AI Discovery and Strategic Planning for Ad Creative Analytics
Problem
A client in the digital advertising industry sought to improve their platform by integrating advanced Artificial Intelligence (AI) capabilities. They recognized that using AI technologies such as Machine Learning (ML), Data Science, and Generative AI could significantly add value to their services. However, they faced challenges in determining the optimal strategies for AI adoption that would align with their business goals, efficiently utilize their extensive historical data, and manage risks related to cost, scalability, and resource allocation.
Also applicable to
This scenario is common among businesses in data-intensive industries like marketing analytics, digital advertising, media platforms, and any organization aiming to implement AI solutions for data analytics, predictive modeling, or automation. Companies looking to adopt AI technologies, including but not limited to Large Language Models (LLMs), Natural Language Processing (NLP), and data governance practices, often encounter similar challenges in strategy development and technology implementation.
Solution
Our AI team collaborated closely with the client to develop a customized roadmap for AI adoption, aligning with their strategic vision and using their unique assets. At a high level, we focused on two key areas: optimizing prompt engineering for immediate benefits and utilizing their historical data for long-term AI capabilities. Since the client needed to balance high-value opportunities with uncertainties like effectiveness, cost, and scalability while aligning with their strategic goals, we found using the Value/Risk framework to be the most suitable approach, as opposed to alternatives like the Impact/Effort matrix, SWOT analysis, or Cost-Benefit analysis.
Optimizing Prompt Engineering
1) Manual Prompt Engineering
We recommended starting with manual prompt engineering to allow immediate experimentation with existing AI models while evaluating added value to their customer. This approach enabled the client to quickly test and refine prompts, gaining valuable insights with minimal initial investment. It provided an opportunity to improve customer value by iteratively improving the AI-generated outputs.
Risks: Relying solely on manual prompt adjustments could lead to: A) generic outputs, performance plateau causing Ad fatigue and B) scalability issues as demands grow, potentially causing delays in adapting to market trends.
2) Expert-Guided Prompt Engineering
To elevate the effectiveness of AI outputs, a proposed solution was involving their domain experts (advertising strategists) in the prompt engineering process. This strategy improves prompt quality through expert insights, reduces dependence on external creative specialists, and builds a rich knowledge base for future AI initiatives. It also lays the groundwork for developing proprietary models aligned with their industry-specific knowledge and strategies.
Risks: This approach may increase resource demands due to ongoing expert involvement and pose coordination challenges in managing inputs from multiple experts, potentially affecting consistency.
Using Historical Data for Insight
1) Extracting Creative Principles
During the initial audit and research phases of the project, we assessed the client's extensive historical data and identified its potential for uncovering effective creative strategies. By extracting valuable insights from past successful campaigns, they could improve instructions for future initiatives and gain a competitive market advantage.
Risks: There may be data gaps that do not capture emerging trends, and the analysis requires advanced tools and expertise, potentially incurring additional costs.
2) Training an Expert Model
Building on the insights gained, we recommended developing custom AI models trained on their proprietary data and expert knowledge. These models aim to capture and scale their unique expertise, ensuring consistent quality and reducing reliance on individual experts.
Risks: According to our cost estimates and TCO analysis, this strategy requires significant time and resources for development and demands substantial high-quality data to train the models effectively.
AI Roadmap
Throughout the engagement, we emphasized the importance of aligning AI initiatives with the client's business objectives, industry standards, and compliance requirements. Our approach focused on delivering measurable impact and return on investment, crafting scalable, high-impact solutions to drive real business growth.
As part of this effort, we proposed a tailored roadmap for their long-term AI adoption, designed to align with their strategic vision and ensure successful implementation.
Impact
By implementing our recommended strategies, the client stands to achieve significant benefits:
- Accelerated AI Adoption: Immediate experimentation with AI through manual prompt engineering allows the client to quickly integrate AI capabilities into their platform, building innovation and keeping pace with industry advancements.
- Improved Quality and Consistency: Expert-guided prompt engineering improves the effectiveness of AI outputs, ensuring they align with industry-specific knowledge and strategies, which improves customer satisfaction.
- Competitive Advantage through Data Utilization: Using historical data to extract creative principles provides unique insights that inform better decision-making and strategy formulation, giving the client a competitive edge in the market.
- Scalability and Efficiency: Developing custom AI models enables the client to scale their operations efficiently by reducing reliance on individual experts, ensuring consistent quality across their services.
- Strategic Alignment and Compliance: Aligning AI initiatives with their strategic vision and compliance requirements mitigates risks and ensures sustainable growth.
Technologies
- Artificial Intelligence (AI) and Machine Learning (ML): Implementing AI and ML techniques to improve data analytics and predictive capabilities.
- Large Language Models (LLMs): Using multimodal LLMs for advanced natural language processing tasks.
- Prompt Engineering: Crafting and refining prompts to optimize AI model outputs.
- Custom AI Models: Developing proprietary AI models trained on the client's historical data and expert knowledge.
- Data Analytics Platforms: Utilizing advanced data analytics tools for extracting insights from historical data.
- AI Strategy and Governance: Ensuring alignment with strategic vision and compliance through AI strategy formulation and governance frameworks.
Our team of machine learning scientists, engineers, and product experts works diligently to uncover, strategize, design, and implement state-of-the-art AI technologies tailored to clients' needs. By reducing costs, saving time, and ensuring quality, we helped clients navigate the complexities of AI adoption, transforming their vision into impactful results.
Known professionally as Dr. Med with 12+ years of experience implementing AI products. Strategic advisor to multiple successful founders including a unicorn, US patent holder. Ex-SOTI, UC Berkeley PhD.
Ex-Microsoft. 10+ years of experience in AI/ML, MSc from Université de Montréal (Mila). US patent holder.