Six key areas of your BI strategy
In my previous article I discussed first steps to building your BI team. So now, as you reach for your morning cup of coffee you hopefully find yourself surrounded by good people, with a clear understanding of the business requirements driving your strategy. Grab that coffee and gather those good people together, it’s now time for you to figure out how to implement this thing.
I love workshops. You can get through so much content. If you are confident in facilitating a workshop then do so, but if it isn’t your thing find someone who can help you. Facilitation can make or break a workshop. You’ll find people within your company that are skilled and willing to help so just ask around. Plan the agenda, know the intended outcomes and use facilitation to ensure that attendees contribute, stay focused and achieve the intended outcomes. Oh, and make sure you buy donuts. The first thing people look for in a workshop environment are the free donuts. Then you have their attention!
You must leave the room with a strategy, or at least the valuable information which you require to formulate your strategy. So what areas should you focus on to get there? Planning your agenda around discussion of the below six areas is a good start.
Know your user. Your user-requirements are critical to your solution and you need to formulate a plan around those needs. Your users can be diverse or demand differing experiences so you may need to deliver a diverse range of BI solutions. There are many ways to present data and whichever you choose needs to be a great fit for the user requirement otherwise it won’t be used.
It’s tough to find the perfect fit and the user question can digress into a debate about which application to implement. I provide my thoughts on applications below, but don’t get embroiled in an application discussion, particularly at this stage. Rather ensure you understand exactly who makes up your user base and what data experience they require. Defined, guided or self-service experiences are all very different requirements.
Security must underpin your entire solution. Consult with the security team before anyone else in the organisation with regards to what it is you are trying or tasked to do and keep then engaged and updated throughout the process. Look to leverage existing security platforms wherever you can, particularly if your organisation already has a robust security platform. You do not want to be reinventing the wheel with regards to security. At best, you create unnecessary technical complexity for yourself and at worst your entire project falls apart because you’ve implemented a solution that contravenes security policies.
Split your architecture discussion into physical and data architecture.
Physical architecture relates to questions like where to house your solution, whether you require hardware or virtual ware, system specs and how your system will meet current and future needs? Who requires access to the systems and how will they access them? Are there other areas of the business generating data which you need “easy” access to in order to augment your data? The answers to these types of questions will determine where you put your kit, how you configure and access it. Think future-proof, long term and build in system fat. Consult the applicable experts (like security and systems specialists) and don’t cut corners for quick wins because once your solution is in place it’s going to be very expensive (in time and money) to move.
From a data architecture perspective, you may be transforming data from single or multiple sources and you could be responsible for delivering data to single or multiple sources. The list of factors to consider in your data design will be long and very technical, but the key is understanding the construct of the data you are dealing with (like complexity and dimensionality), who is going to use it and what it’s going to be used for. This will help you determine aspects like how consumable or scalable your data needs to be, which will help drive the design.
You must ascertain what platform/s you are going to rely on to extract, transform and load your data (ETL). In layman’s terms (AKA my non-technical mind) I like to think of the platform as the engine room which takes care of the processing, shifting and delivery of your data to the end user. Aspects like your data quantity, structure and its purpose (will it be used for flat reporting, data discovery or real-time insights) will help you choose the correct platform to drive your solution.
At a bare minimum, you are going to need test and production environments, and if your BI solution is external customer facing you will have to work very hard at keeping these environments consistent. The sooner you can get started on these environments (applying the questions you answered in the sections above) the better because environments enable you to develop solutions and you want to get going!
If you are also tasked with analytical development, where you work with cleansed data to understand and track your business progress or measure strategy execution, consider creation of an environment where you can develop reports, models and work with your analysts to surface insights. This environment should sit alongside your main data source as your “scratch patch”. Data from multiple sources can quickly be brought into this environment to create room for ad hoc, quick turn-around, user-focused analysis. The environment will help you bring the technically capable people together with those holding the business acumen to collaborate and solve business problems.
You have a myriad of choices when it comes to BI applications, and to keep it simple by “application” in this context I mean what you’re putting in front of your end-user. Deciding which is the right one is not an easy question, but you don’t need to answer it immediately, so don’t panic.
The vendors will hate me for this but they will enthusiastically sell you the “plug and play” functionality of their applications. “Plug here, connect to your source, drag this and see your BI user experience instantly come to life!” With a simple data source, it looks impressive but your data sets are not simple. Don’t even think about it happening until your data is ship-shape, because your “plug and play” is only going to be as simple as your data allows it to be.
So, don’t get hung up on choosing an application at the outset of your project, it’s not the right time and you don’t want to be married to an expensive application at the start of your process. Rather concentrate on designing and implementing a data solution which is adaptable to changing the front end at any time, then when you are ready, take advantage of the functionality that the applications out there offer.
At your workshop discuss the applications you want to consider and agree the guidelines you will use to compare them. Post workshop you can follow a process in evaluating and assessing what’s available. Consider factors like skill required to use it, how it’s accessed, cost, available features, access to trials and support.
Document this all. Keep it light and living, documenting the Why’s and Why nots to maintain a point of reference as to why you made your strategic decision.
Lastly, have fun! You’re in the fortunate position of being entrusted to drive your business forward with data and insights, and that’s an awesome opportunity!