Quantify commercial upside based on machine learning algorithms. Implement changes quickly at scale based on recommendations.
Challenge: Ad testing is given low priority due to high manual effort involved, but also difficulty to estimate the commercial upside
Objective: Identify actionable recommendations for high performing ads + patterns and implement at scale
Solution: Hybrid approach combining the power of machine learning and expert input. Recommendations are implemented with simple click, rather than painful manual processes
What were the key challenges?
Challenge #2: Learning and optimization by Google does not take into account customer insights or business know-how by experts.
Challenge #3: Rolling out new ad ideas and testing different hypotheses is painful due to high manual workloard
Key Recommendation: Use a hybrid approach of machine learning (with transparent criteria) and expert input / creative input.