How to implement a business intelligence dashboard

Having a BI dashboard to manage aspects of your business indicates that your business operations have reached a level of maturity and stability. 

As a manager, assuming you have the technical and analytical resources at your disposal who can do the detailed work in creating a dashboard (data management and ETL, KPI analysis, dashboard implementation in your tool of choice etc), in this post I want to highlight the important challenges you will need to address in order to achieve your goal of creating a BI dashboard.  

A production grade BI dashboard that really lets you take control of your business with least repetitive effort, has two main features:

i) It is reliable, i.e., you are able to trust 100% in the insights delivered by your dashboard, and 

ii) It is low maintenance - both in terms of frequent hand-cranking (if at all) required to reproduce the dashboard for current  day/week/month/quarter, as well as low maintenance to change and move things around as you develop your thinking over course of time and as you start to scale

Two most important steps for creating such a dashboard are: 
1) Identifying the key metrics for the BI dashboard
First step in order to create a dashboard is to analyze the business needs of the dashboard's audience in order to identify the KPIs that the dashboard will need to represent, along with other parameters for example frequency of refresh. Take a look at my previous article where I've shared tips from my experience on how to go about identifying key metrics for your dashboard.

2) Choosing the implementation method of the dashboard
Now that you have honed in on the KPIs, next step is to decide on best means to implement the dashboard. Implementing the dashboard can be further broken down into two main tasks -

2a) Identifying the tool to use for the implementation 
If your organization is already using a BI tool and if that BI tool serves the kind of visualization and presentation that you require in your dashboard then its a simple choice.

If you do need to choose a tool then there are literally 10 to 20 different options available out there, with license fee ranging from free (open source) to enterprise grade with multi seat licenses for tens of thousands of $$. Perhaps in another post I will talk about these different options in more detail.  

In today's world most BI tools (be it DIY tools like Tableau and Gooddata or more old school BI platforms like Oracle BI and Microstrategy) provide similar levels of visualization - granted some are better than others when it comes to fancy interactive charts.

2b) Productionizing the dashboard refresh process 
This is really where the rubber meets the road. Its all fine and dandy to create the conceptual dashboard, but its no good if you can't reliably productionize it in a way that the whole end-to-end process can run in a near-automated fashion and still provide you with the confidence to act upon its insights. 

If your BI tool is being maintained by your IT then its likely that IT will take care of the ETL type processes, to make sure data is pumped in the BI tool at the right time of the day/week/month in order for the dashboard to be refreshed. However, there are many other things like reference data and QA checks that are not likely to be in your IT department's remit - primarily because your IT guys are not your business process and data gurus.

This is where you will need to charge your BI developers and or business analysts to manage a QA process that ensures not only the correct sets of data have been fed at the right time, but also that the data that has been fed as input to your BI dashboard doesn't contain, or worse, does contain new subsets of data (for example orders coming in via a new app channel in a new market that wasn't part of initial dashboard logic) which may potentially skew your entire metrics. 

Have a look at my other post where I've pointed out a few things you need to be wary of when using BI reporting. 

At the end of the day a good BI dashboard doesn't necessarily need to look sexy - the primary objective of a dashboard must always be to bridge the gap between data to insights to action in the most intuitive and least time consuming way. 

Happy Thanksgiving and happy reporting!