Every company is full of data. They look around and see the industry is innovating. Executives listened to their opinions on the artificial intelligence strategy.

Data is a valuable asset of the company

Management believes competitors are adopting artificial intelligence solutions and incorporating key initiatives into their addressable markets. Because of all these background noises, the management's direct response is to conclude that we must do something about our data and let us hire some data scientists. This is no different than when mobile devices were very popular ten years ago. Our idea was to hire a mobile developer who could launch a mobile app for the company. You assume that machine learning is the solution and is looking for problems that it can solve.

What are the problems we have to solve?

Machine learning is not magical. Machine learning is a solution. People must be clear about what we are solving.

Beware of the temptation to find a problem that can be solved with popular techniques like machine learning. Or find the problem without asking a question, is the problem enough to solve and invest resources. Machine learning is like a drill bit. The drill bit you use depends on the problem you are trying to solve.

Define the problem you are trying to solve, and the business outcomes you expect to achieve and the benefits you are trying to find for your customers are very important. Once you have clearly defined the problem, you can start thinking about the data we need, the model to be created, the algorithm to use, the insight based on predictive insight, and the actions taken.

Product manager collaboration

Defining the problem to be solved is the issue that product managers and business stakeholders need to establish. Product managers should conduct initial customer interviews to understand the key pain points of the customer to verify the problem they are working on.

Gain insight into customer intent

When getting customer feedback, the root cause of the problem must be found based on the use cases stated by the customer. For example, a customer might ask me can I export this data (predictive insight) to a CSV file?

We can think that we need CSV data export. Maybe we can dig deeper, and they might want to do this because they want to load predictive insights directly into their CRM application so they can take action. Therefore, the real feature is to drive some operations through deeper integration with the CRM system, rather than exporting the data to CSV for predictive insight. The role of the product manager is to dig deeper into this insight.

Forecasting Framework: An example of predicting business problems

Product managers consider a framework for adding predictive insights by looking at common/popular issues with good ROI.

The example/frame below is not the model to be built. They are examples of popular problems that we can solve.

Simple sample application

Suppose you have an existing product that is sold in the accounting space. The problems you can solve may make your solution separate from your competitors. These problems are like


  • Does the customer pay by default?
  • Will this customer pay the invoice on time?


  • What is the total expenditure for next month?


  • Segment customers based on demographics and buying behavior to better connect with customers


  • Does this invoice look strange?

If you look at the above, we are not talking about what model to build, what algorithm to use or what data we need. We are identifying which issues to solve and which are valuable to your customers and helping to drive your business.

If the above issues add value to your customers, please verify with your customers. Verify if these contribute to your business outcomes. Once you have some good feedback, you can move on to the next step and simply use the POC to verify the idea of ​​a product market fit.

Final Thoughts

Start with the problem you are trying to solve. Determine the benefits it brings to your customers. Define how it can help drive business outcomes. Don't be afraid to try. Experiments help get early feedback, help save time and money, and help correct product vision and direction.

This article is transferred from medium,Original address