AI for Better Ad Creative: A Strategic Framework for Scaling Results
In the modern digital advertising landscape, the pressure to produce high-volume, high-converting creative has never been more intense. With platforms like Meta shifting away from the strategy of running hundreds of slight variations of a single ad—now treating such tactics as a singular creative unit—advertisers are forced to pivot. The challenge is clear: how can small brands and lean teams keep pace with the algorithm’s demands without succumbing to burnout or blowing through production budgets?
The solution, according to Fraser Cottrell, CEO of the direct-to-consumer ad agency Fraggell, lies in a systematic, AI-driven creative workflow. By treating generative AI not as a "shortcut" but as a sophisticated research and production partner, marketers can level the playing field, generating professional-grade imagery and strategic copy at a fraction of the traditional cost.
The Misconceptions Holding Marketers Back
Before implementing an AI-centric strategy, marketers must dismantle two prevalent myths. The first is the belief that using AI is inherently lazy. In reality, coaxing high-quality, brand-aligned output from large language models (LLMs) requires rigorous effort, deliberate prompting, and a deep understanding of one’s own brand identity.
The second myth is that AI-generated creative is inherently "low quality." While early iterations of generative models struggled with visual fidelity, current image models are capable of producing outputs nearly indistinguishable from professional studio photography. While video generation still presents challenges, for static-image assets, quality is no longer a barrier to entry. For e-commerce brands, this shift represents a fundamental democratization of production: product images that once required expensive studio shoots and professional photographers can now be generated for mere cents.
Step 1: Building a Foundational Brand Knowledge Base
The success of any AI creative workflow depends on a foundational step often skipped by amateurs: the construction of a brand-specific "Knowledge Base." AI, by default, is a generalist; to make it an expert on your specific product, you must provide it with deep, proprietary, and external context.
The Deep Research Phase
Fraser Cottrell advocates for a "deep research" session at the start of any engagement. Rather than asking a standard LLM for surface-level marketing advice, marketers should use the "Deep Research" capabilities of tools like Google Gemini.
The objective is to synthesize an external profile of the brand. This process involves:
- Customer Sentiment Analysis: Identifying not just who is buying, but why, and—critically—why prospective customers chose not to buy.
- Competitive Landscape: Pulling data from platforms like Reddit to surface common complaints, pain points, and geographic concentrations of the customer base.
- Objection Handling: Understanding the primary friction points that stop a sale.
To ensure high-quality research, Cottrell recommends using voice dictation tools (such as Whisper Flow) to feed a precise, multi-layered prompt to the AI. The goal is to move beyond "quick answers" and force the model to browse the internet to construct a comprehensive, multi-page report on the brand’s standing in the market.
Verifying and Augmenting Data
Once the AI generates this research document, it must be vetted. Cottrell employs a "verification loop" by pasting the document into Claude and instructing it to act as an interviewer. The AI asks questions one by one to confirm the accuracy of the facts, claims, and characterizations. If a detail is incorrect, the human user corrects it on the spot.
Crucially, this is the stage where the user must add "proprietary insight"—the information the internet doesn’t have. This includes internal customer data, nuances of product functionality, and insights gleaned from sales calls. By blending AI-sourced external research with human-verified internal knowledge, the marketer creates a "source of truth" that the AI can reference in later stages.
Step 2: Training a Dedicated Claude Project
Once the research document is polished, it should be uploaded into a "Claude Project." Unlike a standard chat session, a Claude Project functions as a dedicated, persistent workspace. It acts as a contained environment where the AI retains only the context you provide, ensuring that your brand guidelines don’t get diluted by unrelated conversations.
The Components of a Robust Project
To effectively train the AI, the project workspace should be populated with four key data pillars:
- The Deep Research Document: The foundation built in the previous step.
- Voice-of-Customer Data: Exported reviews and testimonials. This provides the AI with the actual language used by your buyers, which is invaluable for writing authentic ad copy.
- Internal Brand Guidelines: A manifesto or "style guide" that defines what your brand stands for, how it communicates, and what constitutes a "good" ad.
- Performance Data & Visual Context: Top-performing ads from the past quarter. Using tools like Poppy, marketers can analyze visual pacing, on-screen action, and hooks, feeding these findings back into the project to inform future creative.
By curating this data, you are essentially "teaching" the AI the difference between a high-performing creative and a dud.
Step 3: Execution and Iteration
With a trained Claude Project serving as the "brain" of the operation, the execution phase becomes highly efficient.
The Hybrid Approach to Image Creation
Cottrell recommends a hybrid workflow: generate the visual asset via AI, but layer the copy manually. This allows marketers to test multiple headlines against a single visual without the need to regenerate the entire image every time.
When prompting for headlines, the process is iterative. Ask the project to generate a set of headlines based on the uploaded reviews. From that list, identify the winners and losers, and—most importantly—explain why to the AI. Because Claude Projects compound knowledge, this feedback loop trains the model to understand your specific stylistic preferences over time.
Generating Visual Assets
For image generation, the goal is to provide specific, professional-grade briefs. If you want a studio product shot, upload a reference photo of your product to the project and ask the AI to generate a prompt for an image generator (like Nano Banana 2 Pro). The prompt should specify the background, lighting, and mood, ensuring that the final output aligns with the brand’s visual identity.
The Role of AI in Video Ideation
While current AI video generation may not yet meet the high standards required for final commercial delivery, it is an elite tool for scripting and ideation. By describing the target persona and the scenario (e.g., a UGC-style marathon runner talking about shoes), the AI can produce a detailed, timestamped script in seconds.
While a human copywriter will still need to refine the script for nuance and tone, the AI reduces the "blank page" problem, getting a creative team 30% of the way to the finish line in a fraction of the time.
Implications for the Industry
The shift toward AI-integrated creative is not just a trend; it is a structural change in how marketing teams will function in the coming years.
The Democratization of Production
For smaller brands, this approach eliminates the prohibitive cost of traditional production. When high-quality image generation is available for a negligible cost, the barrier to testing new ad concepts is lowered, allowing smaller players to experiment with as much agility as industry giants.
The Rise of the "Creative Strategist"
This methodology highlights a shift in the role of the marketer. The focus is moving away from manual creation and toward "creative management." The marketer of the future is less of a pixel-pusher and more of a conductor, training models, refining outputs, and synthesizing data.
Navigating Algorithmic Changes
As platforms like Meta continue to refine their algorithms—prioritizing meaningful, high-performance creative over mass-produced variations—the ability to generate diverse, high-quality content is paramount. By using AI to systematically develop and refine assets, brands can maintain the volume required by the algorithm while simultaneously ensuring that every piece of creative is rooted in deep customer insight.
Final Thoughts
The path to better ad results in the age of AI is not found in a single prompt or a shortcut button. It is found in the preparation. By building a robust knowledge base, training persistent AI models, and maintaining a tight feedback loop between performance data and creative production, brands can stop "chasing the algorithm" and start driving results that resonate with real people. As Fraser Cottrell notes, AI does not replace the creative process—it accelerates it, provided the marketer is willing to put in the work to define the goal.
