BI users aren't complicated. They want information that helps them make better decisions. To make better decisions, business intelligence tools must fit three criteria. The tool needs to be accurate, easy to navigate, and contain relevant information. If all three of these criteria are not met, then BI tools frustrate end users and they refuse to use them. Our job as BI professionals is to create products that make our customers’ lives easier. What can we do to ensure we create products that simplify our customers’ lives? Let's understand the frustrations end users experience with each criterion. We'll also discuss some solutions that can help reduce or remove those frustrations.
Problem # 1 – Inaccurate Data Destroys the Trust of End Users
Accuracy is one frustration end users experience with BI tools. Stakeholders use the phrases "data don't match" comes up often. Daniel Pasquarelli, Founder/President of BigSquare states “One of the quickest ways for BI implementation failure is from low user adoption rates due to a poorly built, undependable and inaccurate BI knowledge base.” Data is not matching the source system or numbers don't match across reports. Either way, end users are questioning the data. As a BI developer, you're reliant on the data made available to you. Don't let bad data prevent you from building good BI tools. Figure out what the problem is and be more active about helping to clean up the problem.
Solution # 1 – Be a Subject Matter Expert and Trusted Partner of Your Stakeholders
There are a few ways you can help with data inaccuracies. First, become an expert about data sources. Ask to join meetings where people use data. Speak up during meetings and represent the needs of your BI users. Be clear about what users need and why it's important. It is beneficial if you can tie those needs to critical business outcomes. The team at Predictive Analytics Today suggests tying outcomes to ROI and invest BI commitment there first. Define the business problem first, then collect the data and build the tools to solve them.
Second you should know explain to BI users about what's possible. Want to understand what's possible? Become an active learner about your stakeholders’ domains. Then, leverage you data expertise to provide options. An example would be BI tools for law firms. Build a BI system that focuses on the critical issues that attorneys care about [Pasquarelli, 2017]. Make your case on why it's important you're invited to meetings where data is being used. Attending meetings and building relationships will help you better understand user data needs. Joose, Biljon, and Mentz confirm it is critical to increase knowledge and understanding about the context of use (the working environment). This requires user involvement. The more you know about end user needs, the better BI tools you can build.
Finally, offer solutions to improve the accuracy. What are the missing data needs of the stakeholders? Provide short-term and long-term solutions. A short-term solution could be using text files to augment available data. Of course, this isn't best data management practices. But, the alternative is doing nothing. Besides, it's a short-term solution that can prove importance and viability of a data set. If the data becomes critical, then you have a great use case for the engineering team add it to the data warehouse.
Problem #2 – BI Tools Aren’t Built Simple or Intuitive
Navigation is another frustration end users experience with BI tools. Users complain information needed to run the business was difficult to find. One hurdle is actually finding the data in the business intelligence environment. A second hurdle is navigating the tool to get required data. Navigation issues arise due to bad filtering logic or malfunctioning drilldown features. No matter the reason, the end users shouldn't be responsible for figuring out how to get the data they need. As a developer, you must design the BI tools in a way that is intuitive for the end user.
Solution #2 – Create Product Features That Don't Need Explanation
Landing pages are an intro page to the reports available to users. Create them with interactive buttons that direct end users to important information. I've also used report tags to help find reports important to the stakeholder. Tagging reports is also a way to help end users search for the reports they are looking for. I've found using common language hashtags make it easy to find the reports. Hashtags such as #salesreport, #humanresources, and #operations are simple words used for search. Use business language stakeholders are familiar with to create more intuitive tags.
Provide simple filtering logic. It's easy to throw every field in a data table and allow the end users to filter reports for their preferred views. Too many filters causes analysis paralysis or inaccurate data outputs. This is where your expertise as a developer comes in. Explain the issues having too many filters can cause end users. A system with too many options actually makes choices more difficult [Pasquarelli, 2017]. Guide them to pick only the filters needed for a certain department. Do different departments need the same data? Create separate reports for each group. Only include the filters needed for a specific department’s business questions. Yes, build one for sales and build one for operations to reduce confusion across teams. Also, create logical drilldown features. Dashboards provide summarized views of totals and trends. End users won't trust the data unless they can drill into it to confirm accuracy. They only want to compare against source systems for increased confidence. Provide the drilldown capabilities to see the next level of data. Test the function to confirm it works as expected.
Problem #3 – The Data Provided Isn’t Relevant to the Audience Using It
Relevancy is a third frustration users encounter with business intelligence tools. Stakeholders find they don't have the complete information to run the business. Define the business initiatives and the required data to support them. That is the data you need available to end users. [PAT, 2019] In my experience, this is an issue driven by trying to build one tool everyone can use. An executive requests a report that provides a high-level view of the business. The entire department then gets access to the report. Leadership expects managers and associates to use the report to operate their business. Unfortunately, what the executive needs is not what managers and associates need. As end users drill deeper into the data, they find numbers that don't add up (see accuracy issue above). What is a developer to do to combat this issue?
Solution #3 – Include End Users in Design Phase and Incorporate Feedback
Including stakeholders in the design process combats relevancy issues. Yellowfin BI recommends involving your key customers in the design process. User-centric design activities identify upfront what reporting information is relevant to users. Yellowfin states you should choose department representatives to gather input from. Conduct user interviews, wireframing, and user testing exercises in the design phase. These activities help you build data tools that resonate with your audience.
User interviews help the developer gather requirements about how a report should work. It's critical that you know what decisions end users are trying to make. Now is the time to dig deep into their needs. Ask them what business questions or insights they would like to see from the reports. Capture this feedback and integrate later in the design process. A BI knowledge platform that complements that mode of thinking, makes a balanced pairing of user and knowledge with a much more successful outcome [Pasquarelli, 2017].
Wireframing is the process of brainstorm with end users on potential design layouts. There is software out there that helps with wireframing (Figma is the wireframing tool mentioned by the Tableau community). But, don't let a lack of software prevent you from hosting a wireframing session. I've used whiteboards and sticky notes. Since the pandemic, I've used PowerPoint templates to test different design layouts. Wireframing creates different layouts fast to find which design users most preferred.
User testing of prototypes provide feedback on what isn't working for end users. It's better to identify user frustrations early on with subject matter experts. Biljon, Jooste, and Mintz performn usability evaluations to measure the extent to which a BI tool achieves specific goals with effectiveness, efficiency, and satisfaction within context of use. The goal of the evaluations is to ensures tools are easy to learn, effective to use, and enjoyable from a user’s perspective [2013]. The alternative end users entering feature change requests after deploying a BI tool. The first option is the best because it doesn't cause a poor user experience for the entire end user group. It's easy to lose trust and much harder to earn it back. Use the testing phase to identify any bugs and remove them. That way end users don't encounter them later down the road.
In summary, you have to know your stakeholders and how they use reports. Dashboard development isn't capturing requirements, building a tool, and delivering it. BI projects should be run in some sort of Agile approach and there needs to be constant communication between IT and business users [Stangarone, 2015] . Be a subject matter expert of the data architecture, design better BI user interfaces, and include end users in the design process. Do these three things and BI users will find tools more accurate, navigable, and relevant. As a BI professional, people use your products because you improved the BI experience. Stakeholders will have better decision support tools. They will look to you as a master of your craft. In turn, you will feel more valuable at work and enjoy higher job satisfaction.
References
Pasquarelli, Daniel (2017). Why BI Fails in Law Firms. BigSquare. Visited on 8/21/22 at https://blog.bigsquare.com/why-bi-fails-in-law-firms
Unknown (2019). Top 10 Guidelines for a Successful Business Intelligence Strategy. Predictive Analytics Today [PAT]. Visited on 8/24/22 at https://www.predictiveanalyticstoday.com/top-guidelines-for-a-successful-business-intelligence/
Biljon, J., Jooste C., and Mentz, J. (2013). Usability Evaluation Guidelines for Business Intelligence Application. South African Institute for Computer Scientists and Information Technologists Conference. Visited on 8/21/22 at https://www.researchgate.net/publication/262403964_Usability_evaluation_guidelines_for_Business_Intelligence_applications
Yellowfin Team. (2011). Top 5 Keys to a Successful Business Intelligence Program: Macquarie University. Yellowfin BI. Visited on 8/18/22 at https://www.yellowfinbi.com/blog/2011/05/yfcommunitynews-top-5-keys-to-a-successful-business-intelligence-program-macquarie-uni-104857
Stangarone, Joe (2015). 5 Problems That Create “Unintelligent” Business Intelligence. Michaels, Ross, and Cole, LTD. Visited on 8/19/22 at https://www.mrc-productivity.com/blog/2015/09/5-problems-that-create-unintelligent-business-intelligence/