What We Still Need To Learn About AI In Marketing.
π Here are the key takeaways and insights from podcast episode of HBR IdeaCast
π Introduction
- The promise of AI is to make decisions.
- For marketing, companies look for profitable use cases of the technology.
- If not implemented rightly, the use of AI may cause loss to a business.
π AI and Marketing
- Many leaders are leaning towards AI for their marketing strategy without getting to know it.
- The simple use of AI in marketing is to use consumers data and do some forecasting.
- The sophisticated use of AI is to build real-time pricing models like Uber.
- They collect real-time data, analyze it, and make a decision thatβs optimal for them.
π Misalignment Problem
- To leverage the value of an AI, the prediction must be aligned with the decision the company would make in the end.
- The data science team will build a system that can predict customer churn.
- Whereas the marketing team wants to know what offers may persuade these customers to stay.
π The Street Light Effect
- The AI team is more gravitate towards predicting the behavior.
- The marketing team wants to decide to change that behavior.
- The miscommunication between both teams is causing this misalignment problem.
- AI is not causing the problem, but it is a sort of enabler.
π Decision under Uncertainty
- Companies use AI to predict the unknowns that have uncertainties.
- Companies have to understand the cost of a decision made using those uncertainties and integrate those accordingly.
π What Companies Need To Do
βThe data science team did not even know how their predictions had been used.β
- 3 step framework to help companies make better decisions using AI:
- The data science team and marketing team must work together and try to understand what problem they are trying to solve.
- Identify the wrong questions and replace those with the right questions to be asked from data.
- Analyze the data to answer those questions.
- Understand that the predictions of the AI model deviate from our real world. And how this deviation can cost a company in terms of money.
βCompute the cost of doing one thing wrong to the cost of doing other thing wrong.β
π Humanly Problems
- Humans do what they know how to do. The Street Light Effect
- Humans are reluctant to change.
- Humans are not good at showing what they donβt know.
- A good mindset is to improve by doing iterations, do not aim for perfection.