AI Max for Search

AI Max for Search is an optional suite of AI-powered targeting and creative optimisation features that can be enabled within new or existing Google Search campaigns. It is not a new campaign type — existing campaigns can have it toggled on without rebuilding. Announced at Google Marketing Live on 21 May 2025, it entered open beta on 27 May 2025 and reached general availability (GA) on 15 April 2026. (Google Ads Developer Blog, 2025-05-27; blog.google, 2026)

Three components

AI Max comprises three independently toggleable features; activating all three yields the highest automation level but each can be enabled alone. (Google Ads Help, support.google.com/google-ads/answer/15910187, as-of 2026-07)

Feature	What it does
Search term matching	Uses Broad Match and keywordless technology (LLM-based intent modelling from keywords, creatives, and URLs) to match ads to queries beyond the existing keyword list
Text customisation	Dynamically generates headlines and descriptions per query using generative AI
Final URL expansion	Routes users to the most query-relevant landing page on the domain, scoped to pages related to the ad group's theme

Text customisation must be enabled for Final URL expansion to activate; the two features are interdependent. (Google Ads Help, support.google.com/google-ads/answer/15910187, as-of 2026-07)

Priority and controls

When an existing keyword matches a query, that keyword takes priority over AI Max expansion — AI Max is designed to surface net-new queries not captured by existing keyword lists. (Search Engine Land, searchengineland.com/ai-max-for-search-everything-you-need-to-know-462923, 2025–2026)

Available controls:

Negative keywords — applied before AI expansion logic; a query matching a negative is excluded regardless of semantic match. (Google Ads Help, support.google.com/google-ads/answer/15913066, as-of 2026-07)
Brand exclusions — prevent AI Max from serving against competitor brand queries or cannibalising the advertiser's own branded Search campaigns. (Google Ads Help, as-of 2026-07)
URL exclusions — block specific pages or URL patterns (exact or URL_CONTAINS) from Final URL expansion. (Google Ads Help, as-of 2026-07)
Search Partner Network — can be disabled at campaign level. (Google Ads Help, as-of 2026-07)
Smart Bidding required — AI Max does not function with manual CPC; requires Target CPA, Target ROAS, Maximize Conversions, or Maximize Conversion Value. (Google Ads Help, as-of 2026-07)

Search Themes

When keywordless matching is active, advertisers can add Search Themes as intent signals to guide AI Max targeting — these are not hard keyword constraints but nudges toward relevant query territory. (Aaron Young / Define Digital Academy, youtube.com/watch?v=carICkG7reE, 2025-06-02)

Performance benchmarks

Google's official claims (vendor-reported — treat with caution)

Advertisers activating the full AI Max feature suite typically see 14% more conversions or conversion value at a similar CPA/ROAS compared to using search term matching alone. (as-of 2025-05, Google Ads Help, support.google.com/google-ads/answer/15910366)
For campaigns still predominantly using exact and phrase match keywords, the stated uplift is 27% more conversions at similar CPA/ROAS. (as-of 2025-05, Google Ads Help)
At Google Marketing Live 2025 cherry-picked case studies cited conversion increases of 263% and CPA reductions of 73%. (as-of 2025-05-21, blog.google/products/ads-commerce/google-marketing-live-2025/ — selective examples, not median outcomes)

Google's efficiency claim vs SMEC's independent study. Google's official benchmark states 14% more conversions "at a similar CPA/ROAS" — implying efficiency is maintained. An independent Smarter Ecommerce (SMEC) study of 250+ retail campaigns found median revenue uplift of +13% but median CPA increased +16%; ROAS varied from +42% to -35% across campaigns; only 22% of campaigns came close to their original ROAS targets. (as-of 2026, Smarter Ecommerce via Search Engine Land, searchengineland.com/google-ai-max-revenue-higher-cpa-study-470928 and searchengineland.com/google-ai-max-performance-tests-471366) Google's metric tracks conversion volume; SMEC's data tracks cost efficiency — they are not equivalent, and the framing gap between them is significant. Note SMEC sells campaign management software (mild conflict of interest).

A poll of PPC professionals found only 16% saw good performance with AI Max; 84% reported neutral or negative outcomes. (as-of 2026, PPC Live, ppc.live/library/google/ai-max/googles-ai-max-for-search-what-the-data-actually-shows-in-2026/ — self-selected sample, unknown n)

Structural risks

Four structural risks identified from SMEC's analysis of 1M+ AI Max impressions across 250+ retail campaigns. (as-of 2026, smarter-ecommerce.com via searchengineland.com)

1. Broad match cannibalization — occurring in up to 63% of campaigns; AI Max serves queries already covered by existing keywords with no net-new value. (as-of 2026)

2. Competitor brand hijacking — one case study showed competitor brand queries consuming 69% of Search impressions share. (as-of 2026)

3. Reporting overload — search term reports can reach tens of thousands of rows, making manual review impractical. (as-of 2026)

4. Search Partner Network underperformance — one campaign generated 500,000 monthly SPN impressions at 0.07% conversion rate vs 3.04% on Google Search. (as-of 2026)

What drives AI Max expansion? General practitioner commentary assumes AI Max expansion is primarily Broad Match-driven. SMEC's impression-level analysis of 1M+ impressions found ~80% of expansion originates from Exact Match keywords, ~20% from Phrase Match, and less than 1% from Broad Match — contradicting the dominant mental model. (as-of 2026, Smarter Ecommerce via Search Engine Land, searchengineland.com/google-ai-max-revenue-higher-cpa-study-470928)

Relationship to adjacent concepts

vs Performance Max (PMax)

Dimension	AI Max for Search	Performance Max
Inventory	Search only (+ optional Search Partners)	All Google inventory (Search, Display, YouTube, Gmail, Discovery)
Keyword controls	Retained (negative keywords, brand exclusions)	Not available
Query transparency	Full: query + headline + landing page per row	Limited
URL expansion scope	Ad group theme-restricted	Full domain

AI Max is characterised by practitioners as a "middle path" — less control than standard keyword-targeted Search, more control than PMax. (Seer Interactive, seerinteractive.com/insights/from-pmax-to-ai-max-what-to-consider-before-testing, 2026)

vs Broad Match

AI Max's keywordless targeting is broader than Broad Match alone: it uses intent modelling informed by landing page content and creative assets, not just keyword-to-query semantic similarity. Broad Match is a match type; AI Max is a feature layer that includes broad-match-style intent modelling plus dynamic creative and URL expansion. Practitioners recommend pausing manually-set Broad Match keywords when enabling AI Max search term matching to avoid duplication — though SMEC data suggests cannibalization persists even then. (WordStream, wordstream.com/blog/ai-max-vs-broad-match, 2026)

vs RLSA

RLSA lists can be layered on AI Max campaigns for bid adjustments. The precise interaction between RLSA audience signals and AI Max's keywordless expansion is not fully documented in official Google sources — available only via practitioner observation. (Seer Interactive, 2026)

vs Dynamic Search Ads

DSA campaigns are being deprecated and auto-upgraded to AI Max. See Migration timeline below.

AI Overviews eligibility

AI Max-enabled Search campaigns are eligible to serve in Google's AI Overviews (AI-generated search result summaries) — a placement not available to standard keyword-targeted Search campaigns. Google cited AI Overviews as a primary driver for why AI Max's keywordless expansion matters for future traffic capture. (AdExchanger + blog.google/products/ads-commerce/ai-max-new-features/, as-of 2026)

2026 new features (post-GA)

AI Brief — Gemini-powered feature letting advertisers provide natural-language descriptions of business, messaging priorities, and target audience to steer AI creative generation. (blog.google/products/ads-commerce/ai-max-new-features/, as-of 2026)
Text guidelines — natural-language instructions to shape AI-generated ad copy and keep messaging on-brand; expanded globally across AI Max and Performance Max (PMax) in 2026. (Search Engine Land, searchengineland.com/google-expands-ai-max-text-guidelines-globally-470294, as-of 2026)
Network-level placement reporting — segmentable by Google Search vs Search Partners, available from approximately April 2026. (as-of 2026-04, volatile)
AI Max for Google Shopping — extension to Shopping campaigns announced and in rollout as of 2026. (blog.google/products/ads-commerce/ai-max-for-shopping/, as-of 2026)

Migration timeline (DSA → AI Max)

Google announced Dynamic Search Ads will be automatically upgraded to AI Max:

Original deadline: September 2026 (for campaigns using ACA, campaign-level broad match, or DSA settings)
Revised deadline: February 2027 — delayed on 11 June 2026 to avoid disrupting advertisers' Q4 planning; new DSA campaign creation restored from 15 June 2026. (Search Engine Land, searchengineland.com/google-delays-dynamic-search-ads-migration-to-ai-max-480049, as-of 2026-06-11)
Google published voluntary migration tools and recommended running AI Max experiments ahead of the February 2027 deadline. (blog.google/products/ads-commerce/dsa-upgrade-to-ai-max-2026/, as-of 2026)

One Google Business Accelerate page still referenced September 2026 as of July 2026; the June 2026 Search Engine Land delay announcement supersedes it. Verify current deadline against official Google Ads Help. (business.google.com/us/accelerate/announcements/dsa-upgrade-to-ai-max-2026/)

Naming collision

Microsoft separately launched its own "AI Max" for Search — expanding query matching across Copilot and Bing surfaces. This is a distinct Microsoft product with no relationship to Google's AI Max for Search. (Search Engine Land, searchengineland.com/microsoft-launches-ai-max-and-new-ad-tools-for-the-agentic-web-era-474939, as-of 2026)

Practitioner approach

Practitioners recommend:

Run AI Max as a built-in Google Ads Experiment (50/50 split, minimum 4 weeks) before full rollout. (Google Ads Help, support.google.com/google-ads/answer/16450159)
Wait at least 2 weeks after enabling before adding negative keywords, to allow the learning period. (Google Ads Help, support.google.com/google-ads/answer/15909989)
Layer automation one component at a time — enable one feature, measure over 2–4 weeks, then add the next. (Jyll Saskin Gales / Lunio Paid Media Lab, lunio.ai/podcasts/ai-max-in-google-ads, 2025-07-02)
Disable the Search Partner Network if SPN conversion rate diverges significantly from Google Search rate. (SMEC study)
Add brand exclusions before enabling to prevent competitor brand query spillover.

Key terms

Term	Meaning
Keywordless targeting	AI Max's ability to match ads to queries without an explicit keyword in the account
Search Themes	Intent signals that can be added to guide AI Max keywordless matching direction
Final URL expansion	AI Max feature that dynamically selects the most relevant landing page
AI Brief	Gemini-powered tool for providing natural-language guidance to AI creative generation
Text guidelines	Natural-language instructions constraining AI-generated ad copy tone/messaging
SPN	Search Partner Network — third-party sites that show Google Search ads
DSA	Dynamic Search Ads — being deprecated, auto-upgraded to AI Max by February 2027

Benchmarks (as-of 2026-07)

Metric	Google's claim	SMEC study (250+ retail campaigns)
Conversion volume	+14% at similar CPA/ROAS	+13% revenue
Cost efficiency	Similar CPA maintained	CPA +16% (worse)
ROAS range	Not disclosed	+42% to -35%
Campaigns hitting ROAS target	Not disclosed	22%

Google's figures are self-reported vendor benchmarks. SMEC is an independent agency with mild conflict of interest (sells campaign software). Both figures should be treated as indicative, not definitive.