Implementing a BI solution can propel your organization to the next level, if done properly. It can provide a seamless platform that helps your employees get real-time insights from various corners of your business, and achieve operational excellence. It can also empower key stakeholders to perform ad-hoc analysis to discover growth opportunities and analyze data to validate their ‘gut feeling’.
However, more often than not, BI projects deliver mixed results and even fail miserably, in some cases, leading to inflated budgets, data issues, tremendous delays and unhappy end users.
Why do BI projects fail?
Here’s how a BI project typically navigates to a dead end. The BI development team meets with all the stakeholders to gather project requirements. Then they spend months, if not years, building a BI system that fits all the requirements and roll-out all deliverables at once. Since everything is well-documented and planned, it seems like the ideal thing to do and the BI team thinks it’s going in the right direction. But it seldom works out for the following reasons:
Each team, department, and business process requires a report (or a deliverable) to be able to reap the fruits of your BI project. These different reporting areas take forever to agree upon the key metrics they need to measure, and the business rules to calculate them.
Data sources required for your BI project may be incorrect, and a huge part of your BI implementation may simply be about building data validation processes.
Your data sources may be insufficient. For example, some business rules may require you to purchase data from third party data vendors. This means additional expenses and permissions, not to mention the integration problems associated with adding a new data source to your system.
The look-and-feel of your reports may be different than what’s required. For example, some of the senior management may require client-facing dashboards to be well-polished, something that may not supported by your platform, when it is launched. In fact, in one of our implementations, the stakeholders took 17 iterations just to finalize the report design!
Similarly, the first version of your BI project may not have sophisticated functionality required by some of your more tactical end users. For example, data analysts may require advanced features such as drill-downs and filters, which may not be supported at the outset. This may decrease the adoption of your BI project across the organization.
Requirements are misinterpreted and different teams (such as data validation, data processing, and reporting) work in silos. In the end, it becomes difficult to integrate the various modules to deliver a unified BI system.
Delays can occur because different teams are working on different schedules.
As a result, the stakeholders end up being dissatisfied, and the business users lose trust in your solution.
How to Successfully Implement BI Project in a Phased Manner
One of the most effective ways to implement an enterprise-level BI Project is to execute it in a staged or phased manner. It enables you to spot issues early on and do course correction before it’s too late. Here are 7 things to consider to avoid any pitfalls while implementing your BI project
1. Collect analysis & reporting requirements up front – Make a list of all the key stakeholders, decision makers and end user groups (business users) who will be impacted by your BI solution. Schedule interviews to gather all the requirements from the them. You may not be able to get all of them in one meeting at the same time. Schedule a series of meetings and follow-ups with each user group to flesh out all the requirements. In case you come across inconsistent or contradictory requirements, get the concerned users in one room and sort it out beforehand. It will help you avoid a lot of extra work down the road.
2. Organize requirements by business areas & phases – This makes it easy to execute your project by breaking it into manageable parts, instead of being overwhelmed by one large deliverable. It also allows you to clearly identify interdepartmental dependencies and organize your solution better. For example, payroll reports can’t be generated unless the sales system processes the sales data and calculates the rep payouts. With this information, you can sequence the data processing units correctly.Also, if you hit a roadblock in one area of your project, you can keep the other parts moving. It will also allow you to develop your project iteratively, regularly interact with stakeholders to keep them informed of your progress, and incorporate any feedback that you get from them.
3. Create mock ups and take a sign off – One of the most common reasons why BI projects get delayed is because the stakeholders simply don’t sign off on the deliverables. So create mockups of your reports and dashboards, present it to the decision makers and ask them to sign off on it, so you can get to work. Otherwise, you’ll get change requests right in the middle of your implementation. Ensure that you get the approval formally in writing – either as an email or through a word document. This will help you push back unnecessary requests that can potentially delay your project.
4. Work backwards – Once you have a list of all the users, the deliverables they need, and what goes into each deliverable, the key is to work backwards to design the processes required to report those numbers. Make a list of all the business rules, processes as well as the input data required to calculate the reporting information and generate your deliverables.
5. Prioritize your phases – Once you have identified the different processes & building blocks of your BI project, you’ll need to schedule & prioritize them.
1. Which phases will impact your organization the most
2. Which business areas will benefit the most from each phase
3. Which are the simplest, and most complicated parts of your project, as far as execution is concerned
4. Which teams and departments are most likely to adopt your solution
5. Which departments have well-defined reporting metrics that you can implement quickly
6. Which departments can give you the flexibility & time to work on a phase
6. Validate your data – If you feed faulty data into your BI system, it will give incorrect business insights and generate inaccurate deliverables. Consequently, the end users will lose confidence in your solution and stop using it altogether. So it’s essential to build in both automated and human-based data validations at every stage of your project to ensure that data issues are caught as early as possible and not allowed to pass through your system.
7. Gradually roll-out your project – Instead of having a grand organization-wide launch of your project, roll it out silently to a couple of departments or teams, ones who are the most ardent supporters of your project. This will help you remove any initial kinks in your solution and refine it quickly before rolling it out to others. It will also allow you to focus your time & resources on 1-2 departments and deliver a great user experience, something essential to help spread good word about your project. This will get other departments excited about your project and increase its adoption across the organization.
It’s important to remember that BI projects are not a one-time thing. The key is to fine-tune your BI system as your business grows and changes. Having an iterative approach to your BI project will ensure that it constantly provides useful insights to end users and keeps delivering value to your organization.