A Three-Step Guide for BI Professionals
Launching a new business intelligence initiative or project can be tricky: while we’re staunch believers in agile BI and quick wins, it’s important for both the BI person and the business executives to first align their needs and expectations, and to understand what the organization hopes to achieve. The good news is, an efficient business analyst can get it done in a day or two.
After all, it makes sense that before you dive into schemas, calculations, and charts, the first thing you’ll want to do is actually understand what the business hopes to achieve. This might sound bafflingly obvious, but you’d be surprised how many times I’ve seen organizations skipping this step and going straight to building their KPI dashboards, without stopping for a second to think whether these KPIs are even relevant to the current project.
Often this happens due to an executive or analyst who has some preconceived notion of the end result, acquired from a previous company or project, which does not necessarily apply to the matter at hand. Remember: The process should always start with the business and serve the business. The metrics need to fit the organization and not the other way around. You have to be flexible enough to accommodate the BI solution that the business actually needs, rather than the one that’s easiest for you to create.
So without further ado, here’s how to put together a successful BI project plan:
Step 1: interview key stakeholders
What you’ll need: undivided attention
Time to completion: 1 – 2 days
Deliverables: summary of business requirements
When doing business intelligence project management, the best way to understand what the business hopes to achieve through the BI project at hand is through face-to-face meetings with the relevant stakeholders (or, less preferably, via phone or Skype). These would include whichever executives, managers or analytical users who will actually be looking at the data on a regular basis.
Don’t skip this step or make do with written specifications! These few hours of meetings will make your job that much easier down the line and will greatly increase the chances of the project being a success.
Questions to ask during the interviews:
The wh- questions: Why is a particular dashboard needed? Who will use the dashboard, and who will receive its outputs? Where (and on what device) will they do it? When will the dashboard be used?
Current and desired decision processes: How are decisions currently made? How would they like to make decisions in the future? Which data is currently missing, or hard to access, and how would it affect the decision making processes?
Pain points: What did they always want to know but couldn’t find out? Why is data too difficult to find or analyze? Where are analytical\IT resources currently going and how could they be used more effectively?
Step 2: to the drawing board!
What you’ll need: a whiteboard and markers, pen and paper, or Powerpoint\Visio
Time to completion: 2 – 3 hours
Deliverables: list of crucial business questions to answer
Once we’ve finished interviewing our key stakeholders and understood what they expect the business intelligence project to look like, we’ll want to start visualizing — not the data (yet), but the processes themselves. What we’ll be drawing here is not graphs, widgets or visualizations, but a flowchart of sorts.
Build a diagram illustrating the business process workflow for each relevant process. It is at this BI planning stage that we understand whether we need multiple dashboards, schemas, etc. By the time you’re done, you should be able to describe the way in which the organization makes decisions, and how it measures the quality of these decisions (e.g.: Marketing is focused on campaign performance, whereas Customer Success is concerned with account health, the CEO wants to see ACV, etc.).
Tips for creating your business process flowchart
Top-down approach: start from the most high-level KPIs, the ones that might concern the most senior decision maker – ACV, growth, recruitment, etc. The lower level metrics should be derived from the way you measure the high-level ones.
Keep it simple: Even for bigger projects, you should aim to end up with a number of metrics that you need less than two hands to count. Your guideline should be: more information, less indicators. Try to identify the one measure that matters (e.g., weighted average is always better than 5 different numbers).
Step 3: Mock things up
What you’ll need: same ingredients from step 2.
Time to completion: 2 – 3 hours
Deliverables: draft of your first dashboards; list of facts, dimensions, and filters
For simpler projects this step might be considered optional: if the business questions are relatively few or simple, you can go ahead and fire up your business intelligence tool and start modeling your data and dashboarding. However for bigger projects I always suggest creating an initial draft of the dashboard first, so you’ll remember what the finish line looks like (your plan for how to get there is the deliverable from step 2) — and also to ensure you get sign-off from the relevant stakeholders.
It’s at this stage, after we’ve identified the business questions, that we’ll want to start getting our hands dirty: where is the data coming from? What are the data sources, how will I connect to them (with Sisense this is incredibly easy using our built-in connectors)? Will you need to connect to unstructured data sources? Which data is not currently available and will need to be calculated?
The answers to these questions should give you a clear idea of your dimensions, facts, and filters. Then you create a mockup of which widgets and data visualizations you’ll want to apply in order to best present the data.
Once you have a mockup you’re happy with, go back to the stakeholders and get their approval: after all, they’re the ones who will be regularly viewing these dashboards, so they should understand why they look like they do and how they represent the relevant data. And for you, getting early feedback will save the time you would spend making adjustments and changes after already building the dashboard. Once you have approval — congratulations, you’re good to go!
Things to Consider Before your BI Project Plan
Before you can buckle down and follow the three steps from above, you need to make sure the tool you’re using is able to handle your organization’s needs.
What are your biggest pain points (either with the tool you already have or that you wish to resolve by integrating a tool)? Maybe you are looking to cut time and resources by moving from manually generated reports to automated, drill-down reports that you can set up once and rely on to be accurate and on time, every time. Or, maybe you are looking to make sure your organization has a single source of truth in order to make sure everyone, across your organization, is on the same page.
If you’re an enterprise trying to tackle digital transformation and a move to the cloud, then you have your own challenges ahead. You’ll want to ensure any analytics solution you bring on board is cloud agnostic. This way, your developers won’t be confined to a single cloud structure. Instead, your analytics platform should be built on a containerized microservices architecture for complete flexibility.
Whatever your challenge is, you also should pay attention to how much a BI solution is going to cost. There is a long list of things to take into account when figuring out the total cost of ownership of BI. But to begin, make sure the platform you’re evaluating does everything you need and has the ability to handle all of your data as you grow. Next, try to estimate how much you’ll need, internally, in terms of budgets and resources, new staff, and opportunity costs of moving current employees off of other projects to make sure implementation runs smoothly.
Closing (and a disclaimer of sorts)
Needless to say, no two organizations or BI initiatives are created equal. For some, this entire process can take an hour or two, and for others, it might take weeks. However, I would strongly recommend following this framework for each and every BI project you embark on, whether there is already a business intelligence system in place or not — because when you go in with a plan, you’re that much more likely to come out with a win.