Have you ever worked for a big company that wants to utilize Data Science, but doesn’t know how? Have you ever been given a large dataset with multiple data sources without a clear understanding of what is expected at the end? Does that sound familiar? This is a story on how to bring value to the client when the requirements are not clear from the very beginning. The client is a large investment fund of one large western country that wants to assess the investment attractiveness of startups. By itself, this is a good business question, but rather vague Data Science one. How to transform it into more mathematical terms? How to find a target variable in the pool of sparse data? If you've ever worked with a hard-to-communicate client or noisy data without a target variable - this presentation might be particularly interesting for you. It brings insights and possible solutions to the above-mentioned problems and shows how to survive in a difficult professional setting.