It will come as no surprise to anyone that most advertisers are reluctant to do this. use auto-generated creatives from advertising platforms. Auto-generated ads fall into the following categories:
- Customer in the loop (CITL): Assets are generated based on inputs such as a website URL or user prompt. The advertiser always has the choice of whether or not to include these elements in their campaigns.
- Dynamic composition: Ads are composed at the time of delivery in different formats based on existing asset groups, with successful winners selected and scaled (i.e., How does Performance Max work?). May or may not include AI-generated assets depending on client preferences.
- Automatically generated: New assets or ads are generated after launching a campaign based on inputs like URLs, search queries, or existing videos to improve performance. These items are not reviewed or approved by advertisers prior to delivery, but can generally be viewed and monitored in reports.
Even advertisers who adopt automation of bidding, targeting and budget allocation often draw a firm line when it comes to creativity.
This resistance generally comes from several places:
- Quality issues due to generic rather than product/service specific copy.
- Brand compliance requirements.
- A strong desire to retain creative ownership.
- Discomfort with serving ads without a human approving each variation.
Yet, auto-generated designs can sometimes perform as well or better than human-created assets. A A 2025 study found that auto-generated ads had 19% better CTR.
These performance gains are not new; AI ads reach or exceed human creativity by 2018.
This performance advantage comes from two main advantages.
First, auto-generated creatives are highly adaptable. It can adapt to different formats and locations in a way that would be time-consuming or difficult for humans to manage manually.
Second, it is unbiased in its desire to apply the creativity most likely to work for humans searching profitably, rather than the semantic syntax we think will succeed.
This article is not intended to state that auto-generated designs are good or bad. There is no universal answer. The relevance of considering this will always depend on commercial constraints, brand rules and personal comfort level.
What we are What we’ll do is walk through a practical framework you can use to decide whether auto-generated creatives are worth testing for your business, and how to use the platform’s tools to better understand how well your site and messages are being interpreted by AI systems.
Before we get to the heart of the matter, an important revelation. I am an employee of Microsoft Advertising. The advice presented here is intended to be platform-agnostic, but I will reference a few Microsoft-specific tools that are free to use and particularly useful for understanding how your site is interpreted by machines and humans.
The Case for Using Auto-Generated Creatives
The main reason to consider automatic creation is simple: to save time.
At its core, the auto-generated creative takes your existing assets and adapts them to meet the formatting and placement needs of different inventories. Instead of creating custom creations for each surface, you allow the system to reassemble what you already have in a way that reaches more people with less manual effort.
Auto-generated creative assets typically come from your website, your existing ads, and in some cases, concepts that are proven and widely applicable to all advertisers. You can also apply brand style guides to ensure fonts, colors, and creative (including tone of voice) conform to brand standards.
Advertisers who are able to say yes to auto-generated creatives often see faster ramp-up of their campaign. Eligibility for more placements means more opportunities to participate in auctionsand reducing bottlenecks makes it easier for the system to test and learn which creative works best in which contexts.
Because auto-generated creatives allow advertisers to be eligible for more placements (ad rank determines which ad is displayed), they naturally have access to more impressions. More impressions create more opportunities to win auctions, which can translate into additional volume that would have been difficult to capture using only tightly controlled, manually created assets.
Auto-generated creatives don’t have to be all or nothing. There is also a hybrid approach in which humans partner with AI systems. This may involve using tools built into Google or Microsoft’s platform, or external AI tools, to help generate ideas, titles or variations which are then manually reviewed, approved and uploaded.
Some marketers make a distinction between AI-assisted ideation and auto-generated creative. In practice, if you are using AI at any point to create or shape advertising messages, there is already an element of automation in the process.
The Arguments Against Using Auto-Generated Creatives
There are very valid reasons to walk away.
The most urgent thing is respect for the brand. If your organization requires explicit approval for each creative item before spend can be made, allowing systems to dynamically generate variations may simply not be allowed.
That said, many platforms offer preview tools that show examples of how creatives might appear.
If you’re willing to explore these overviews and rely on tools like brand kits that apply fonts, colors and tones, it may be possible to gain internal approval where it seemed impossible before.
Another reason advertisers are hesitant to use auto-generated creative is the reliance on proven assets with no tolerance for variation. Sometimes budget approval depends on using a specific creative that has already demonstrated performance, and it is not possible to test alternatives.
It’s worth noting, however, that auto-generated creatives are already largely dependent on your existing assets. If the main concern is avoiding untested messages, allowing your site content and proven ads to inform the system can help mitigate this risk.
Bonus Tip: Use Auto-Generated Creatives to Understand How AI Sees You
One of the most underrated benefits of campaigns like Performance Max, Dynamic Search Ads, and other feed-based or keyword-free formats is that they reveal how well platforms understand your site and landing pages.
If you strongly disagree with the creative shown in the previews for AI Max, Maximum performanceor similar formats, this is a warning sign. Assigning a budget to these pages risks confusing users if the system’s interpretation doesn’t match your intended message.
These tools can function as diagnostic instruments, not just delivery mechanisms.
You can go further by combining them with behavioral analytics tools like Microsoft Clarity, which shows how users actually interact with your site. When creative interpretation and user behavior mismatch, the problem is often not the ads, but the underlying content.
Another advantage of modern campaign creation tools is their built-in AI editing capabilities. Even if you never allow auto-generated creatives to go live, you can still use these tools to explore tonal shifts, rewrites, and messaging ideas that inform your manual creative work.
There are many use cases for these systems beyond just automation. Generating information is one of the most valuable.
Takeaways
Basically, the decision to rely on auto-generated creatives depends on whether your brand is allowed to conduct testing.
If the answer is yes, there is little downside to experimenting. Auto-generated creatives are largely built from your existing assets, and poor results are often a sign that your landing pages or messages need refinement anyway.
If the answer is no, whether due to brand compliance, limited testing bandwidth, or the need to spend on proven creative, it’s entirely reasonable to opt out.
Used wisely, it can save you time, increase scalability, and reveal insights into how your brand is understood by machines and users. Used blindly, it can create risks. The goal is not blind trust, but informed experimentation.
I hope this was helpful and I’ll see you next month for another edition of Ask the PPC.
More resources:
Featured image: Paul Poetry/Search Engine Journal
