Business Analytics Specialist
John Daniel Associates, Inc.
In this series of posts I will discuss the challenges I have found in the self service BI strategy and give tips for overcoming and avoiding common problems.
For interest of categorization I am going to break up the stakeholders in this process into the following buckets; Information Technology (IT), Analysts and Business users. The lines between these roles are often blurred, but to keep it simple, I will speak directly to the challenges of each role – not necessarily an individual who wears multiple hats. Each post in this series will discuss challenges faced by each of these groups. The perspective is given from the IT point of view, the group responsible for delivering information and reporting toolsets to their organization.
The idea of self-service Business Intelligence has been around for years. Its meaning has varied as technology has changed. In its most current form, self-service BI is described as flexible toolsets that allow business users to answer hundreds of question on large data sets without a dependency on an IT staff, typically through the use of visualizations. This all sounds wonderful- the business users win by getting access to their information they want faster and IT wins by freeing up time to do “other things” that don’t involve talking to business users. With all this promise, why are organizations still struggling with a very simple concept; the right information to the right people at the right time?
Assumption 1: Users want/need mountains of data to analyze
One of the most touted benefits with self-service BI is the myriad of directions a user can take while asking and answering questions.
Simply put, this creates noise, and noise can lead to paralysis by flooding a user with options. The more choices/information available, the longer it will take to make a decision. This idea has been studied and mathematically described as Hick’s Law. The law describes the time it takes for a person to make a decision as a result of the possible choices he or she has; increasing the number of choices will increase the decision time logarithmically.
You may argue that the increased time in decision making is offset by “better” decisions due to more information. In almost all cases that is not true. Often times speed is more important than absolute accuracy. If 5 indicators are pointing in a certain direction, does a user really need to analyze 10 more to “make sure”? Good enough is probably good enough. Stop wasting the user’s time.
Assumption 2: These tools are designed for business users and easy to learn
Want a sure fire way to frustrate your business user community? Give them another tool set to learn, manage and interact with. The business users are not dumb, stubborn or unwilling to learn. Rather they are pressed for time. They have spent years perfecting the most efficient way to do their job. Part of that is finding and using information. They are comfortable with their current process and know they can reliably get things done on time.
Out of the box, these tools are not designed for business users but rather business analysts. Confusing these two roles or assuming they are one in the same is a guaranteed path to failure for your self-service BI tool. There are a few people in each organization that can proficiently do both, but as a rule of thumb, assume they cannot.
Assumption 3: Business users understand how to manipulate and join data
I cringe when I see business users mashing up multiple data sources. The data is probably an output for another report. It has already been manipulated and summarized to give certain information and without thought, it is repurposed and mixed with something else. These toolsets make it easy to join multiple sources of data together but that doesn’t mean much. Knowing how to do something is much different than knowing why you do it.
The first thing uttered by a user when they get incorrect results is “this self-service BI tool gives wrong answers.” I have heard all types of things from users but it always come back to blaming the toolset and ultimately IT.
Assumption 4: Because they know their business they are able to analyze data about their business
What I often see is that users are very good at finding inconsistencies in data. They know what should be there and can easily spot when things are wrong. Do not confuse this quality assurance skill with analytical skill. This is a huge time suck that distracts users from their actual responsibilities. What’s worse is that many of the issues are self-inflicted due to poor manipulation of the data.
The second problem here is a gap in skill.
Put a collection of data and visualizations in front of 10 users and they will each draw their own interpretation. This is what makes self-service BI powerful but also dangerous. This happens for many reasons- self-serving bias, lack of analytical skills, lack of understanding of basic statistics, etc. If a picture is worth a thousand words then it is probably worth a thousand interpretations too. Information presented visually is open to interpretation. Give users the ability to choose which pictures they look at and they will tell the story they like best.
In contrast, analysts are trained to understand what data is telling them and not telling them, and they know which results to include and which to ignore. This is usually driven by mathematics, not by “gut” feel or by self-serving bias. They also distill their findings into actionable and useful information for their business users.
The notion that these tools make every user an analyst is false, just as knowing how to drive a car doesn’t make one a mechanic.
The overarching issue with everything listed above for many employees is that time is shifted away from things they do well to things they don’t do well. Do not fear…
Most of the benefits of self-service BI tools are achievable but the implementation, roll out, training and maintenance of these tools requires a specific approach. Just because these are easy-to-use tools doesn’t mean they require less attention from IT. Business users are resourceful; they will find a way to do their jobs with or without these tools. Making sure they have a good experience with these new tools is imperative for adoption and more importantly- better decision making.
Next post I will cover some of the tactics to help overcome the issues discussed in this post!