If your Facebook ads are tanking despite swapping creatives, audiences, and copy, it’s time to stop guessing and start diagnosing. I built an AI prompt to tell you exactly what’s wrong with your ads and how to fix them fast.
This isn’t about throwing more money at failed campaigns or endless trial and error. It’s about getting clear, actionable feedback from AI so you can reduce wasted spend and scale your campaigns with confidence. This process works for any niche, budget, or objective.
Why Diagnose Your Facebook Ads with AI?
Most marketers throw spaghetti at the wall when their Facebook ads don’t perform. They swap images, tweak audiences, rewrite copy, and cross their fingers. What they miss is a clear diagnosis of what’s actually broken.
AI can spot problems you won’t see. It can analyze the synergy between your copy, creative, and audience targeting, then tell you what’s working and what’s not. This means fewer wasted dollars and less guesswork before you scale.
Diagnosing ads with AI isn’t about predicting success with 100% certainty. It’s about understanding where your ad fails to connect, what your audience really wants, and how to sharpen your message to get clicks.
Choosing the Right AI Model
Not all AI models are created equal. Many are text-only and can’t “see” your ad image. Since Facebook ads combine visuals and copy, you need a model that understands both.
I use GPT-4 Mini, a vision-enabled AI model that analyzes images and text together. This lets it evaluate the entire ad experience, not just the copy. Depending on when you read this, newer or better models may be available. The key is to pick an AI that processes text and images simultaneously.
The Anatomy of My AI Diagnostic Prompt
Here’s the breakdown of what I feed the AI:
- Role definition: The AI is told upfront to act as an expert Facebook ad diagnostician and strategist.
- Ad objective: Whether it’s traffic, conversions, leads, or sales, you specify the goal so the AI can evaluate the ad’s effectiveness against it.
- Audience description: You must describe your target audience in detail. The AI needs to “become” this audience to judge how they’d react to the ad. For example: “Urban Gen Z thrift hacker who flips vintage finds online and tracks every carbon gram saved.”
- Ad creative: You provide the image or video for the AI to analyze.
- Ad copy: The headline, body text, and any calls to action.
This is a summary. I’ve made this prompt available for free at CEOworkbench.com. You just fill in your ad info, paste your creative, and let the AI run the analysis.
How the AI Breaks Down Your Facebook Ad
Once you hit go, the AI generates a detailed report. Here’s what it covers:
1. Overall Impression and Clickworthiness
The AI describes the vibe of your ad – is it calm, exciting, minimalist? It then estimates the likelihood your target audience will click, based on your objective.
For example, the AI might say: “Warm minimalist flat-lay of zero-waste bathroom staples. Feels calm and intentional. Click likelihood for traffic objective is moderate.” This instantly tells you the ad isn’t terrible, but it’s not grabbing enough attention.
2. Visual Analysis
The AI evaluates your image’s effectiveness. It might flag a muted color palette as an issue because it gets lost in a fast-scrolling feed. This is a detail you might miss but could be killing your engagement.
3. Copy Analysis
The AI critiques your headline and body copy separately. It looks for:
- Whether the headline highlights a tangible benefit or metric.
- If the copy creates curiosity and urgency.
- Whether the message matches what the audience expects.
- If the copy lowers the commitment barrier.
- Any missing proof points or social validation.
For instance, the AI might point out your headline lacks a promise that justifies a click or that your body copy misses proof such as carbon grams saved.
4. Synergy Between Copy and Creative
The AI assesses whether the image and text work together or send mixed signals. This helps you avoid disjointed ads that confuse your audience.
5. Actionable Recommendations
AI doesn’t just critique your ad; it gives you specific copy tweaks and new headline options designed to boost performance. For example, “Swap One Plastic Item, Save 42 g of Waste Today.”
This quantifies the benefit and lowers the friction for your audience to act.
Testing Headlines with AI
Beyond diagnosing, the AI can generate multiple headline options tailored to your audience to maximize click-through rate (CTR). For example, after inputting a suggested headline, you can ask the AI for five punchy alternatives tuned for eco-savvy Gen Z scrollers.
Each headline pairs a single, low-friction action with a concrete benefit to drive clicks. This gives you ready-to-run A/B test options without guessing what might work.
Limitations and How to Overcome Them
There’s one downside to this method: you’re asking a single AI to impersonate an entire audience. That’s a lot of weight for one model to bear and can limit accuracy.
I’ve addressed this in another approach using a swarm of AI’s, each acting like a specific customer persona. It produces even more precise feedback. If you follow me on X/Twitter or LinkedIn I show how I do that.
But even this “quick and dirty” cut-and-paste prompt delivers far more insight than blind guesswork.
What To Do Next
For more on running winning Facebook ads and marketing smarter, check out these resources: