June 14, 2018

4 Key Steps to a Successful Business Intelligence Project

We were recently engaged to help a client propel their various Business Intelligence (BI) initiatives forward. Before beginning the “nuts and bolts” of BI development, this client rightly wanted to perform some initial analysis of their current reporting and data infrastructure to determine the best path forward. The client brought in two highly-skilled and BI-specialized Business Analysts (BAs) to help with the data discovery and definition process.

One of the client’s initial concerns was that our BAs wouldn’t have enough specialized knowledge in the BI space versus a traditional business analyst. They had struggled with several “stops and starts” on previous projects with their own internal BAs. The client also wanted to ensure that we had a disciplined process in mind for helping to ensure a successful delivery of their BI goals.

As a member of the BA team, I made the following 4-step process to ensure a successful BI project implementation:

Step 1: Understand the Business Context (High Level Scope)

As we kick off any BI initiative, it is critical to understand the current state and business context of the data. To do this, we put together a context and a data flow diagram to confirm our understanding of the current state with key stakeholders. Along with this initial discovery, we also sought to understand the following:

  • What kinds of business questions/problems are you trying to answer/solve?
  • What business processes are most critical to measure?
    • Why is this information needed? (Think about the broader process)
  • What are the current goals of each department?
  • What are the sources of company data?
  • Where does the company’s data originate from?
  • How does the data flow between departments? 
  • What are the data relationships? (e.g. Does the data need to feed into other processes?)

Understanding the high-level context is critical for laying a good foundation and identifying department dependencies and relationships before diving into any BI project. Time spent in this up-front analysis will result in money saved later in the form of re-work or missed requirements.

Step 2: Understand the Current Users of the Data (Stakeholder Analysis)

As we dive deeper by interviewing key stakeholders, we try to better understand the current users of the data. We conduct a thorough Stakeholder Analysisto make sure we are interviewing key users from each department and eliciting the critical requirements for success. As part of this analysis, we create User Persona sketches of our main customer types to help us confirm our understanding and ensure that any data solution we recommend/implement will be suited for and adopted by our users. 

We seek to answer the following questions in this Stakeholder Analysis:

  • Who are the key user groups of this data/report?
  • How are they currently utilizing/interacting with the data? (What form)
  • Do they need to do analysis of the data or just review static reports?
  • What is the level of data savviness of the users?
    • Have they done self-service analysis before?
    • Have previous data analysis/implementation projects succeeded?

Only after understanding the high-level business context and stakeholder analysis are we ready to dive into a more technical look at the data.

Step 3: Identify Key Data Source(s) and Data Model

Next, we try to understand what the system(s) of record are or the source(s) of truth as utilized by the company. In many siloed organizations, this can vary widely by department and who you talk to. Regardless, we need to understand where the data originates from so we can identify its “purest” form and then seek to understand/create the data model.

As we understand and define the data model, we are looking for the following:

  • Dimensions/Facts – key ways the data is structured. (Commonly used dimensions are people, products, place and time.)
    • What are the primary keys used to identify this data?
  • Measures – The core of the dimensional model and are data elements that can be summed, averaged, or mathematically manipulated
  • Granularity of data – finest level of detail that is implied when the fact and dimension tables are joined.
    • Is the data at the individual record level or only at a summary-level?

The answers to these questions will help us define the core data model to be used in our data reporting/BI structure.

Step 4: Other Key Data Questions

Some other critical questions that need to be answered as part of a successful BI project are the following:

  • History – What amount of historical data is available and required for analysis?
  • Distribution Needs – How do users receive the data? (PDF, Print, Email)
  • Distribution Frequency – How often? (hourly/daily/monthly/annual)
  • Data Latency – How up-to-date must the data be compared to the live source system? (minutes, hours, days, months)
  • Data Transformation – What business rules/logic are applied to the data? Are any data transformations required?
  • Data Security – Any special security restrictions? Who can view the data?

Answering these questions in the format laid out above have proven to be a very helpful guide towards completing a successful BI project. If we can be of any assistance in your BI/BA projects, feel free to reach out.

Subscribe to our Newsletter

Stay informed on the latest technology news and trends

Relevant Insights

Should You Disrupt Yourself to Accelerate Digital Transformation?

It has been interesting to watch Microsoft transition from a company that makes its money via licensing to one that...

Cybersecurity Myth Busted: Tools Are the Solution

When thinking about security, people often gravitate towards implementing various security tools, solutions, or products. If you bring up a...

Time to Reconsider MP-BGP EVPN for Your Datacenter Network?

VxLAN was defined in 2014 by RFC 7348 and has been used as a component in several SDN (software defined...
X