Testing Google Ads AI Campaign Setup: Can Automation Really Build a Winning Campaign?
Google is all-in on AI. Everywhere you look inside the Google Ads platform, there are prompts encouraging advertisers to use automation, AI-generated assets, and machine learning-powered campaign types like Performance Max.
In Google’s ideal world, advertisers would simply provide two things:
- A website URL
- A valid payment method
Then Google’s AI would supposedly handle everything else.
The promise sounds appealing, especially for small businesses overwhelmed by the complexity of Google Ads. But how close are we really to that reality?
To find out, this article breaks down a real-world test of Google’s AI campaign creation tools using a brand-new Google Ads account. The goal was simple: determine whether Google’s automation can genuinely create an effective campaign from scratch — or whether advertisers still need hands-on expertise to get good results.
Why Google Is Pushing AI So Aggressively
Google’s motivation is obvious.
The easier it becomes to launch ads, the more businesses will advertise on the platform. Removing the learning curve around:
- Keyword match types
- Campaign structure
- Ad copywriting
- Conversion tracking
- Audience targeting
…allows Google to onboard far more advertisers at scale.
For new businesses, Google Ads can feel intimidating. AI aims to reduce that friction by simplifying setup and automating decision-making.
But there’s a major catch:
Automation only works if the AI systems are genuinely effective.
And that’s exactly what this test aimed to uncover.
The Test Setup
To simulate a realistic beginner experience, a brand-new Google Ads account was created with no previous campaign history or conversion data.
The chosen business was an electrical contractor website offering services such as:
- Emergency electrical work
- Fire alarm systems
- EV charger installation
- General electrical services
The campaign objective was straightforward:
- Generate leads
- Use a standard Search campaign
- Rely heavily on Google’s AI prompts and automation features
The idea was to behave exactly like a first-time advertiser would.
Initial Campaign Setup: A Surprisingly Smooth Start
The early stages of campaign creation were relatively simple.
Google guided the setup process by prompting for:
- Campaign goals
- Geographic targeting
- Budget settings
- Conversion objectives
One positive point was that Google no longer automatically enables the Display Network for Search campaigns by default — something that previously caused many poorly optimised campaigns.
The campaign was targeted specifically to London using presence-based targeting, which helps avoid irrelevant traffic from users merely “interested” in the location.
So far, the process looked promising.
The First Major Problem: AI Couldn’t Read the Website
The first serious issue appeared when trying to use Google’s AI asset generation.
The test attempted to use a specific landing page for EV charger installation, but Google repeatedly returned an error:
“Landing page not supported.”
This happened even though the page was publicly accessible and perfectly functional.
Multiple attempts were made, including:
- Using the exact service page URL
- Using the homepage instead
- Adding additional business information manually
Nothing worked.
Eventually, a completely different electrician website had to be used before Google’s AI would finally accept the URL and generate assets successfully.
This raises a serious concern.
If Google’s automation refuses to work with perfectly valid websites, many small businesses could immediately hit a dead end before even launching their campaigns.
What Google AI Generated
Once a compatible website was found, Google automatically generated:
- Ad groups
- Keywords
- Headlines
- Descriptions
- Sitelinks
At first glance, this looked impressive.
The AI correctly identified major services from the website, including:
- Emergency electricians
- Fire alarms
- Storage heaters
- Emergency lighting
However, deeper inspection revealed major flaws.
Keyword Targeting Problems
One of the biggest weaknesses was keyword relevance.
For example, an ad group specifically designed around emergency electrical services included keywords such as:
- Electrical repair London
- Electrical service
- Electrician in London
These are far broader than genuine emergency searches.
This creates a serious problem because users searching for general electrical services could end up landing on an emergency-only page — reducing relevance and likely harming conversion rates.
The keyword selection felt:
- Too generic
- Too limited
- Poorly segmented
Instead of tightly grouping keywords around clear user intent, Google blended emergency and generic searches together.
This is exactly the kind of structural issue experienced Google Ads managers actively avoid.
Overall, the keyword generation scored poorly in the test because relevance is critical for Search campaign performance.
AI-Generated Ad Copy: Functional but Generic
The headlines and descriptions generated by Google were acceptable — but only just.
Some examples included phrases like:
- “Electricians 24/7”
- “Power outage help within the hour”
- “Fast, reliable, professional service”
These are relevant to emergency electrical work, but they lacked originality and emotional impact.
The ads failed to fully leverage the business’s real strengths, such as:
- 17+ years of experience
- Industry certifications
- Trust signals
- Customer reviews
- Specific competitive advantages
Another issue was that many descriptions were unusually short, despite Google allowing up to 90 characters per description line.
This meant valuable selling opportunities were being wasted.
The result was ad copy that felt:
- Bland
- Generic
- Forgettable
Not disastrous — but certainly not exceptional.
Budget Recommendations Were Unrealistic
Google also provided estimated campaign performance figures, including projected conversions and costs.
However, these estimates appeared highly questionable.
Because the account had:
- No conversion history
- No existing data
- No established performance signals
…the AI was effectively making assumptions without meaningful evidence.
This is potentially dangerous for inexperienced advertisers who may trust Google’s forecasts too heavily.
New advertisers could easily believe they are guaranteed strong performance when the estimates may be wildly inaccurate.
Conversion Tracking and Lead Forms
One area where Google performed reasonably well was prompting for lead tracking.
Since no conversion tracking existed in the account, Google encouraged the creation of a lead form extension during setup.
This at least gave the campaign a basic conversion mechanism.
However, the platform still didn’t properly guide users toward implementing more robust conversion tracking — something essential for long-term campaign success.
The Biggest Automation Failure: Additional Ad Groups
The most revealing problem came later.
After publishing the initial campaign, the test attempted to create additional ad groups using AI automation.
This is where the system completely broke down.
While Google successfully generated the first ad group during campaign creation, it could not repeat the process afterward.
Attempts to use the built-in AI assistant to generate:
- New keywords
- New headlines
- New descriptions
…failed repeatedly.
The system could generate some keyword suggestions, but it could not create properly tailored ad copy for new services.
Instead, it simply duplicated the original emergency ad content across unrelated ad groups.
This is a major limitation because effective Search campaigns rely heavily on:
- Highly relevant ad groups
- Tailored messaging
- Precise keyword segmentation
Without that, campaign quality quickly deteriorates.
Final Verdict: Google Ads AI Is Not Ready Yet
So, can Google’s AI fully build an effective Search campaign from scratch?
Right now, the answer is no.
The automation showed flashes of promise, particularly during the initial setup process. But the weaknesses were impossible to ignore:
The Biggest Problems
- Website compatibility issues
- Weak keyword targeting
- Overly generic ad copy
- Poor ad group segmentation
- Inability to scale campaign structure
- Unrealistic performance estimates
Most importantly, the AI still lacks strategic thinking.
It can generate content, but it cannot yet understand campaign intent deeply enough to build high-performing account structures consistently.
What This Means for Advertisers
Small businesses hoping AI will completely remove the need to learn Google Ads are likely to be disappointed — at least for now.
Even with automation enabled, advertisers still need to understand:
- Campaign structure
- Keyword intent
- Conversion tracking
- Ad relevance
- User psychology
Without that knowledge, automated campaigns may launch quickly but perform poorly.
Google will almost certainly improve these systems over time. Eventually, fully AI-generated campaigns may become genuinely competitive.
But today, human expertise still matters enormously.
Automation can assist with setup, but it cannot yet replace proper Google Ads strategy.
And for businesses relying on paid traffic to generate leads and revenue, that distinction matters a lot.
