❗ Never start with the solution (or the data), always start with the problem.

👉 Companies often want to use ML because they think they need to - and they do to stay competitive but still they need to make sure to use it in the right way such that it generates a business value (not just do it for the sake of going with a hype).

💡Here is a step by step guide how to add value with ML:

1. Define the problem you want to solve Describe and define exactly what the problem or pain point is that you want to tackle. This can be difficult and design thinking workshops can be a good way to narrow down the pain points and problems from a user perspective.
    
2. Quantify the impact Decide which metric is suitable and use it to estimate the impact of the implemented solution, i.e. saved hours of manual work, reduced customer defect.
    
3. Identify potential solutions List potential solutions starting by pure heuristics up to more sophisticated solutions, try to estimate the impact in terms of your chosen metric of a simple solution vs a more sophisticated one (run small experiments if needed).
    
4. Quantify the effort Estimate implementation effort for each solution to understand the investment.
    
5. Make decision Decide which solution has the best impact to investment ratio.
    
6. Start small and iterate Narrow the solution down to part of the problem and then extend it, i.e. start with one language, customer group, market.