Google Ads is testing a campaign-creation screen that asks "What's your goal for this campaign?" and lets advertisers answer in a free-text box, after which Google AI suggests goals and assets. Google's Ads Liaison has confirmed it is an experiment, and availability, supported campaign types and level of control are not yet public. Use it, if you see it, as a drafting aid you review line by line, not as an autopilot. As of 14 October 2026.
Key Takeaways
- The test was spotted by an advertiser and reported by Search Engine Land and Search Engine Roundtable on 12 October 2026; Ginny Marvin, Google's Ads Liaison, said "this is an experiment."
- There is a "Create campaign manually" option, so classic setup is still available in the screenshots seen.
- Unknowns: which advertisers get it, which campaign types it supports, and how much control you keep over suggestions.
- It sits on top of automation you may already be running, especially AI Max, which has documented controls and cautions worth knowing first.
- Safe adoption means a written brief, an approval checklist, a budget cap, and a two-week review rhythm.
- Do not let AI-suggested assets go live without checking claims, prices, and regulated-category wording.
What has been confirmed
According to Search Engine Land's report by Anu Adegbola, Google Ads is experimenting with a flow where, rather than starting from predefined objectives, the advertiser describes the goal in their own words. A screenshot shows a large text box and the line "Google AI will suggest goals and assets based on your input." Search Engine Roundtable's coverage credits an advertiser who shared the screenshot on X and quotes Ginny Marvin's reply on X: "this is an experiment. We welcome advertisers' feedback around how and where this type of workflow would be helpful."
That is the entire confirmed surface. Search Engine Land says Google has not shared availability or timing. I saw no official Google Ads Help page or blog post describing the feature as of this date, so everything beyond the screenshot is inference. I have kept the inference labelled below.
The reported test replaces a menu of objectives with a free-text goal box and a manual fallback.
Why Google is doing this (inference)
Campaign setup asks advertisers to understand objectives, campaign types, conversion actions, assets and bidding before they spend. A natural-language step lowers that barrier. It fits the broader direction: Google has been moving Dynamic Search Ads, automatically created assets and campaign-level broad match into AI Max. In Google's own announcement, published 15 April 2026 and updated 11 June, automatic upgrades for ACA and broad match campaigns begin from September 2026 and the DSA sunset moved to February 2027. A goal-first creation flow is consistent with that shift, though Google has not linked the two.
The risks to plan for
Search Engine Land lists the open questions: supported campaign types, how much control advertisers keep over recommendations, and how advanced requirements are handled. From a practitioner view, the risks are these.
- Goal drift. A vague goal ("get more customers") can map to a campaign type you did not intend. Be specific on conversion definition, geography and margin.
- Asset claims. Suggested headlines may state benefits, prices or guarantees you cannot support.
- Conversion-tracking dependency. AI optimises towards whatever conversion is set up. If your account counts micro-conversions, the campaign may chase them.
- Budget exposure. Automation tends to expand reach. Without a cap you may scale spend before you have learnt anything.
- Regulated categories. Education, health, finance and similar verticals have advertising rules; AI wording does not exempt you.
Know AI Max controls first
If the flow creates Search campaigns, you are likely to meet AI Max features. Google's AI Max help page documents three: search term matching, text customisation and final URL expansion, plus brand controls, location controls and URL inclusions and exclusions. It also lists cautions I would read before launching anything:
- AI Max is described as not effective if campaigns are limited by budget.
- With final URL expansion on, pinned RSA assets may not be respected.
- Final URL expansion requires text customisation to be on.
- Tracking templates must be compatible; Google says to use the {lpurl} tag and test landing pages.
- Turning off broad match or moving to AI Max changes keyword prioritisation, so exact-match copies may be needed.
The reporting side is covered in Google's reporting help page, which describes the search terms report with an "AI Max" match type, the combined search terms and ad combinations view, the landing pages report with "Selected by", and asset reports that label Google-generated text as "Google AI." For the new combined view, see my guide to reading the AI Max search terms report.
A cautious adoption checklist
Use this if the test appears in your account.
Before you click anything
- Write a one-paragraph brief offline: goal, conversion event, target CPA or ROAS, geography, audience, exclusions, and claims you may and may not make.
- Confirm conversion tracking and that the primary action is the one you want.
- Set a daily budget cap you would be comfortable losing in a bad fortnight.
During creation
- Paste your brief rather than typing a short phrase.
- Read every suggested goal and asset. Delete anything with an unverified claim, price or guarantee.
- Check the campaign type chosen. If it is not what you wanted, use the manual route.
- Turn off features you are not ready to monitor (for example final URL expansion) until you have a baseline.
After launch
- Record the exact settings and the date in a change log.
- Review search terms and landing pages every one to two weeks. Google's reporting guidance suggests waiting at least two weeks after enabling AI Max before making changes such as adding negatives, so the campaign can learn.
- Judge on conversions, cost per qualified lead and customer quality, not clicks. Search Engine Land makes the same point about AI Max: more visibility does not automatically mean better optimisation.
A decision table
| Situation | Use the AI flow? | Why |
|---|---|---|
| New advertiser, small budget, simple offer | Yes, with review | Helps with structure; low downside if capped |
| Established account with tight CPA targets | Manual, test AI in a duplicate | Preserves control and a clean baseline |
| Regulated vertical (education claims, health, finance) | Manual first | Wording risk; review obligations |
| Multi-region with different offers | Manual | Complex requirements are an unknown |
| Agency managing many accounts | Pilot on one account | Learn failure modes before scaling |
Treat AI-suggested setup as a draft: write the brief first, review every asset, log the settings.
The brief template
Copy this into a doc and fill it before you open Google Ads.
- Business goal: (e.g. 40 qualified enquiries per month)
- Conversion event: (e.g. form submit with verified phone)
- Target economics: (max CPA or min ROAS and why)
- Geography and language:
- Audience and exclusions:
- Offer, price and proof allowed in copy:
- Claims that are banned: (guarantees, outcomes, comparisons)
- Landing pages allowed / excluded:
- Brand terms to include or exclude:
- Review cadence and owner:
If the AI flow produces something that contradicts this brief, the brief wins.
What it means for organic teams
Two practical effects. First, AI-driven campaigns pull copy and URLs from your landing pages, so thin or inconsistent pages show up in ads. Cleaning product, pricing and fee pages helps paid and organic. Second, the search-terms data AI Max exposes is a cheap source of real query language for content planning. For how Google's ad surfaces are shifting inside AI answers, see my notes on AI Mode ad slots and AI Max with Conversion Lift and margin-based ROAS.
What I do not know
I do not know whether the flow applies to Search, Performance Max, Demand Gen or all of them. I do not know if it will respect existing brand-safety settings, or whether suggested assets are labelled the way other Google-generated assets are (Google's Demand Gen help says ads with AI-generated or edited assets may carry AI labels in the EU, India and New York, which is a related but separate feature). I will update this post as Google documents the feature.
FAQ
Is Google Ads natural-language campaign creation available to everyone?
No. Google has called it an experiment and has not shared availability or timing, per Search Engine Land and Search Engine Roundtable on 12 October 2026. Only some advertisers have reported seeing it.
Can I still create campaigns manually?
Yes. The reported screen includes a "Create campaign manually" link that skips the AI-assisted workflow. Keep using it for complex or regulated campaigns.
Which campaign types does it support?
Not public. Search Engine Land lists this as an open question. Do not assume it supports every type; check the campaign type the flow selects before proceeding.
Is this the same as AI Max?
No. AI Max for Search is a set of features (search term matching, text customisation, final URL expansion) inside Search campaigns. The natural-language flow is a creation experience that may produce campaigns using such features. Google has not confirmed the relationship.
Will the AI write my ad copy?
The on-screen text says Google AI will suggest goals and assets. Treat suggestions as drafts and review each for accuracy, claims and policy compliance before launch.
Should agencies use it for clients?
Pilot on one account with a clear brief and a change log. Do not roll it across a portfolio until you understand how it behaves and what controls you retain.
Does it change how I should track conversions?
It makes clean conversion tracking more important. The system will optimise towards the conversions it sees, so confirm the primary conversion action is the one that matters to the business.
How long should I wait before judging results?
Google's AI Max reporting guidance suggests waiting at least two weeks before making changes so the campaign can learn, and reviewing search terms every one to two weeks. Give any AI-built campaign a similar window, subject to budget.
What about education or health advertisers?
Use the manual route and keep human sign-off on every claim. Regulated categories carry advertising rules, and AI-suggested wording does not remove your responsibility.
Where can I follow updates?
Watch Google Ads Help and the Ads Liaison's posts, plus Search Engine Land and Search Engine Roundtable coverage. I will revise this post when Google publishes official details.
Talk it through
If you want help setting up a safe AI-assisted paid and organic workflow, visit younusfardeen.in to see my work and use the contact form. I bring 4+ years of marketing experience and would be glad to review your brief and checklist with you.