Business intelligence can open a world of insight and opportunity for your organization. It gives you access to the big data that can help you make better decisions, pinpoint problems, run your company more efficiently, and give you better insight into your customers.
Researching business intelligence software is always so inspiring. Landing page after landing page of beautiful dashboards, visualizations, and lists of benefits and features. This is because business intelligence (BI) software companies are good at marketing their products. They position their software as user friendly – so friendly that you don’t need an IT person to pull your data together! As if those complicated dashboards and visualizations are ready to be created at the click of a button. It seems too good to be true, right?
Well, it kind of is.
You see, there’s a catch to buying into business intelligence. One solution that business intelligence software doesn’t offer (and it’s the key to getting all those benefits and beautiful dashboards) is getting your data ready for a business intelligence solution.
So maybe you don’t need an IT person, but what you will need is a data scientist to prepare your data.
Unfortunately, many companies fall victim to the belief that they can purchase business intelligence software like Tableau or Salesforce and then just point the new tool to their data sources and go. The truth is, there’s a lot of analysis and aggregation involved in preparing data for a BI tool.
In fact, data scientists have been reported to spend 50-80% of their time collecting and preparing digital data before it can be used for business intelligence. This makes the hopes of an install-and-go solution very unrealistic.
The reality of the situation is that collecting data from multiple sources and then using it for business intelligence can be messy. Raw data is what you need to be working with and raw data is … well, raw. It needs to be prepared before it can provide any insights into your business operations. What do we mean by “prepared”? There are three steps to getting data ready for BI.
- Sourced (wrangled)
- Transformed and cleaned (formatted)
First, data needs to be sourced and original data needs to be collected
Your organization aggregates data from a multitude of sources – accounting, logistics, warehouse, marketing, etc. The first step in preparing your data for a BI tool is sourcing all of the data that needs to be collected, and then wrangling the raw, original data from each source.
Each data set within your organization has its own unique purpose and is therefore structured differently. On top of that, many business programs and applications are structured in a way to show end-users a summarized or simplified version of raw data.
The ultimate goal here is to clean and source the data at its lowest level – not at a high aggregated summary level. This way, you maintain the structure of any existing data model that you might want to import. This is what we mean by data “wrangling”. If you want to maintain the integrity of your data, it’s important you are capturing raw data from these sources and not the simplified versions you have been using within the application.
Second, data from multiple sources needs to be integrated
Now you’ve got multiple data sources and a lot of raw data. These data sources need to be integrated with one another – basically knitted together so each data set can be cross-referenced to discover relationships between them.
This is a very important step in preparing data, because it’s the only way you will be able to see the big picture. Let’s take a hospital for example. Hospitals collect a myriad of data day in and day out. Lab results, admissions, readmissions, claims, demographics, medical history, scheduling, etc. But just because the data is being collected does not mean that it can be connected across multiple sources. The data has to be knitted together in a way that allows each data set to reference the other. This is the only way hospitals can have a full picture of each patient across different views.
Let’s say a hospital wants to understand more about why a patient had an unexpected outcome. Looking into that patient’s continuum of care is going to require data from multiple, disparate sources, such as the hospital electronic health records (EHR), ambulatory clinics, and post-acute care providers. While this patient may have records in all these places, there needs to be a way to verify that you are looking at the same patient’s records – a common identifier that matches the information across multiple sources. This is what integrating the data means.
Third, the data needs consistent terminology
Not all data systems are written in the same language nor are they stored in the same format. For example, raw data can be in the form of numbers separated by dashes, numbers separated by decimal points, or simply written data. On top of that, raw data is typically not usable to the end user.
The last step in preparing data for a BI tool is cleaning it up and transforming it into a usable format and overall structure so that your business intelligence software can consume it efficiently.
Once your data is consistent, standardized, and integrated, it can be used to find patterns, trends, and insights within your organization. Without taking steps to prepare your data so that it is analysis ready, you will be losing out on seeing the big picture. Your data may lack integrity, and you will not be able to create those amazing dashboards or visualizations you were shown when you purchased your business intelligence software.
Have questions or comments about what you just read – and how it might apply to you? Contact us to start a conversation.