Data-driven approach

What is a data-driven approach
Why is a data-driven approach needed
What research is conducted for decision-making
What data is collected
How data is collected

What is a data-driven approach

A data-driven approach is a concept in which management decisions are based on data and facts rather than intuition or the opinions of individual employees. Translated from English, this term means "data-driven," and it is also referred to as a culture of decision-making based on factual information. The main idea is that human errors are inevitable, while data and facts remain constant. Relying on the opinion of the founder, manager, or other team members means risking making decisions based on subjective opinions.

An example can be a situation that occurred at the company "Avito." They decided to change the interface of their application based on employees' opinions that it was outdated and inconvenient. However, the results of testing the new interface showed that users started placing fewer orders. A deep analysis and interpretation of the data helped to uncover the true reasons, confirming the importance of focusing on facts.

Why is a data-driven approach needed

The main goal of using a data-driven approach is to improve the effectiveness of decision-making. For example, in a pizza delivery company, there are often complaints about delays in order deliveries. Different employees suggest their solutions to the problem:

Hire additional couriers;
Replace current couriers with new ones;
Implement a motivation system for couriers;
Change the transport used by couriers;
Optimize communication between production, couriers, and customers;
Encourage workers to prepare pizza faster;
Change the pizza preparation process to reduce time.

A data-driven approach allows for a justified selection of the most effective solution based on research results, which may include:

Measuring the time required for pizza preparation and delivery;
Creating a test group of new couriers;
Introducing a trial motivation system;
Changing the pizza preparation process without compromising quality.

The results of such research provide objective data that can be measured in minutes or the number of orders, helping to identify the most optimal solutions.

What research is conducted for decision-making

To ensure the correctness of decision-making, companies conduct various types of research. The main types include:

Quantitative research: The results of such research are measured in numbers. An example can be A/B testing or surveys without open-ended questions. The advantage of quantitative research lies in its massiveness, as it can involve hundreds or even thousands of respondents.
Qualitative research: This provides hypotheses and insights that require further verification using quantitative data. This includes focus groups and interviews, allowing for deeper information gathering.

Research can also be classified into desk and field studies. Desk studies use already available data, allowing for quicker results, while field studies generate data from scratch, providing more accurate results for specific tasks.

What data is collected

For data-driven decision-making, the collected information must be:

Relevant: Metrics should correlate with business goals;
Objective: Data should be collected according to uniform rules and in a single format;
Understandable: Information should be accessible for interpretation not only by specialists but also by other employees.

At the initial stage, all possible metrics may be collected to later determine which of them are most useful for decision-making.

How data is collected

Implementing a data-driven approach involves several stages. At the initial stage, data may either not be collected or be stored in simple spreadsheet editors. The next step is to systematize and segment the data, which involves creating a common methodology for data collection.

Once data management becomes centralized, companies implement automated systems, such as BI systems, which assist in analysis and report generation. The final stage includes creating predictive models that use artificial intelligence to forecast future indicators based on the analysis of previous data.