Google Ads’ Goal-First AI Campaign Setup: What Agencies Should Do Before They Trust the Prompt Box

Google Ads is testing a new way to start a campaign. Instead of choosing a campaign type first, some advertisers now see a screen that asks “What’s your goal for this campaign?” Google AI then suggests goals and assets based on what you type. For agencies, this changes where the real work happens: the quality of the brief you type becomes the quality of the campaign you get. Here’s what’s new, what’s still unknown and how to use it without losing control.

What’s actually new

According to reports from Search Engine Land (Anu Adegbola) and Search Engine Roundtable (Barry Schwartz) on October 12, 2026, the test works like this:

• Goal-first prompt: campaign creation opens with the question “What’s your goal for this campaign?”
• AI suggestions: Google AI suggests goals and assets based on your answer.
• Manual escape hatch: a “Create campaign manually” link is still on the screen.

Google Ads Liaison Ginny Marvin confirmed it is an experiment and asked advertisers for feedback. That’s all that is confirmed. There is no help page, no rollout timeline and no list of supported campaign types. Treat it as a test, not a new standard workflow.

Why agencies should care

Brief quality is the control surface

In a manual setup, control comes from settings: campaign type, bidding, locations, audiences, assets. In a goal-first setup, the first thing Google reads is your sentence. A vague prompt like “get more leads” gives the AI very little to work with, so it fills the gaps with its own assumptions. A specific prompt narrows those assumptions. The prompt box is now part of your account strategy, and it deserves the same care as a client brief.

Review time doesn’t disappear

AI suggestions can speed up the first draft, but they don’t remove the need for checks. Suggested goals may not match how the client measures success. Suggested assets may use claims, offers or tone the client hasn’t approved. The time you save on building moves to reviewing. Plan for that in your process and in what you tell clients.

How to write a usable goal prompt

A good goal prompt reads like a short brief. Cover five things:

• Outcome: the one action that matters, such as booked calls, form leads or online sales. Be specific about what counts.
• Offer and audience: what you’re selling and who it’s for.
• Where and when: target locations, service areas and any seasonal or time limits.
• Constraints: budget range, claims to avoid, brand rules and anything that must not appear in ads.
• Landing URL: the exact page people should reach, so suggestions are based on the right content.

Here’s a template you can adapt:

“Goal: generate [qualified leads / booked calls / purchases] for [product or service] from [audience]. Target [locations] during [time period]. Monthly budget around [amount]. Send traffic to [landing URL]. Only use claims shown on that page. Do not mention [restricted terms, competitors, unapproved offers]. Tone: [plain, professional, friendly].”

Keep a copy of every prompt you use in the client file. If results go wrong, you’ll want to know exactly what the AI was told.

8-point QA checklist before you launch

1. Goal match: does the suggested goal match the client’s real success metric?
2. Conversion tracking: is the right conversion action set up, firing and set as primary?
3. Campaign type: is the campaign type Google chose the one you would have chosen manually?
4. Locations and language: are targeting settings correct, including presence versus interest?
5. Budget and bidding: are daily budget and bid strategy within what the client approved?
6. Asset accuracy: are headlines, descriptions and images accurate, approved and free of unsupported claims?
7. Landing page: does every ad send people to the intended URL, and does that page support the ad’s message?
8. Automation settings: have you reviewed automatically created assets and other auto-applied options, and set them on or off deliberately?

When to stay fully manual

The “Create campaign manually” link is there for a reason. Use it when:

• The client works in a regulated or sensitive category where every claim needs approval.
• You’re rebuilding or restructuring an account with a proven setup you need to keep.
• Reporting depends on a fixed naming, structure or tracking setup.
• You need tight control over keywords, placements or exclusions from day one.
• You can’t yet tell which campaign types the test supports for your use case.

A sensible approach is to test goal-first setup on a low-risk campaign, compare the draft against what you would have built manually and note the gaps.

Tie-in for Digital PR, guest posting and paid

AI-led setup reads your landing pages and brand signals to suggest goals and assets. That makes the content behind your ads more important. Clear landing pages, consistent messaging across your site and accurate claims give the AI better material. Digital PR and guest posting help here too: they build the brand awareness and trust that make paid clicks more likely to convert. Paid, organic and PR work best when they all tell the same, accurate story.

Bottom line

Google’s goal-first campaign setup is an experiment with no official documentation yet. It may save time on first drafts, but it moves control into the prompt and the review. Write prompts like briefs, check every suggestion against a fixed QA list and stay manual when the stakes or the unknowns are high.

FAQ

Is goal-first campaign creation available to all advertisers?

No. Google Ads Liaison Ginny Marvin confirmed it is an experiment. There is no rollout timeline or help page yet.

Can I still build campaigns the old way?

Yes. The test screen includes a “Create campaign manually” link.

Which campaign types does it support?

Google hasn’t published a list of supported campaign types. Check what the screen suggests and compare it with what you would choose yourself.

Do I need to review AI-suggested assets?

Yes. Check every suggested headline, description and image for accuracy, approved claims and brand fit before launch.

How do I give feedback to Google?

Ginny Marvin asked advertisers for feedback on the experiment. If you see the test, share specific notes on what worked and what didn’t through your usual Google Ads feedback channels.

Want more practical guides like this? Read more on the Tech Hustler blog, explore our services or book a free strategy call and we’ll review your Google Ads setup with you.