Using a Data-to-Decisions Framework – A Detailed Guide

Data science is a great tool to have when running a business.

However, analytics will only help if it drives impact. This impact could be anything from company growth, better products, or increased revenue.

Using analytics to make decisions in your business is known as data-driven decision making. This involves collecting data, extracting patterns and facts, and making inferences.

It is definitely more popular now to invest time and resources to make a majority of your company’s decisions data-driven.

Despite this, surveys show that gut feel still factors into the decision-making process.

A major factor in this is the lack of a proper decision-making framework in the organization.

This article will introduce the BADIR framework, and how you can use it to create actionable, data-driven insights for your business.

BADIR Data to Decisions framework

The BADIR framework is a highly effective data-to-decision framework designed to solve business problems.

It’s simple to adapt and works for any industry. It aims to combine data science and decision science together into one easy-to-follow framework.

Aryng, a well-known data science consulting, training and advising company devised this data-to-decisions framework.

Today, various Fortune 500 companies for their digital transformation initiatives have adopted BADIR.

Key Features of Data-to-decisions Framework

• Provide actionable data-driven insights
• Formulate a hypothesis-driven analysis plan
• Facilitates data specification to make dat
• Insights derived from pattern recognition techniques in Machine Learning and statistics
• Present actionable recommendations to stakeholders

The Five Steps in the Data-to-Decisions Framework

The BADIR data-to-decisions framework involves five steps that must be followed in order.

Business Question

Before we do any sort of data extraction or analysis, we must first understand the context of the problem we’re trying to solve. This will help reduce the number of iterations needed down the line.

This involves asking the right questions. The framework encourages us to ask the six basic questions (who, what, where, when, why, and how).

For example, we need to make sure that we understand what decision needs to be taken.

Is this decision urgent?

We need to know when we’re expected to come up with a final recommendation.

Lastly, we need to know who our stakeholders are.

Should the data be shared with the marketing team as well as the logistics team?

How many stakeholders need to know the results of our analysis?

In effect, we try to convert very basic asks into proper questions. For example, you might have the following data request: “customer data by country, product, and feature”.

A better and more useful request should look like this: “What are the reasons we’re losing customers after launch? What actions can the sales and marketing department do to address this loss?”

Analysis Plan

After deciding on a concrete business question, our next step is to formulate an analysis plan.

We should create SMART goals. SMART is an acronym that stands for Specific, Measurable, Achievable, Relevant, and Time Bound.

Next, we should formulate our hypotheses. These are statements that we aim to prove or disprove using our data. Along with these hypotheses, we should set the criteria needed to prove each one.

We also need to look into the methodology needed during data analysis. Common methodologies include:

• Aggregate

• Correlation

• Trend

• Estimation

After deciding on the methodology, we also need to decide on the data specification.

Will we use data from the past year or all-time data?

Will we primarily be using financial data or marketing data?

These questions are important because this will make the data collection process easier later.

The final output of this step is a project plan. This includes all resources needed to run this analysis as well as the timeline for each step in the process. The project plan also specifies who the stakeholders are as well as the various roles within the team.

For example, let’s say that we have the following hypothesis: “Our company is losing customers because of a less successful marketing campaign in the past quarter”.

To prove or disprove this analysis, we’ll have to pull marketing data from the past year.

We can use correlation methodology to determine whether a metric like CTR is correlated or can predict the number of customers for each quarter.

Data Collection

Data collection is now much easier since we could describe the data specification during our Analysis Plan step. This will prevent unnecessary data from being retrieved.

This is especially important if we’re dealing with a significant amount of data since it will save time when performing our chosen methodology.

The data collection step also involves data cleansing and validation. Data cleansing refers to manipulating data to make it usable.

We need to perform data validation to make sure that the data we have is accurate.

Derive Insights

Our next step involves the actual deriving of insights from our data.

In this step, we review patterns in our data.

For example, in correlation analysis we can start with a univariate analysis which looks at the distribution of the key metrics. If applicable, we can also find out if there is a difference between a test and a control population.

Using the criteria we set in the second step, we also try to prove and disprove our hypotheses.

Finally, the output of this step should be our findings. We should present our findings regarding quantified impact.

For example, you can mention the dollar impact of a particular percentage drop to engage your stakeholders.

You might say that a percentage drop in customer acquisition may result in a $1 million revenue drop.

Recommendation

Recommendations are the most important step in the BADIR framework. These recommendations must be actionable.

They are the main reason we went through each step in this framework.

In this last step, we want to achieve multiple things. First, we have to engage with the target audience. This means that you should present short and insightful recommendations.

A credible and sound recommendation will also lead to you being perceived as an effective business partner.

Lastly, your recommendation should drive your audience toward action.

If you’ll be in charge of presenting the recommendations, it’s important to build a slide deck that has all your findings.

The creation of a slide deck is iterative, starting with all your findings, and progressively streamlining the flow of the deck.

The final slide deck should have a concise executive summary. We can add any additional information in an appendix.

Conclusion

Adopting a data-to-decisions framework is a great way to make sure that you can gain actionable insights from your business data.

Combining data science with decision science allows for a dialog between all stakeholders involved. Each step in the BADIR data-to-decisions framework leads to an effective final output: actionable recommendations.

Let us know how your business or team can benefit from this type of framework!