Data-Driven Decision Making: Top 9 Best Practices

Datamation content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.

The phrase data-driven decision making – certainly popular in the field of data analytics – may seem redundant. After all, nearly everything is driven by Big Data or we wouldn’t have petabytes of databases in public and private data centers around the world.

So what exactly does it mean to be “data-driven?” It’s quite straight-forward. Data-driven decision making (DDDM) is the process of making organizational decisions based on actual data analytics rather than intuition, anecdote, or observation.

Business intelligence (BI), another popular data term, is entirely data-driven decision making. Using enterprise data analytics applications like TeraData or Microsoft Power BI, IT managers and business people process data, extract facts, figures, and patterns from that data, and make decisions based on the cold, hard facts, not gut feelings.

DDDM is the art and science of using facts, metrics, and other data to guide strategic business decisions to meet your company’s goals and objectives. Done right, DDDM helps you make better business decisions and spot strategic opportunities.

So how do you do DDDM right? You start with making it the norm. Your organization needs to make data-driven decision-making standard operating procedure. Sure there is room for gut instinct but first and foremost you need a culture of analytics. That’s why analytics has become so predominant in technology. As data has exploded, so has the opportunity for insight from that data, whether it’s through business intelligence, Big Data, data warehouses or data lakes.

Data mining results in general fall into two distinctive types: qualitative analysis and quantitative analysis, and both are equally valuable to making a data driven decision.

Quantitative data analysis is what DDDM is all about. It is measured analysis that focuses on numbers and statistics and other elements such as median and standard deviation. Qualitative analysis focuses on data that isn’t defined by numbers or metrics, such as images, videos, and social media.

Qualitative data analysis is observational while quantitative is factual. Both qualitative and quantitative data should be analyzed to make smarter data driven business decisions.

The good news is that employees across the board can participate in DDDM. While some of the more arcane data science disciplines belong to data scientists with advanced degrees, there are plenty of DDDM-related business applications for mere mortals, starting with Microsoft Excel.

Data-Driven Decision Making: Best Practices

Beyond apps, companies have to develop data skills through practice and application through best practices and business models with security and governance to watch over things. To effectively utilize data, professionals must take several steps:

1. Know your end game

If you don’t know your destination, how can you get there? That should be the first step in any DDDM scenario: ask yourself what are you trying to solve. Identify and understand your goals thoroughly. You need to do this before you begin collecting data so you know what data to collect and not to collect.

To get the most out of your data, companies should define their objectives before beginning their analysis. As Sun Tzu said in The Art of War, “Victorious warriors win first and then go to war, while defeated warriors go to war first and then seek to win.” Set a strategy to avoid falling into traps through Key Performance Indicators (KPIs) as measures of success or failure.

2. Coordinate among teams

Your DDDM project will involve at least two stakeholders: the business unit looking for insight and the IT people who will run the computing. But there may be others with a vested interest. Other departments or C-level executives might want to know the results as well. And adding new people might mean changes in the data collected. A new stakeholder could meant a new data variable added to the mix.

3. Democratize the process

We all have unconscious biases and we all have blind spots. We might even be guilty of seeing the data we wish was there instead of what’s really there. Therefore, make this a team effort and bring multiple eyes to the project. They will have their own biases, sure, but hopefully they won’t be the same as yours.

A 2010 McKinsey study of more than 1,000 major business investments showed that when organizations worked at reducing the effect of bias in their decision making processes, they achieved returns up to 7% higher. By eliminating bias, you open yourself up to discovering more opportunities.

4. Clean and organize your data

According to Gartner, data scientists spend 79% of their time collecting, cleaning, and organizing data and only 20% actually performing analysis. Not surprisingly, this is the least favorite part of the job for data scientists but it must be done.

The term “data cleaning” is the process of preparing raw data for analysis by removing or correcting incorrect, incomplete, or irrelevant data. In data warehousing, this is known as “schema on write,” where you apply such filters before you store it. The process involves creating a data dictionary, a table that defines each of your variables and translates them into what they mean to you in the context of this particular database. Once you have a dictionary it is available for reuse on other projects.

5. Find the data needed to solve these questions

Look at the data you have already gathered and try to focus on your ideal data, that which will help you answer the questions you are asking. Once you determine the data needed, check if you already have this data or if you need to set up a way to collect it or acquire it externally.

6. Perform basic statistical analysis

If you are new to analytics and DDDM, it really isn’t a good idea to involve a multi-petabyte database for your first project. Start small and learn. You are testing your models to see if they are providing the answers you need. Testing different models such as linear regressions, decision trees, random forest modeling, and others can help you determine which method is best suited to your data set.

From there, you can come up with three types of reports:

  • Descriptive: Just the facts.
  • Inferential: The facts, plus an interpretation to provide context.
  • Predictive: An inference based upon results.

7. Draw conclusions

The last step in data-driven decision making is coming to a conclusion or conclusions based on the findings. The conclusions you drawn from your analysis will be the basis on which your organization will make an informed business decision and plot strategy moving forward. The final step of DDDM is always the human element.

8. Present the data in a meaningful way

Digging through data is what computers do but humans want something more than rows and columns of numbers or their eyes glaze over. With the help of a great data visualization application you can present the data in a meaningful way to people less technologically skilled. Apps like Databox, Zoho Analytics, Tableau, Infogram, ChartBlocks, Datawrapper, and many more provide an easy to use GUI environment to tell your data story.

9. Revisit, review, revise, and reevaluate

Once you have your models and data dictionaries in place, you can’t rest on your laurels. Do not be afraid to step back and to rethink your decisions, review your models revise them, and ask how can I do this better? Optimization is always possible and nothing is bug-free. Keep revising your work and make it better.

Subscribe to Data Insider

Learn the latest news and best practices about data science, big data analytics, artificial intelligence, data security, and more.

Similar articles

Get the Free Newsletter!

Subscribe to Data Insider for top news, trends & analysis

Latest Articles