Business Intelligence: A Practical Checklist
How to take the first steps right now
by Anthony Leshinsky | Jan, 2018
These days every person and organization are exposed to mountains of data. Making use of that data, turning it into useful information, is one of the core challenges of our time. Paths to this usefulness are many, some as simple as a quick exercise in Excel, others as complicated as Artificial Intelligence. Navigating this path is like walking through an evil haunted forest: there are a lot more ways to die then there are ways to find treasure.
Selecting the wrong tools, investing in the wrong skills and people, underestimating data munging, not clearly defining goals, not being aware of technological limitations, etc. are all very expensive detours. So is not doing anything, an organization that chooses to ignore Business Intelligence altogether also incurs the cost of waste it cannot identify while missing opportunities and strategies its more aware competitors are pursuing.
So, how to do Business Intelligence well? Well, the first step is acknowledging that it’s something you want to do AND that it will require some careful thinking. From there, the paths diverge wildly depending on your organization and goals. The diversity of options and objectives is one of the core challenges, but below is a very practical checklist of considerations to get started on the right foot.
Since this process has been so useful for so many of our clients and colleagues, we’ve created a standard spreadsheet to use specifically for it. This example version is pre-filled with information for Helix Pharma which you can use as a reference, and we use to illustrate the idea below. You may have more/fewer data sources, users, etc. Don’t worry too much about exactness, the point is to start and break ground on what your organization concretely wants out of data.
1. Your Data
Where and what is your source data? For many organizations it will reside in multiple systems, QuickBooks for financials, a CRM system for customers, some advertising data from an online platform, social network data from company accounts, external data, like market and competitor information provided by vendors.
Create a simple itemized listing of these sources with a brief description of the data you want to use. Fill in the rest of the Primary information using the drop-downs provided. Fill in any Secondary information if you are able.
2. Your Users
Who needs the information that can be derived from the data you’ve identified? Perhaps you are the owner of a small business and this is solely for your use, or maybe you are the IT manager and need to make this information available to your executives. The information can also be presented to clients, or other users throughout your organization.
As you see in the Helix Pharmaceutical example, create a simple itemized listing of your target users, indicate the type of user and include a brief note about what they are seeking to accomplish with the information you will be providing them.
3. Your Delivery
Now, the actual way in which you will get the information to your users. You want to focus on meeting their goals identified in the previous step. Here you want to consider the skill level and context of your users. Excel reports are great for many users, and they are very cheap! But if you need to provide an executive on the go with constant updates, they aren’t great in mobile and keeping them always updated is complicated. On the other hand, if you need to provide information to a hands-on analyst, a mobile dashboard is just going to frustrate them.
In the Helix example on the right you can see how we’ve listed the deliveries that will meet the goals of our users. It’s ok to be rough and experimental at this stage, if you aren’t sure how something will work yet, that’s ok. The important thing is to capture the idea, capability you desire and why. Refinement and prioritization will come later. Finally, this is a great time to circle back with your users and ask deeper questions about what they want and wish to accomplish if you find yourself struggling to clearly define it.
Having completed the steps above you should now have a good sense of your inputs (data sources) and your outputs (your users and their delivery). All that’s left now is all the hard work in the middle! =)
If you look at your list of inputs and outputs, and if how to get from one to the other seems overwhelming – that’s ok, but acknowledge that you will need some help and reach out to an expert. Not to toot our own horn, but we are awesome at this and always happy to give advice. Especially if you’ve already done the groundwork of understanding what you are trying to accomplish and what you have to work with. Sometimes all it takes is pointing a person in the right direction, which takes 5 minutes.
If the process of going from your inputs to outputs actually doesn’t seem that bad to you, you’re in great shape. The best thing you can do now is to walk the data path yourself, that is, actually take your input data, and create your outputs manually. Excel and PowerPoint are your friends here. It will be challenging, but it is a tremendous investment and increases your chances of success. Having created these prototype deliveries you next (1) seek detailed feedback from your users, (2) deeply understand the data process, and (3) have a de facto set of requirements and an example that can be followed for a full-scale implementation.
Business intelligence/data work projects are themselves complex, come with a host of secondary considerations and vary wildly from case to case. The cost of implementation can be between hundreds and millions of dollars. The cost of making wrong choices, especially early, can be extreme. It is all overwhelming, to put it mildly.
A practical path is to start by identify your inputs and outputs, and make a decision to either start walking the path yourself or bring in expert help to guide you.
Finally. This is maybe the single best piece of advice I can offer, whether or not you decide to use an approach like what’s outlined above: DO NOT START WITH A FANCY ADVANCED ANALYTICS THING. Like, NLP, Artificial Intelligence, Big Data, Blockchain, etc. They are all fine and good, when appropriate, but if it’s the starting point of what you are seeking from data for your business – you’re likely being suckered by someone selling one of those things. Your Data. Your Users. Your Needs. Those are your priorities. Keep things simple and your eyes on the prize, that’s how you find the data treasure – and not die along the way.