The essential 10-minute read on successfully implementing BI
September 15, 2016 | By Lisa Dare, TEKsystems Digital Content Strategist
We asked BI experts from all stages of BI projects, from solutioning through delivery, about the most common problems they see preventing BI initiatives from meeting their potential. They shared some insightful answers and proven strategies for enabling business intelligence projects to succeed.
- User adoption
- Planning for technology obsolescence
- Moving fast enough
- Trusting your data
A four-part path to user adoption
1. Involve key users right away
“You have to take user adoption into account from the beginning,” says Delivery Director Vyke Adepu. “You can’t treat it like a black box and to try deliver all at once.” Adepu suggests involving business users from Day 1, periodically meeting with them to get their feedback on the look, feel and usability of the product.
2. Set up an effective training plan
User training—demos, Q&A sessions, clear documentation, formal training sessions—is key to getting business value out of the tools. Users need clear demonstrations of how to use the tools in their specific business context, and to understand and get excited about how it can help them achieve their goals. If you have a BI implementation partner, consider building training into your SOW. And don’t neglect the training of your team who will maintain the tools; at a minimum, your IT team needs to know how to build new reports and dashboards.
3. Monitor usage
“It’s very important to monitor your key business users—are they using the BI tools?” says Srini Swaminatha. “Check weekly or monthly to see, and if they’re not, discuss their constraints and concerns and try to accommodate them.”
4. Deliver fast performance
If your programs are slow—even taking an extra few seconds to produce answers—your users will find take the data offline to analyze it, probably in Excel. This has two consequences: First, users will employ their own methods to analyze the data, hindering efforts to get a unified view of the truth. Second, the business won’t get the full benefit of the BI tools, from deeper analysis to rich data visualizations that can be reused with new data.
“We’ve seen many instances where performance tuning to deliver faster results for users makes all the difference in adoption,” says Swaminatha.
Planning for technology and business evolution
Data science Practice Manager Ashesh Parikh says the biggest risk an enterprise must address is planning early for changes in technology and your business model. “The technology you use will become obsolete—and replaced with more nimble tech—so you must account for change in your decisions.”
Parikh cites JetBlue as an example. “All the major airlines had manual call centers and heavy investments in technology, but over time the solutions got better. Along came JetBlue, which had no burdensome legacy system, so used the latest and greatest tech and left the big carriers scrambling to catch up.”
The moral is that companies can reap a lot of rewards by investing in early technology—but that same investment can slow them down when the tech and business models change. Over-investing in custom technology to address gaps in existing out-of-the-box solutions, over-investing in internal talent—these are the enemy of agility.
“BI—or really any technology solution you buy—has an expiration date. Plan to look at the value delivered 2-5 years down the line, max,” says Kaushik Katrapati, an engineer.
Moving fast enough
You need to balance having accurate, dependable data with moving fast enough to provide useful, timely results. And you also need to balance adequate requirements gathering with speed.
“BI is ever-changing; it’s not like traditional implementations like ERP projects, with processes that are well-defined and stable,” says Ramesh Vishwanathan, Practice Director, Big Data and Advanced Analytics. Vishwanathan advises starting with out-of-the-box BI products to get on track quickly. “Off-the-shelf products might not do all the things you need, but you could be 40 percent up and running in a matter of 10-12 weeks. You iteratively customize from there.”
BI accelerators can also help you speed your implementation process by, for example, automating parts of the requirements-gathering, generating information about customizations in your linked technologies, and discovering impact analysis and data lineage.
Trusting your data
Do you want a business intelligence operation to provide the truth—or the next-best thing?
If you want your business to get maximum value out of your BI tools, decision makers need to trust its conclusions. And that means you need to do three things:
1. Build a flexible data quality strategy: Data stewards set rules for cleansing and maintaining the data, and use a process to bounce data against that set of rules when it enters a database. “You need a maintainable knowledge base of rules that can be augmented over time, and an actual human to look at the rules,” says Dimitry Borochin, a Data Services practice architect for TEKsystems.
2. Get leaders to understand and trust in the quality of data. Provide a high-level overview of the steps your organization has taken to cleanse and integrate data.
3. Make sure users don’t mess up the data. “It’s a Catch 22,” says Arvind Sambaraj, a regional delivery director. “You want business users to explore the tools, but it can be like a kid with his hands on a new toy he’d like to explore. That’s where things start falling apart; suddenly you have reports from BI being exchanged all over the place with no one managing them, and it can lead to the wrong conclusions.” Think carefully about how you much control you want users to have, and provide adequate governance around reporting.
Related: The future of BI
The future of business intelligence: A Q&A with BI expert Srini Swaminatha