Fast, scalable early insight into advertising performance
TEST YOUR AD SCREEN - TURNING ADVERTISING DATA INTO PREDICTIVE INSIGHT
Test Your Ad Screen is an AI powered Early Signals product designed to provide fast, directional insight into how finished advertising creative is likely to perform.
THE CHALLENGE
Advertising effectiveness has traditionally relied on human testing. The opportunity was to create a faster and more scalable way of providing useful early insight, without positioning AI as a replacement for human judgement.
THE APPROACH
Bringing together product, AI, engineering and governance to turn predictive modelling into a clear, scalable customer experience.
EARLY SIGNALS
Directional insight designed to help customers understand which creative is worth taking forward
FOUR PREDICTIONS
Predictive outputs across Star, Spike, Fast Fluency and overall effectiveness
BUILT TO SCALE
AI based evaluation that can operate quickly and consistently across finished advertising creative
CASE STUDY
TURNING AI PREDICTIONS INTO A CUSTOMER PRODUCT
CONTEXT
Making advertising performance insight faster and more scalable
Test Your Ad has traditionally used human response to understand advertising effectiveness. The opportunity was to introduce a complementary AI based capability that could analyse finished advertising creative and provide fast, directional predictions based on patterns learned from historic advertising performance.
The aim was not to replace human testing. It was to create an Early Signals product that could provide useful insight quickly and at scale, helping customers decide what creative was worth taking forward.
CHALLENGE
Turn a complex prediction model into something customers can understand and trust
Building the model was only part of the challenge. The prediction needed to become a usable product with a clear customer journey, meaningful outputs and defined boundaries around what the AI could and could not tell customers.
This required aligning product, engineering, data science, commercial and legal thinking while ensuring the experience remained simple for the customer.
APPROACH
Bridge AI capability with product and delivery reality
I worked across product, engineering, data science, commercial and legal teams to shape the capability from requirements through to production.
This included defining the product requirements, customer journey and reporting structure, translating complex model behaviour into understandable product language and establishing clear governance around how predictions should be interpreted.
I also worked on the evolution of the underlying model, defining how additional creative signals could improve prediction quality, confidence and interpretability over time.
OUTCOME
A scalable Early Signals product for advertising effectiveness
The result was a working AI product capable of analysing finished advertising creative and producing predictive outputs aligned to established advertising effectiveness measures.
Rather than presenting AI as a definitive creative judge, Test Your Ad Screen provides directional evidence that supports better and faster decision making alongside human judgement.
DELIVERABLES
WHAT I DELIVERED IN THIS WORK
PROPRODUCT DEFINITION
Defined the requirements, customer journey and product framework needed to turn the prediction capability into a usable customer product.
GOVERNANCE & TRUST
Established clear boundaries around what the AI could and could not predict, supporting responsible use and customer trust.
AI PREDICTION EXPERIENCE
Shaped how predictive outputs across Star, Spike, Fast Fluency and overall effectiveness were presented and understood.
MODEL EVOLUTION
Defined the next stage of the model, introducing richer creative signals designed to improve prediction quality, confidence and interpretability.
CONTRIBUTION
HOW I SHAPED THIS WORK
Led product definition and requirements across the AI prediction capability
Translated complex AI behaviour into clear customer and product language
Established governance and responsible AI boundaries
Bridged product, engineering, data science, commercial and legal teams
Defined the customer journey, reports and interpretation of predictive outputs
Shaped the future evolution of the model through richer creative signals
PRINCIPLES
HOW IT WAS DELIVERED
AI SHOULD SUPPORT DECISIONS
Predictions provide directional evidence to support people making decisions, rather than attempting to replace human judgement.
MAKE COMPLEXITY SIMPLE
The technology underneath can be sophisticated, but customers need clear outputs they can understand and use.
TRUST NEEDS BOUNDARIES
Being clear about what AI cannot understand is as important as explaining what it can predict.
BUILD FOR EVOLUTION
The product framework was designed to remain stable while the underlying prediction capability could continue to improve.