Venturing into a new Business Intelligence (BI) project can be as exciting as it is daunting. For many people, BI is something they know they need, but they don’t quite know where to get started. One of the things I see time and time again are enterprises and software vendors that jump right into developing their first BI solutions without the necessary planning in place. For example, these companies neglect to:
- Establish a clear picture of what the problem they are trying to resolve is
- Understand who their actual target audience is or what they want
- Outline a clear set of deliverables to target completion
All of these issues lead companies down a spiraling path of endless development, redefining of requirements, and wasted effort that rarely results in successful solutions.
Throughout my time working with various companies both new and experienced in BI, I have developed a basic guide to ensure that any BI project is a successful one. This approach aims to minimize endless documentation and instead focuses on gathering a clear understanding of the problems that need solving along with the end-users expectation. So, without further ado, here is my 4-step guide to implementing a successful Business Intelligence solution.
Step 1: Understand the Business Needs
Outcome of this Step: Determine a list of business requirements
Before you start getting into ‘how’ you are going to implement your BI solution, you need to understand ‘why’. The purpose of every solution is to solve a problem, and Business Intelligence is no different. Without an understanding of the problem you are trying to solve is, there is nothing to guide your implementation. By understanding the problem, you can begin to outline the requirements of your solution.
How to get there: Understand your audience
At the end of the day, you are building a solution for your end-users and stakeholders. If the solution does not meet their expectations, then it won’t be considered a success. That’s why it’s important to understand who your audience is, what problems they currently have, and what they are hoping to gain from a BI solution. I recommend meeting with members of each of your user groups (face-to-face whenever possible) and having these discussions to better understand their needs. During these meetings, you’ll want to ask questions such as:
What problems and pain points are you hoping to alleviate with a BI solution? As mentioned before, without knowledge of the problems, there is no way to develop a solution. Who better to tell you their problems than the people you are building the solution for?
Which current business areas are lacking visibility? Sometimes it goes beyond knowing what problems you are currently facing, and instead recognizing there are problems that you may not even be aware of. Business Intelligence opens the floodgates when it comes to possibility, and that often means being able to get clarity on avenues that were previously unavailable.
What kind of experience are they looking to have with a BI solution? You can develop what you believe to be the greatest dashboard(s) of all time, perfect for being viewed on large, mounted monitors. After days of work, the deadline comes, and you present your baby to the stakeholders. Within a couple of seconds, all your hard work is torn to shreds as they utter, “Can you make this work on my mobile devices?”. I have seen this specific scenario and many others like it happen countless times. I can’t stress enough how easy and important it is to avoid such wasted efforts by simply talking to your audience, understanding the type of experience they are looking to have, and making sure your solution matches their expectations.
Step 2: Keep it SMART
Outcome of this step: Determine a list of SMART objectives
One of the best ways to ensure the success of a project is to determine what its objectives are. Objectives provide you with a clear set of goals that you can guide your solution towards. By meeting those objectives, you can say that your solution achieved exactly what it is supposed to.
With that being said, any list of objectives just won’t cut it. For example, let’s say you proceed with the objective of making more sales. How do you actually determine if an objective like this has been achieved? Are you trying to make more sales compared to the previous year? Are you trying to make more sales during the month following the implementation of the solution compared to the previous month? If you make one more sale than in the previous period, does that count? This is where SMART objectives come in. SMART stands for:
With SMART objectives, we can eliminate any ambiguity, progress becomes much easier to track, and unrealistic objectives are avoided.
Now, let’s revisit the example objective I had, but this time I’ll make sure it meets the SMART objective criteria: Our objective is to increase the total year-end sales dollar amount by at least 10% compared to the previous year.
As you can see, this is a much more specific objective. I have a timeline to achieve it by (end of the year), and a clear target I am trying to hit (110% of last year’s total sales). Any objectives that do not follow the SMART guidelines should not be considered objectives of your project until they have been revamped to do so.
Step 3: Determine your Deliverables
Outcome of this step: Determine a list of project deliverables and sub-list of KPI definitions for each deliverable
At this point, you should have a solid understanding of why you need a BI solution. Now, it’s time to think about how you will get there. Before any development begins, you need to know what you are actually developing. Are you developing a dashboard, multiple dashboards, or an automated report? Are you preparing data for self-service analysis? A good BI tool will offer you limitless possibilities in terms of what you can create, so before you get lost in the development process, determine what the deliverables of the project will be. It’s important to distinguish and separate your project objectives from your project deliverables. Remember, objectives focus on elements that are external to the project, while deliverables focus on elements internal to the project.
How to get there: Use what you know
You’ve spoken to your stakeholders and target audience and based on these discussions you should have an understanding of how many different user groups you are dealing with, what problems they are looking to resolve, and what type of experience each wants. Use this knowledge to determine what deliverables will best achieve the results your audience is looking for. Think about how much overlap you have between each of your user groups and their requirements. Can some of these be addressed within a single deliverable, or are they separate pieces to the overall solution?
For example, your target audience might include members of your Business Development and Finance departments, as well as upper management who oversees both those teams. The individual members of those two departments are looking for a way to monitor and track their specific and detailed day-to-day operations and performance, while upper management wants to receive scheduled updates on the overall performance of both departments. Upper management also wants the ability to see the detailed, day-to-day information, but only when there are any issues. In this case, your deliverables could be:
- A detailed-level Business Development dashboard
- A detailed-level Financial dashboard
- A scheduled, high-level report on both departments
At this stage, you will also want to start thinking about which KPIs you’ll include on each of your deliverables. As this is a Business Intelligence solution, the driving force of the solution’s information will be the KPIs that it presents. Now that you know what deliverables you will be providing, think about which KPIs can best address the business requirements and SMART objectives associated with your solution.
Remember, real-estate on dashboards and reports is precious, so every KPI should serve a purpose. If you cannot tie a KPI back to one of the solution’s business objectives, then it probably is not needed. Then, for each of your deliverables, come up with a sub-list of the KPIs it will include, along with the business objective each KPI will address.
To make the development of each KPI more seamless, putting the right amount of planning into defining each KPI is critical. Here are some things I recommend considering when defining your KPIs:
- What measures and dimensions should the KPI include?
- How can the KPI be filtered?
- Are there any targets or comparisons to be made?
- Are there any conditional states that signify if the results are good or bad?
- Where is this data coming from?
The idea is that these definitions should be clear enough that even users without a technical background should be able to easily understand them. By creating these definitions, the transition to developing your KPIs will be a smooth one, as it will simply be a matter of applying what has already been defined.
Step 4: To the Drawing Board
Outcome of this step: Create mockups for each deliverable
The fourth and final step to this guide is to create mockups for your deliverables. In this phase, you’ll want to begin by laying out each deliverable’s set of KPIs in a drawing. Yup, a physical drawing! These drawings can start out roughly (treat them as you would a brainstorming session) as you’ll want to easily be able to make changes. I prefer working on a whiteboard at first.
To simplify this step, I like breaking it down into two parts:
Determining your data visualizations. At this point, you should have your KPIs well defined; you should know what they are going to include and what business objectives they are going to serve. This is where you will want to utilize your data visualization skills to determine the best visualization option to make understanding the data as easy as possible. A user should be able to tell if a KPI is meeting its objective or not within a simple glance, and the right visualization will allow for that.
Design the layout of your KPIs and user interface (UI). By now you should know who the target audience of each deliverable will be, so use this information to create a suitable design. Think about each group of users that will be interacting with specific dashboards or reports, what their skill-sets are, and what their user stories would be. Use this information to create mockups that address each user group’s desired user experience. You should also know how many KPIs your deliverable will have, as well as what types of filters you will want to include. Based on these details, you can determine how to divide that available space and whether you want to have a scrolling view or organize items onto different layers.
This stage is one that tends to be revisited and reworked quite frequently, so be sure to communicate with your stakeholders before making any final decisions. Any feedback that they provide can go right back into adjusting your mockups, which is probably much easier than trying to adjust a finished solution
The point of these steps is to guide you in the right direction when starting your next Business Intelligence project. What’s important to remember, is that this guide provides you with a solid plan to base your development on, rather than going in blindly. Many organizations might already have steps in place to get the same results, however, these are ones that I have seen lead to successful BI solutions countless times.
Are there any steps in this guide that you have put into action before? Are there steps that you feel are missing that you can’t start your BI projects without? We’d love to hear from you your success following these steps along with other approaches you’ve followed in the past.
About the Author
Dany Nehme is a Senior BI Consultant at Dundas Data Visualization. A key component of Dundas’ Professional Services team, Dany provides guidance to both veteran and new Dundas BI clients, enabling them to implement Dashboards and custom solutions in a best practices manner.Follow on Linkedin