analysis — see the Google Ads manager account (MCC) guide. And if you are scaling buying and need working accounts plus support behind them, look at Google Ads agency account rental and the rest of the PPC Rebels services.

FAQ

What is the unified journey view in AI Max?

A report that shows the search term, the assembled ad combination and the delivered landing page in a single row. In the API it is ai_max_search_term_ad_combination_view, released on 23 September 2026.

Where do I find AI Max reporting in the Google Ads interface?

The search terms report now has a “Search terms and landing pages from AI Max” view with an “AI Max” match type filter and a “Source” column. Also check the “Selected by” column in the landing pages report and the expanded final URL assets tab.

Can I add metrics from the journey view and the search terms report together?

No. The views overlap, many terms appear in both, and summing them inflates clicks, cost and conversions. Choose one source per question.

What does AI_MAX_KEYWORDLESS mean?

That the match happened with no keyword involved: the system used your page content, assets and demand signals. It is the newest traffic layer and needs its own evaluation thresholds.

Why does the report show a different URL than the one on my ad?

Final URL expansion can serve a more relevant page. The delivered address sits in expanded_final_url, and that is the field your joins and page groupings should use.

How do I tell whether a campaign was migrated?

Check campaign.ai_max_setting.enable_ai_max at campaign level, or the AI Max indicator in campaign settings. Record the migration date separately — comparisons are unreliable without it.

Should I block all keywordless traffic?

No — that removes the point of the migration. Add negatives for specific irrelevant themes instead. If the layer is structurally unprofitable, manage it through expansion settings rather than a wall of negatives.

What keywordless share is normal?

There is no official benchmark. As a reference, broad-catalogue accounts tend to settle between 15% and 35% of spend. Outliers in either direction are a prompt to check assets and expansion settings, not a crisis.

Is there historical data from before the migration?

The new views do not carry deep history. If you did not capture a baseline, compare aggregates from the classic reports and focus on the trend within each layer.

When do Dynamic Search Ads migrate?

A separate wave is scheduled for February 2027; it was not part of the September window. Preparing structure and feeds before then is worthwhile.

CPA rose after migration but volume is flat — what first?

Split spend by match source, compare each layer against allowable CPA from margin, and only then touch bids. The problem is usually inside one layer rather than across the campaign.

Do I need to rebuild dashboards because of the new fields?

Minimally, yes: add match source and delivered URL, keep the two reports in separate tables, and move product joins to the delivered URL. Otherwise discrepancies accumulate without anyone noticing.