Przejdź do treści
whitead

AI Max for Search

słownik google ads zaktualizowano 2026.05.06 6 min czytania
// contents · 9 sekcji
  1. What AI Max for Search is
  2. How AI Max works under the hood
  3. What changed in 2025-26
  4. Common misconceptions
  5. Is AI Max the same as Performance Max?
  6. Should you turn AI Max on for every Search campaign?
  7. Does AI Max replace responsive search ads?
  8. Related concepts
  9. Sources

AI Max for Search is an opt-in setting on existing Google Ads Search campaigns that layers Gemini-powered query, asset, and URL matching on top of your keyword list. Announced at Google Marketing Live in May 2025 and rolled to general availability in late 2025, it is a feature toggle — not a new campaign type (Google Ads Help, 2025-10-21).

What AI Max for Search is

AI Max for Search is a campaign-level setting that lets Google's matching engine expand beyond the literal keyword and ad-asset definitions you supplied. With it enabled, the system can match your ads to relevant queries that your keyword list would not normally cover, generate or rewrite headlines and descriptions using your final URL and existing assets as input, and bias toward (or away from) specific landing pages via URL inclusions and exclusions (Google Ads Help, 2025-10-21). It is opt-in per campaign and is positioned as the Search counterpart to the AI-driven matching that already powered PMax. It is not a new campaign type; your existing keywords, negatives, audiences, bid strategy, and budget all carry over.

How AI Max works under the hood

The toggle bundles three sub-controls that you can tune independently after enabling:

  • Keyword + URL matching expansion — broadens which queries your ads compete on, using the campaign's keywords, landing pages, and assets as intent signals. Google describes this as a generalisation of broad match plus URL-driven matching from DSA (Google Ads Help, 2025-10-21).
  • Text customization — Gemini generates additional headlines and descriptions per query, anchored to your existing RSA assets and final URL. Pinned assets remain pinned, and brand-restriction settings still apply.
  • Final URL inclusions/exclusions — a URL allowlist/blocklist that scopes where the engine may route traffic. Without exclusions, AI Max can send a query to any indexable page on your domain, including blog or legal pages.

Reporting surfaces the new traffic in the search-terms report and a "search terms insights" panel; AI Max-generated headlines appear in the asset details view with a system-generated label (Search Engine Land, 2026-01-27).

What changed in 2025-26

AI Max moved from open beta (announced May 6, 2025) to general availability on April 15, 2026, after roughly eleven months of global testing (Google blog, 2026-04-30). Google's original launch communications claimed advertisers who turned it on saw "an average of 14% more conversions or conversion value at a similar CPA or ROAS" — a number Google quietly revised down to ~7% in the GA announcement (Google blog, 2026-04-30).

Independent operators tell a more uneven story. PPC Land's August 2025 rollout coverage cited early agency tests where Monks found "99% of impressions have 0 conversions" across ~30,000 AI Max-matched search terms, alongside Google-published wins like L'Oréal Chile's 2× higher conversion rate at 31% lower CPA (PPC Land, 2025-08-17). A broader November 2025 analysis of 250+ retail campaigns by Smarter Ecommerce found AI Max delivered conversions at roughly 35% lower ROAS than exact/phrase/broad match, with cost-per-conversion of $100.37 versus $43.97 on phrase match in one four-month head-to-head (PPC Land, 2025-11-08). Search Engine Land's pre-test checklist independently warned that query expansion treats every keyword as broad match and then expands further, so accounts without disciplined conversion tracking and proven broad-match success tend to underperform (Search Engine Land, 2026-01-27).

The honest read: Google's headline lift is a real average across opt-in advertisers, but the distribution is wide and the floor is well below zero for accounts without disciplined negatives or brand guardrails. Treat the headline number as a directional ceiling, not a guaranteed lift.

Methodology note. Whitead's audit engine flags AI Max-related risk through three signals on opted-in Search campaigns: (1) negative-keyword count below a per-campaign floor scaled to monthly impressions, (2) absence of brand-as-negative on non-brand campaigns, and (3) missing URL exclusions on accounts with public blog or legal subdirectories. When any two of those signals fire together on an AI Max-enabled campaign, the rule escalates to HIGH severity because query expansion drift and off-brand asset generation become measurable rather than theoretical risks. We do not assert frequency claims pending sample-size disclosure in a later wave.

Common misconceptions

The bundling tempts operators to flip the toggle and walk away. The three sub-controls behave very differently in practice — keyword expansion alone often lifts volume cleanly, while text customization and URL exclusions are the levers that determine whether the lift is on-brand and on-page.

Is AI Max the same as Performance Max?

No. PMax is a separate campaign type that runs across Search, Display, YouTube, Discover, Gmail, and Maps with no keyword targeting. AI Max keeps your keyword-targeted Search campaign intact and only expands matching within the Search inventory. You still see the search-terms report, you still control negatives at the keyword level, and your bid strategy continues to optimise against your existing conversion goals (Google Ads Help, 2025-10-21).

Should you turn AI Max on for every Search campaign?

Not without prep. The accounts where AI Max underperforms in third-party reports share two traits: thin negative-keyword lists and no brand exclusion on non-brand campaigns. Audit those first. Do not enable AI Max on a brand-only campaign — the expansion will pull it into generic queries that belong to a different campaign with a different bid strategy. Run it on generic Search campaigns where you already trust broad match, and add brand terms as negatives before flipping the toggle (Search Engine Land, 2026-01-27).

Does AI Max replace responsive search ads?

No. Your RSA assets remain the foundation. Text customization generates additional headline and description variants per query, but pinned assets stay pinned and your existing ad strength scoring still applies. If your RSAs are weak (poor ad strength, missing pinned brand headline) the customization layer amplifies the weakness rather than fixing it.

AI Max sits between Smart Bidding (which optimises bids on the queries AI Max surfaces) and Negative Keywords (which scope what AI Max may match). Pair it with Search Query Mining as a recurring workflow once enabled, and review the Smart Bidding Learning Phase thresholds before flipping the toggle on a low-volume campaign — query expansion plus a re-entered learning phase is a worse combination than either alone.

Sources

// czy to było pomocne?
// anonimowo · nie zapisujemy danych osobowych
← Wróć do Wiki