Read Time: 6 minutes
We’re going deep on Meta ads today.
But before I do, I want to call out a data point I got this week, for you to share with the AI skeptic in your life:
There are currently 200 Million active users on ChatGPT every week. That is a lot of bosses who are getting lied to in 25% of companies that have banned AI.
On Wednesday, we looked at ten steps to improve your Meta ads by 40% using AI, which hopefully got you thinking that you need a repeatable ad process to get consistent results.
What steps do you take? 👇🏼
I’m going to assume you have a clear goal set.
For example: ‘I need more leads for my roofing business, based in Portsmouth, NH.’
I would start with a prompt like this:
“You specialize in assisting Alec’s Roofing sales and marketing team as an expert Meta ad strategist to prepare a Meta ad campaign for prospects needing new GAF roofs in Portsmouth, NH area. What are the pain points, motivations, priorities and challenges of a home owner who needs a new roof. Exclude buyers who need roof repairs as Alec’s Roofing does not do repairs we are looking to meet the consumer with 5 clear hooks and 5 eye-catching meta creatives to make Alec’s Roofing their clear choice for a new roof backed by our 50 years no questions asked warranty.”
You can see ChatGPT results it produced here.
I would then refine with this prompt:
“What other trends could I incorporate into my ad content from social media to emotionally connect with my roof buyer for Fall?”
Then, I would take those results and use one of my 24 psychological frameworks with a prompt like this:
“Rework this Meta ad campaign using the ‘Social Learning’ framework to showcase the successes and benefits of using our GAF Roofing for $1 million plus homes in the Portsmouth, NH area. Describe the positive outcomes others have experienced with Alec’s Roofing over the last 41 years, and provide incentives for the readers to transform their homes with GAF products.”
Yes, you need to be bossy.
I don’t just run ads with them… I also use Later to schedule my ad creatives as organic posts on LinkedIn.
It only takes a few clicks and it helps me:
Here’s an example of the process…
Frame 1: Before Image (Top-left)
Scene: A luxury home in Portsmouth, NH, with an older, worn-out roof. The house is large and stately, with lush greenery and tall trees in the background. The roof appears faded, with some visible wear and tear, which detracts from the home’s overall appeal.
Caption: “Before: An Outdated Roof Can Hold Back Your Home’s Full Potential.”
Frame 2: Transition Frame (Center)
Scene: A mid-installation shot with roofers working on the home. You can see GAF roofing materials stacked nearby, with roofers in action, using professional equipment.
Caption: “Transform Your Home with GAF Roofing” in a modern, elegant font.
Details: Use arrows or transition effects to show the progress from the old roof to the new roof.
Frame 3: After Image (Top-right)
Scene: The same home with a brand-new GAF roof, shining under a bright blue sky. The house looks rejuvenated, with a modern, sleek roofing design. Manicured landscaping and the clean lines of the roof make the home stand out as a true luxury property.
Caption: “After: A Stunning GAF Roof Brings Your Home to Life.”
Frame 4: Call-to-Action (Bottom)
Scene: A bold call-to-action framed with a clean, elegant font over a subtle background of Portsmouth’s skyline or a neighborhood of high-end homes.
No.
Did it take me two minutes?
Yes.
If I need better quality, I load them into mid-journey and version/edit the images.👇🏼
Meta performance is driven by testing your new creatives. I run tests in a separate test campaign with 3-4 creatives per Ad Set.
If you are new to Meta ad testing, this is a good place to start:
How to Test Facebook Ads Creative in 2024
If you need help on campaign structure watch this:
The BEST Facebook Ads Campaign Structure for 2024
But this is where the fun starts.
70%+ of your tests are going to fail.
Yep, you heard that correctly.
This is why you need a process for creating new creative.
The #1 issue I see is creative burnout. We all get lazy and run an ad too long. No one wants to see you ad 1,000 times.
They zone out on your old ads.
Next, we need to get busy on measuring what’s working and why.
I’d start with a prompt like this:
“You are an expert Facebook Ads analyst for the roofing industry, optimizing campaigns across various products. I’m going to provide you with my ad performance data, and I need your expert analysis and recommendations.
Follow these steps…
Data Overview:
Summarize the key metrics I’ve provided, including impressions, clicks, CTR, CPC, conversions, conversion rate, and ROAS.
Performance Analysis:
Identify the top-performing ads, ad sets, and campaigns based on ROAS and conversion rate.
Highlight any significant trends or patterns in the data.
Point out any anomalies or unexpected results.
Audience Insights:
Analyze the performance across different demographics, interests, or behaviors.
Suggest any audience segments that appear to be particularly responsive or unresponsive.
Creative Analysis:
Evaluate the performance of different ad formats (image, video, carousel, etc.).
Identify any common elements in high-performing ads.
Suggest improvements or tests for underperforming creatives.
Optimization Recommendations:
Provide specific, actionable recommendations for improving overall campaign performance.
Suggest budget allocation strategies based on the performance data.
Recommend any changes to bidding strategies or optimization goals.
Scaling Opportunities:
Identify campaigns or ad sets that show potential for scaling.
Suggest methods for scaling these campaigns while maintaining efficiency.
Testing Ideas:
Propose 3-5 specific A/B tests based on the data to improve performance further.
Industry Benchmarks:
Please compare the performance to industry benchmarks and suggest areas for improvement.
Future Strategy:
Based on all the above analysis, provide a brief strategic plan for the next 30 days to improve overall account performance.
Please provide your analysis and recommendations in a clear, structured format. Use bullet points where appropriate for readability.”
THEN, we iterate what we get back using ChatGPT’s memory, digging into areas of promise to cycle the next set of creatives for testing.
Ask it questions until you are clear on what you need to do.
If I can do this, you can do it too.
I have zero extra time; welcome to teaching 4 classes at UNH, running a portfolio of companies, and hugging AI skeptics twice a week in this newsletter.
BUT, I still run my own ads. 🫣
Have a process for Meta > automate 40% of that process with AI = have more backyard time with my 3-year-old daughter. Getting better ads consistently and doing less work.
Take action. The only thing holding you back is fear. That goes away 1% at a time, with one Ad set tested and one AI prompt done.