From data to decisions: how to get started with analytics and business intelligence
Many companies collect data they never use to make decisions: sales in the accounting system, orders in spreadsheets, customer conversations in email or on WhatsApp. Turning that data into useful knowledge is a process, and it pays to work through it in stages.
Why it’s worth it
Some of the strongest evidence comes from the U.S. Census Bureau. Using data from a representative sample of manufacturing plants, Erik Brynjolfsson (MIT) and Kristina McElheran found that data-driven decision-making nearly tripled between 2005 and 2010, from 11% to 30% of plants. In a companion study, plants that adopted it showed, on average, 3% higher value added than those that did not, after controlling for other factors of production. That is an average observed in U.S. manufacturing plants, with differences by type of company, not a guaranteed gain for every organization.
That 3% does not come from buying a tool. In the study, a plant was considered “data-driven” when it had data available, actually used it to make decisions, tracked its key performance indicators and set short- and long-term targets. In other words, the benefit comes from a management habit, not from software.
1. Assessment
Which decision do you want to improve, and with what data? Without a clear business question, a dashboard is just decoration. A good question has three parts: the decision (for example, how much inventory to order each month), the person who makes it and the data that would support it. If seeing the data wouldn’t change what anyone does, you don’t need that metric.
2. Organize and clean the data
Before any analysis, you need to know where the data lives, who maintains it and how reliable it is: duplicate customers, dates in different formats, products with several names. This stage is unglamorous, but it determines the outcome. According to a Gartner survey published in February 2025, 63% of organizations either do not have, or are unsure whether they have the right data management practices for AI. In the same release, Gartner predicted that through 2026, organizations would abandon 60% of AI projects unsupported by AI-ready data.
3. See what is happening (descriptive)
Dashboards and key performance indicators (KPIs) that show the real state of the business: sales, funnel, operations. This is where patterns, trends and anomalies that were previously invisible come to light. A few well-defined metrics are worth more than dozens of charts: each one should have a written formula, a source and an owner.
4. Anticipate (predictive)
With a solid foundation, statistical and machine learning models make it possible to anticipate demand, risk or behavior and to move from reacting to planning. A model is only useful if its prediction reaches the decision-maker in time and is regularly compared with what actually happened.
Start small and measure
Choose a high-impact use case, set a baseline and compare before and after. The goal is not the flashiest dashboard but a demonstrably better decision. For example, if the problem is stockouts, measure how often they happen over a few weeks, implement an inventory dashboard with alerts and measure again using the same method.
Common mistakes
- Starting with the tool instead of the business question.
- Building dashboards no one reviews because they are not tied to a meeting or a decision.
- Jumping to AI before getting the data in order.
- Not measuring the starting point, which makes it impossible to prove the result.
At Wasivia, we turn your data into decisions: analytics, dashboards and predictive models. Request a proposal.
References
- Brynjolfsson, E. and McElheran, K. (2016). “The Rapid Adoption of Data-Driven Decision-Making.” American Economic Review, 106(5)
- Brynjolfsson, E. and McElheran, K. (2016). “Data in Action: Data-Driven Decision Making in U.S. Manufacturing.” U.S. Census Bureau, CES 16-06
- Gartner (February 26, 2025). “Lack of AI-Ready Data Puts AI Projects at Risk”
- Gartner IT Glossary — “Predictive Analytics”
