Top Down and Bottom Up styles in Data Processing, which one?

You might have this question upon starting working on a data analysis project presented to you by your manager.

Shall I start analyzing the data right away? Or I might start looking at the big picture and interviewing the management?

• Top down style: Just like the name implies, you start from the top management details and work your way down into more granular level, say you start by looking at the fiscal year reports, financial report for the organization and reaching down to the datasets finally. This style is mainly used by consulting firms, which agrees with the business setup they follow, in data science you need to understand that no matter how your report and data analysis process supports your claim, there is the management final decision which can can quite possibly be irritated of you findings.
• Bottom up style: In this style, you start from a granular level i.e. looking at data sets, process, analyze and work your way up in the system to communicate your findings. It is a style that mostly followed by people who deal directly with data. Yes as you guessed it, data analysts and data scientists. You may want to keep in mind that your findings, Piecharts, Barcharts might not impress the management.

Which one is better? bottom up or top down?

There is not style better than other, it is just what the environment setup limits you to or it could be a best practice results. In a clearer way, consulting firms tend to follow a top style as they approach the organization from the top level, while data science teams tend to start from the dataset and working their way up to the top level.

Resource: Confident Data Skills: Master the Fundamentals of Working with Data and Supercharge Your Career, Book by Kirill Eremenko