How to build a free static ad machine with Gemini 3

Transcript

In this video, I'll be showing you exactly how to create a static ad machine for your business using Gemini 3 Nano Banana Pro. This machine allows you to test creatives faster than ever before and is essentially free. So, the first thing we're going to do is go into our ad account, go to columns, compare attribution settings, and select incremental... Read More

Key Insights

• X is incremental attribution, which is used to measure and compare the performance of ads over time to identify true winners.
• The key to testing efficiency is to duplicate top performing templates and swap the product or model while keeping core elements intact.
• The process relies on Gemini 3 and Nano Banana Pro to generate high fidelity image swaps with minimal loss of realism.
• Removing text from images is highlighted as a simple way to produce clean variations that avoid logo and message clutter.
• A workflow is shown where competitor templates are used as inspiration, then adapted to the brand’s products and models for faster creative development.
• The approach emphasizes testing at scale by generating many variations quickly rather than relying on a single variant.
• Incremental ROAS is used as a benchmark to prioritize ads that exceed the average threshold for spending and revenue impact.
• The overall strategy combines data-driven attribution with template-based creativity to maintain consistency while exploring new visuals as fast as possible.

Questions & Answers

Q: A real search query question about the video (e.g. 'How to...', 'What is...', 'Why does...'). Question 1

Incremental attribution helps determine which ads truly drive revenue beyond mere impressions by showing a higher return on ad spend when compared to a baseline. This allows marketers to pause underperforming ads and reallocate budget toward creative variants that have demonstrated stronger incremental performance, enabling more efficient scale over time.

Q: Question 2

The video demonstrates using a single top performing ad template and swapping in different products or models to create multiple variations. This preserves the core composition and creative feel while testing what combinations yield the best incremental ROAS, helping to discover scalable assets without starting from scratch each time.

Q: Question 3

Nano Banana Pro is used to generate edits of existing ad images, such as substituting a jacket for a shirt or removing on image text. The process involves feeding prompts that describe the desired edit, then allowing the tool to produce a new image that matches the original template but with the requested changes, improving iteration speed.

Q: Question 4

The method includes removing text like sale messaging to create clean image variants. This makes it easier to reuse templates across different campaigns or platforms, reduces clutter, and helps in testing how product visuals perform without advertising copy influencing viewer interpretation.

Q: Question 5

Competitor templates are analyzed for inspiration, then adapted to the brand by replacing the model or product while maintaining the underlying template. This approach accelerates creative development, enables rapid diversification of ads, and supports building a broader library of adaptable templates for ongoing campaigns.

Q: Question 6

The flywheel described involves a cycle: create new ads, analyze with incremental attribution every 7 to 14 days, pause duds, scale winners, and brief the best scaling ads to produce new variations. Repeating this loop helps sustain performance gains and keeps campaigns aligned with evolving audience responses.

Q: Question 7

The video emphasizes keeping core elements like the model presence while varying the product or jacket to maintain recognizability. This balance helps preserve brand identity in generated variants and increases the likelihood that viewers respond similarly to original ads while exploring new product combinations.

Q: Question 8

A practical takeaway is to use prompts that describe the template and the exact change, such as swapping the product while keeping pose and layout intact. Clear prompts help Gemini 3 and Nano Banana Pro deliver high fidelity substitutions with predictable outcomes, reducing creative guesswork during iterations.

Summary & Key Takeaways

• A method is shown to create multiple ad variations by swapping products in a single template while preserving core elements, enabling rapid testing and iteration. It emphasizes using incremental attribution to identify winners and to pause underperformers, then scale the best performers with new variations. The approach leverages Gemini 3 and Nano Banana Pro for image generation and editing.

• The speaker demonstrates how to extract templates from competitor ads, apply them to own products, and maintain branding by changing only the model and product while keeping the template intact. The workflow includes removing text, adjusting visuals, and using prompts that describe the desired changes. This creates scalable, copyright-conscious ad variations.

• A flywheel is outlined: analyze ads with incremental attribution every 7 to 14 days, pause duds, scale winners, and brief the best performing ads for new variations, using outside sources as inspiration to continuously generate new content.