The short answer

You can create a Google Search campaign with AI by giving an MCP-connected assistant the business context and campaign inputs it needs: account, goal, budget, location, bidding strategy, landing page, ad-group themes, keywords, negatives, and ad copy.

The safest workflow is staged. Create the campaign in a paused state, verify its settings, add ad groups and targeting, create responsive search ads, then review the complete build before enabling it. That keeps AI useful without turning “build a campaign” into “surprise me with a live campaign.”

With HireOtto, the new Search campaign is paused by default. It uses a dedicated daily budget, targets Google Search only, presence-based location targeting, and a start date of the following day. You can then continue the build in the same conversation.

What AI should and should not do

AI is good at translating a structured brief into Google Ads entities, keeping track of campaign and ad-group IDs, enforcing ad character limits, and carrying context across the build.

It should not invent your economics or positioning. The operator still owns the decisions that determine whether the campaign deserves to exist:

  • The offer and landing page

  • The conversion action that matters

  • The acceptable cost per conversion or return target

  • The geographic market and language

  • Keyword intent and exclusions

  • Claims, compliance, and brand voice

  • The final decision to enable the campaign

Before you ask AI to create the campaign

A good prompt cannot rescue a weak brief. Prepare these inputs first:

  • A Google Ads account with billing active and the required permissions

  • A clear campaign goal: leads, sales, qualified traffic, or visibility

  • A landing page that matches the search intent

  • Reliable conversion tracking before using conversion-based bidding

  • The daily budget and the maximum CPA, CPC, or ROAS constraint you can defend

  • The target locations and language

  • A first-pass keyword list grouped by intent

  • Obvious negative keywords

  • Proof points, differentiators, and approved ad claims

If keyword research is not ready, do that first. AI can generate ideas and historical metrics, but you should still review relevance, commercial intent, and likely CPCs before adding anything.

Use a campaign brief, not a vague command

“Create a Google Ads campaign for my business” leaves too much room for interpretation. Start with a brief that gives the assistant the operating constraints.

I want to build a paused Google Search campaign for [OFFER].

Account: [ACCOUNT NAME OR ID]

Goal: [LEADS / SALES / TRAFFIC / VISIBILITY]

Primary conversion: [CONVERSION ACTION]

Location: [LOCATION]

Language: [LANGUAGE]

Daily budget: [BUDGET]

Bidding approach: [STRATEGY AND OPTIONAL TARGET]

Landing page: [URL]

Ad-group themes: [THEME 1, THEME 2, THEME 3]

Important negatives: [NEGATIVES]

Approved proof points: [PROOF]

First show me the proposed structure and any missing inputs. Do not enable anything. 
After I approve the plan, create the campaign one step at a time and return the IDs created at each step.

That last instruction matters. A staged build creates natural review points and makes errors easier to isolate.

Step 1: Create the paused Search campaign

At minimum, provide the campaign name, daily budget, location, and bidding strategy.

Create a paused Search campaign called “US_NonBrand_ProjectManagement_Search” with a $100/day budget, targeting the United States in English. 
Use Maximize Clicks with a $12 maximum CPC. Do not enable it.

HireOtto supports Maximize Clicks, Maximize Conversions with an optional target CPA, Maximize Conversion Value with an optional target ROAS, Target Impression Share, and Manual CPC for Search campaign creation.

Which strategy should you choose?
Match it to the campaign’s job and the quality of your conversion data.

  • Maximize Clicks can be useful when you need controlled traffic and do not yet have reliable conversion history.

  • Conversion-based bidding is more defensible when the right conversion action is tracking consistently.

  • Target Impression Share fits visibility-led use cases.

Step 2: Verify campaign settings before adding more

Ask the assistant to read the campaign back before building on top of it:

Show me the settings for campaign [CAMPAIGN_ID]. 
Confirm the status, daily budget, bidding strategy, locations, language, network settings, start date, and tracking setup. 
Flag anything that does not match my brief.

Review these fields:

  • Status: the campaign should remain paused

  • Budget: correct amount and account currency

  • Bidding: correct strategy and any target CPA, ROAS, impression-share, or CPC ceiling

  • Locations: the intended markets, using presence targeting

  • Language: aligned with the ads and landing page

  • Networks: Google Search on; Search Partners and Display off unless deliberately enabled

  • Start date: appropriate for the planned launch

  • Tracking: correct landing-page and campaign tracking parameters

Fix the campaign here rather than after ads and keywords have been built around the wrong assumptions.

Step 3: Build tightly themed ad groups

Each ad group should represent one coherent search intent and point to a landing page that answers it. Do not group keywords merely because they contain similar words.

Create an ad group called “Construction Project Management Software” in campaign [CAMPAIGN_ID]. Return the new ad group ID.

A useful test: could one set of headlines, descriptions, and one landing page answer every keyword in the ad group? If not, split the theme.

Step 4: Add keywords with deliberate match types

You can add keywords with one common match type or mix match types in the same instruction.

Add these keywords to ad group [ADGROUP_ID]:

“construction project management software” as exact match

“construction management platform” as phrase match

“construction project software” as phrase match

Exact and phrase match can provide more control in a new campaign.
Broad match can expand reach and works best when conversion tracking, Smart Bidding, budgets, and negative-keyword discipline are strong enough to absorb exploration.
The right choice depends on the account – not on a blanket “AI best practice.”

Step 5: Add negatives before launch

Negative keywords are part of campaign construction, not cleanup you postpone until spend appears.

Add “jobs”, “salary”, “course”, and “free download” as campaign-level negative keywords to campaign [CAMPAIGN_ID]. Use the appropriate match types and show me what will be excluded before applying them.

Use campaign-level negatives for exclusions specific to this campaign. Use a shared negative keyword list for exclusions you expect to reuse across campaigns, such as employment or support-related traffic.

Step 6: Create responsive search ads

Responsive search ads can contain up to 15 headlines and four descriptions. Google requires at least three headlines and two descriptions. Headlines are limited to 30 characters, descriptions to 90, and each display-path field to 15.

Create a paused responsive search ad in ad group [ADGROUP_ID].

Final URL: [LANDING PAGE]

Display paths: [PATH 1] / [PATH 2]

Headlines:

[HEADLINE 1]

[HEADLINE 2]

[HEADLINE 3]

[ADDITIONAL UNIQUE HEADLINES]

Descriptions:

[DESCRIPTION 1]

[DESCRIPTION 2]

[OPTIONAL DESCRIPTION 3]

[OPTIONAL DESCRIPTION 4]

Validate character limits and flag repeated or unsupported claims before creating the ad.

Write assets that can make sense in different combinations. Include the primary keyword or intent in at least one headline, vary the benefits and calls to action, and avoid pinning unless a brand, legal, or compliance requirement must appear in a fixed position. Google notes that excessive pinning reduces the combinations available for optimization.

Google currently recommends testing more than one strong RSA per ad group. Whether you create one or several, judge results at the ad-group and campaign level as well as the individual-ad level.

Step 7: Run a pre-launch review

Do not ask only, “Does this look good?” Give the assistant a checklist and require it to surface mismatches.

Review campaign [CAMPAIGN_ID] without changing anything.

Check:

1. Campaign status, budget, bidding, dates, locations, language, and networks

2. Conversion action and bidding alignment

3. Ad-group themes and landing-page match

4. Keyword relevance and match types

5. Missing or risky negative keywords

6. RSA character limits, duplication, pinning, claims, and final URLs

7. Whether every ad group has eligible keywords and at least one complete RSA

Return a table with: check, finding, risk, and recommended action. End with a clear “ready to enable” or “not ready” recommendation.

Then review the output yourself. AI can detect inconsistencies; it cannot accept commercial or compliance risk on your behalf.

Step 8: Enable only after approval

Once the review passes and you have checked the actual account setup, enable the campaign explicitly:

Enable campaign [CAMPAIGN_ID]. Then confirm its status and list any ads, keywords, or ad groups that are still paused, disapproved, limited, or incomplete.

The confirmation step catches an easy mistake: a campaign can be enabled while an ad, keyword, or prerequisite still prevents delivery.

Common mistakes when creating campaigns with AI

  • Using one giant prompt with no review gates

  • Letting the assistant invent the CPA, ROAS, offer, or conversion goal

  • Choosing conversion-based bidding before reliable conversion tracking exists

  • Combining different intents because the keywords look linguistically similar

  • Adding keywords without negatives

  • Accepting generic ad copy that does not match the landing page

  • Over-pinning RSA assets

  • Forgetting to verify Search Partners and Display Network settings

  • Enabling before checking final URLs, claims, budget, and location targeting

  • Assuming “created successfully” means “ready to serve and likely to perform”

Where HireOtto fits

HireOtto is a remote MCP server that connects AI assistants such as Claude and ChatGPT to Google Ads. It lets the assistant create and manage the campaign through the Google Ads API while you stay in the conversation.

For this workflow, HireOtto can create the paused Search campaign and dedicated budget, set location and language targeting, create ad groups, add positive and negative keywords, create responsive search ads, read settings back for verification, and enable the campaign when you explicitly approve it.

Start with the setup guide if HireOtto is not connected yet: https://docs.hireotto.com/quickstart

Use the procedural campaign guide for copy-paste commands and current parameters: https://docs.hireotto.com/guides/create-a-google-ads-campaign-with-ai

Frequently asked questions

Can ChatGPT or Claude create a Google Ads campaign?

Yes - when the assistant is connected to a write-capable Google Ads integration such as HireOtto. The chat interface supplies the reasoning and instructions; the MCP connection supplies authenticated access to Google Ads.

Will the campaign go live immediately?

Not with HireOtto’s Search campaign creation workflow. The campaign is created paused by default. It will not serve until it is explicitly enabled, but you should still verify the statuses of the campaign’s ads, keywords, and other components.

Can AI choose my keywords and ad copy?

It can research, group, draft, and validate them. You should still review search intent, landing-page match, CPC economics, claims, brand voice, and exclusions.

Can I create the whole campaign in one prompt?

An assistant can execute the steps sequentially, but a staged workflow is safer. Pause after the campaign, after the ad-group structure, and before enabling. Those gates make corrections cheaper.

Which bidding strategy should I use?

Use the strategy that fits the goal and available signal. Maximize Clicks can suit controlled traffic gathering; Maximize Conversions or Conversion Value needs trustworthy conversion data; Target Impression Share suits visibility goals; Manual CPC is useful when direct bid control is intentional. Review account-specific evidence before deciding.

About Me

I’m Suyash – badminton junkie, ex‑GroupM ad‑ops grunt, first marketer at a B2B SaaS startup, and creator of Hiretto: Google Ads MCP Server.

My mission: less clicking, more thinking.

Let’s build leverage together.

Keep Reading