Ready to become a data-driven organization?
- blog
- 10 worst BI mistakes
10 Worst BI Mistakes
Originally published January 23, 2020. Updated April 22, 2024
A business intelligence (BI) implementation can be a hard job for a company of any size. Not only is there a new tool to learn, but new processes to set in place and new strategies to consider. The ultimate goal, of course, is faster, more efficient decision-making with a greater depth of knowledge and understanding of the task at hand. When implemented right, an analytics-driven environment will speed up workflows and optimize business as a whole.
Companies shouldn’t only invest in the BI and analytics platform itself but also in new strategies to ensure high user adoption among those who would benefit most from BI. Just as important as adopting new best practices though, is knowing which pitfalls to dodge along the way. Avoiding these ten common BI mistakes will help you reach the goal of a data-driven environment throughout the entire company.
1. You don’t prime the pump
Hazard level: Orange
BI is meant to improve workflows, but with any new addition to office life, there may be some bumps along the way. This is natural in a progressive BI adoption strategy. New habits can be hard to adopt, no matter how enticing the carrot at the end of the stick. The best way to keep employees focused on the end goal, though, is to prime the pump for the ultimate reward. Appeal to employees’ rational sides.
Without BI, there are natural problems and a lack of information for everyday business decisions. Tasks are likely manual, Excel-based, and prone to human error. The data that could lead to greater insight is locked in an ivory tower guarded by IT. Extracting is complicated, and changes to a report or analysis could take days or even weeks.
It shouldn’t be this way. The best way to foster this drive for investment in new tools and processes is to show, with everyday scenarios, how this new tool would fundamentally change the way they do business. A proper communications plan will stimulate the logical side of even the most stoic “technical dinosaur.”
2. There’s no clear roadmap
Hazard level: Red
Every BI implementation should start with a BI roadmap between local and central. Gather input from each department that would benefit from BI – and seriously now, aren’t all of them? Map out a step-by-step process that clearly lays out how each department will incorporate BI into their daily workflow from a practical standpoint, balancing both local and central BI.
Whereas the first mistake revolves around failure to appeal to the rational, practical side of employees with the overall improvement of their workflow, this roadblock is a failure of proper strategic planning.
Even though locally-driven BI initiatives create some value, without proper preparation, companies have a difficult time determining what is most important now, six months from now, and further down the line. A BI rollout doesn’t happen overnight. Companies often set their eyes on the end goal and bite off more than they can chew from a practical user adoption standpoint in the beginning. Think big, but start small. Then, expand the scope of data and users along with the plan.
3. You're measuring too many KPIs
Hazard level: Orange
One of the most common pitfalls we see companies fall into when starting off with BI is attempting to measure too many metrics. With so much information suddenly at decision-makers' fingertips, BI users fall into the slippery slope of trying to monitor everything simply because they can.
Dashboards and analyses end up looking like clown cars packed to the gills. Not only is this bad data visualization practice, but it muddies the water of BI entirely.
Many people make the mistake of examining only results (Key Results Indicators). It’s great to know how you’re doing, but it’s better to know how you did it. These are your KPIs and they should be monitored closely.
Clearly defined goals in your BI roadmap will help clarify which activities have the bigger impact on reaching that final objective. This will help separate the important data from the not-so-important.
4. You don’t consider the users
Hazard level: Red
BI will never reach critical adoption across the company if end-users are not carefully considered. Analytics is not one-size-fits-all. Users across departments and roles will consume BI differently both in form and functionality.
Some technically-minded users will need the ability to create reports and analyses from scratch and access data both inside and outside the organization. But not every user needs advanced analytic functionality – and many shouldn’t even have access to those tools.
Likewise, not every user needs access to scores of KPIs and company data. Many will only need access to a few KPIs delivered to them in simple dashboards and analyses embedded inside the platforms where they already work most. These should be designed to answer the business questions in a way that makes the most sense to those who need to know the answers.
Employees won’t adopt BI if it’s delivered in a way that doesn’t make sense to them. Failure to consider BI from both a licensing perspective and a data visualization perspective will stunt growth and make it difficult to further scope the solution in the future.
5. Departments are islands
Hazard level: Orange
One of the primary benefits of a BI and analytics strategy is to break down the silos that naturally exist in companies. But when working in isolation is ingrained within a company culture, it can be hard to break old habits. Some companies make great strides with their BI and analytics implementation and adoption, only to let that information live within departments. This is not how a data-driven environment is cultivated.
When the Sales and Marketing departments, for example, join forces with data, they’re better able to see what’s working and what’s not and can plan for future initiatives or improvements in current strategies. A common Marketing dashboard should include vision into the sales pipeline. These numbers help incentivize and continuously improve processes.
6. A lack of training
Hazard level: Orange
The popular myth that humans only use 10 percent of their brains may have been busted, but it’s no myth that most everyday BI users only scratch the surface of their analytics platform's capabilities. That’s okay for the users who only need those key dashboards or reports to skim over each day, but it shouldn’t be the case for the company as a whole. You shouldn’t have to hire consultants or wait for IT each time users need a change to a report or have an idea about incorporating external data.
Ultimately, the only type of tool that has a chance of successful user adoption across departments is one that isn’t too intimidating or burdensome for employees to utilize in their everyday workflow. Arming every BI user with the right lessons and information to make the most of the software is a no-brainer. And failure to do so all but ensures the solution will, over time, be abandoned on the roadside.
Only when every user understands the tools in their toolbox can they fully take advantage of the true power of self-service BI and analytics.
7. Not adhering to self-service speed limits
Hazard level: Red
Self-service is good, right? Yes! … And no. Modern BI models have made leaps and bounds with self-service BI capabilities in recent years. It’s fairly easy for users to get under the hood of the car, so to speak. Which also means it’s fairly easy for something to go wrong.
There are myriad ways bad data can end up floating around a data warehouse. Say a salesperson incorrectly enters the name of a city or state for a new lead. We discovered this very case in our own TARGIT database just this week. Or perhaps a valuable external data source stores data in an incongruous format. Incorporating depositories such as in-memory, data lakes, and other external data sources can add real power to a BI solution’s capabilities, but it can also spell real trouble.
Companies need a strict, documented method for maintaining proper data quality. Without one, inconsistencies and otherwise “dirty” data create a chaotic system that will produce inaccurate reports and analyses. Sandbox analytics in a closed-loop system also allows users to prototype new data securely without the help of IT. That data then is distributed for user testing where inconsistencies are discovered, or it is sent on to production. From there, more sandboxing is inspired, and the cycle of experimentation and data discovery begins again. This closed-loop system allows users to verify data before things go into production without technical assistance. Quality assurance increases and users take greater ownership of their data.
8. Driving like a race car, not a pace car
Hazard level: Yellow
BI myopia is a problem that strikes organizations of all sizes. In many cases, this comes in the form of license creep (not planning for the correct amount of users over time) and scope creep (the never-ending project). But companies also fail to plan for the aging layers of their BI application over time. In other words, pace layering.
In order to understand pace layering, consider a building. Every building has a number of layers that will age over different timelines and change to meet the needs of current occupants. The structure itself will last the entire life of the building, perhaps hundreds of years. The exterior may need to be remodeled or repaired every 10 to 20 years. The services — HVAC, electrical, ventilation — will probably need to be updated every seven to 15 years. And at the very bottom level, the elements inside the building — furniture, lamps, art — can easily be changed in a year, a month, or a day. First described this way by Stewart Brand, it helps to consider your BI in the same model.
There are three levels of an overall BI strategy that change at different rates.
- Systems of Record: Mature application packages or home-grown legacy systems that service core processes and manage data. These include the ERP system and other legacy data warehouse systems. They have the longest life cycle of the BI layers, five years or more.
- Systems of Differentiation: Applications that enable company- or industry-specific processes. They typically have a one- to three-year life span to adapt to new business requirements. This would be a company’s email campaign platform, e-commerce platform, site personalization, or other system that feeds into the BI solution.
- Systems of Innovation: New applications and opportunities, such as social communities, mobile apps, and web content, change all the time.
Your BI strategy must work with and around the aging of each of these layers in order to plan the budget and documentation accordingly.
9. Leaving drivers in the dark
Hazard level: Yellow
Just as failure to prime the pump is one of the first mistakes that can be made in a BI implementation, failure to actively incentivize BI adopters can diminish the importance of an analytics-driven environment. When employees can see the direct results of their own personal BI utilization, such as increased conversion rates, they’ll more concretely grasp the importance that BI brings to everyday business tasks. Departments should call out BI success stories in group meetings to incentivize employees and share successes.
At TARGIT, a report on conversions that directly resulted from blog posts is delivered directly to our content manager every morning so the results of her efforts are the first thing she sees when she opens her email in the morning. When she wants more information, she opens the platform directly and easily digs deeper with a few clicks to find the reasons behind the numbers. This helps her better plan her content strategy around what’s working and what’s not. Those concrete numbers give her a boost of motivation to think more creatively about delivering content in a compelling way, which adds excitement each time a conversion rolls in. It’s a win-win.
10. Keeping BI at arm’s length
Hazard level: Orange
BI will never make its way into the daily workflow of decision-makers throughout the company if it isn’t accessible from applications in which they already work. For many users new to the concept of BI, analytics can seem like a complex and intimidating tool to add to the To-do list. And as every manager knows, the more complicated a task becomes, the less likely it is going to be executed well or at all.
Today, BI can be embedded directly into applications such as SharePoint, CRM, and ERP systems. If more information is needed, users can drill down into the analyses with a few clicks. And mobile BI lets users take their analytics with them everywhere they go. The percentage of analytics consumed on mobile devices continues to climb, incentivizing BI solution providers to ensure enhanced power and native feels for mobile BI applications so power and insight are never compromised.
Both embedded and mobile BI remove the traffic lights holding up BI adoption. Unobtrusive analytics capabilities take employees out of the busy city streets to a BI joyride in the wide-open highway.
Avoiding these most common BI mistakes will help analytics take root across the organization, empowering every decision-maker with the data they need. Avoiding bad is good, but so is strictly adhering to the best practices.
