Businesses thrive because they fail in new and exciting ways. Little failures every day add up to big lessons. By examining those little failures, we can determine what doesn’t work. This insight leads us naturally to what does work, and certain patterns emerge. We call those patterns “best practices,” and although they are often learned the hard way, the results are always worth the struggle.
We’ve seen companies struggle and fail, but we’ve also seen some have great success. Here are a few things we’ve learned about embedded business intelligence and what we consider the best practices for it.
1. Have a data management plan
This might seem obvious, but obvious answers are often overlooked. You absolutely must start with the bare minimum: a plan for managing your data. All too often, an organization fails to adequately plan for modeling, efficient data extraction, aliasing, and data stewardship, all of which are important factors in data management. That energy spent preparing and defining the parameters of your embedded BI deployment will pay dividends when the time comes to implement it. Data management isn’t the most thrilling-sounding part of the insight-to-action process, but it’s a vital component.
2. Focus on phases
You can’t eat an apple like a horse or a hippopotamus, you have to take bites. This is also true for embedded bi implementations. A phased rollout method requires fewer resources and less time and almost always results in long-term savings. A gradual embedded BI rollout can include releasing a beta version to a select group of users and gathering feedback. This feedback can help guide the product team as well as become a source of testimonials, case studies, product reviews, and other promotional materials. A phased deployment lets you take little steps, informed by the plan you made in Step 1.
A piecemeal release plan offers resource-strapped teams a way to reach a full application launch without losing momentum.
A piecemeal release plan offers resource-strapped teams a way to reach a full application launch without losing momentum. While the idea of a beta isn’t exactly brand new, its usefulness in product updates and feature launches often gets overlooked.
3. Success factors: focus on them and track them
Part of getting where you want to go is knowing your destination. A clear view of your goals will help you define what factors equal success. This is crucial to the embedded BI implementation’s core objectives: reducing the reporting backlog and improving sales. This is even more important as time passes — your measures of success are likely to change as your requirements and implementations evolve.
Once your initial goals are achieved, your team can adjust its expectations and strive for even higher objectives. It’s important to remember that BI’s ultimate function is to support decision making, not generate reports.
4. Equip and train your users according to their roles
You probably already have a rough idea of what your users’ roles are, but it’s safe to say it isn’t real until it’s written down. Implementation of an embedded BI solution is a perfect opportunity to define which of the people in your organization (and/or your customers’ organizations) will interact with the embedded BI and what they might need in order to maximize its efficiency.
The most obvious delineation between users is their skill levels — power users vs. casual users. A power user typically has specialized knowledge of databases, querying languages, statistical analysis, and modeling. The power user will put your BI solution through its paces. A casual user, roughly 90% of a given population of BI users, are exactly that: casual. That doesn’t mean they won’t use some of the more esoteric features of your embedded BI, but they mostly stick to the basics.
Training and equipping these two groups is the fourth item on our list of best practices, but, like the others, it could easily be number one. If your power users and casual users don’t have what they need to use an embedded BI solution, then you might as well not even have one.
5. Promote a data-driven culture
This is something you probably already do, or at least have thought about a lot. As more and more casual users evolve into power users, and as organizations rely more and more on data to drive the decisions made across the entire company, it serves everybody at your business (and/or your customers’ businesses) to make data literacy a primary goal for your training and implementation efforts. This makes it even more important to have pre-defined success factors when establishing a project. Picking specific, achievable goals is paramount.
When it comes to data literacy, it’s important that you empower your staff to avoid statistical fallacies, especially when they’re convenient and tell you what you want to hear. While data itself has no biases or opinions, humans can unintentionally allow theirs to influence their conclusions. Being aware of these tendencies is an important consideration.
By knowing the importance and value of the data passing through your entire BI deployment, everybody can not only use the tools and reports but also understand why they’re so important.
Originally published with Analytics Insight.