Turning a commercial need into a scalable AI product
AIP - SCALING EARLY CREATIVE DECISION MAKING WITH AI
This work focused on turning a commercial need into a working product.
The goal was to give customers a faster, lower cost way to assess large volumes of creative early in a campaign, before investing in deeper human testing.
THE CHALLENGE
Testing every creative version with full human research was too slow and too expensive.
THE APPROACH
End to end product delivery, from defining the training data through to MVP, iteration, and go to market.
COMMERCIAL NEED
Fast, scalable creative evaluation before full human testing
AI DELIVERY
Prediction engine designed and built with an offshore engineering team
REAL OUTCOME
A scalable, compliant product that complements human insight
CASE STUDY
FROM COMMERCIAL PROBLEM TO SCALABLE PRODUCT
CONTEXT
A need to evaluate creative at scale, earlier
Campaigns are not built around a single video creative. With the rapidly moving technology landscape it is becoming quicker to create many ad versions, different edits, storylines, formats, and lengths.
Testing each version with full human research would be too slow and very expensive.
The opportunity was to introduce a first layer of evaluation. A way to assess and rank a high volume of finished creatives quickly, enabling faster decision making before investing in deeper testing.
CHALLENGE
Balance speed, cost, and trust in an AI driven product
The challenge was not just building a model. It was creating something that could operate at speed, scale and have outputs trusted by the business and customers.
This required:
- Creating an AI model and turning the results into something understandable
- Ensuring outputs were useful for real decisions
- Delivering quickly through an MVP approach
- Maintaining compliance with evolving regulations, including the EU AI Act
The product needed to sit between speed and confidence. Fast enough to scale. Reliable enough to use.
APPROACH
Leading delivery from concept through to MVP and iteration
I took the commercial idea and led the delivery of the product. This included:
- Identifying and structuring the training dataset
- Working directly with an offshore AI engineering team to build the prediction engine
- Coordinating onshore teams across product, legal, and commercial to ensure alignment
Once live, I supported:
- Evolution of the product from feedback captured by product management
- Translation of technical behaviour into clear product messaging
This ensured the product was not only built, but usable, explainable, and aligned to real customer needs.
OUTCOME
A scalable early signals product with a clear commercial role
AI Predictions became a high volume, lower cost entry point into the Test Your Ad ecosystem. It enabled:
- Rapid ranking of multiple creative options
- Earlier stage decision making at scale
- A clear path into full human testing for validation
The product did not replace human insight, it complemented it. This created a strong commercial model — speed and scale at the top of the funnel, confidence and validation through human testing.
The result was a product that connects AI capability with real business value, while remaining compliant, explainable, and trusted.
DELIVERABLES
WHAT I DELIVERED IN THIS WORK
END TO END DELIVERY
Took the initiative from concept through to MVP, working across commercial, product, and engineering.
PREDICTION ENGINE
Worked with offshore AI engineers to design and build the model, including defining training inputs and structure.
MVP AND ITERATION
Delivered a working product early, then refined it through real customer feedback and usage.
AI COMPLIANCE
Ensured the product was explainable, governed, and aligned with regulatory requirements including the EU AI Act.
CONTRIBUTION
HOW I SHAPED THIS WORK
Turning a commercial problem into a deliverable product
Bridging technical AI concepts with customer facing language
Reducing ambiguity in how the feature should be understood and used
Making complex AI behaviour understandable and usable
Supporting alignment across product, engineering, and stakeholders
Ensuring governance, compliance, and trust in the product
PRINCIPLES
HOW I APPROACHED THE WORK
START WITH THE COMMERCIAL PROBLEM
The product exists to solve a real need, not to showcase technology.
DESIGN FOR SCALE AND SPEED
The value comes from handling volume quickly but delivering quality outputs.
USE AI AS A FILTER, NOT A REPLACEMENT
AI supports decision making. It does not replace human judgement.
CLARITY BUILDS TRUST
Products like this must be explainable, governed, and transparent to be used with confidence.