Research has found that between 70% and 80% of business intelligence projects fail. While this number can seem daunting, don’t let it scare you. There are steps that organizations can take to help ensure the success of their business intelligence ventures.
Why projects fail
There can be several reasons that business intelligence projects fail. Two of the most common reasons are that companies aim to be all things to all people or they attempt to solve problems that are not the real issue.
1. Setting yourself up for failure. No company can be everything to all people. When you start a project with this mindset, you are doomed to fail. To be successful, you need to define what you are good at and then get even better at it. In other words, find your niche.
2. Define the real problem. Most often, clients are looking to solve symptoms of the underlying problem rather than tackle the real issue. This is not their fault. Unless you have done some investigation into the symptoms, defining the real problem can be difficult. However, until you solve the real problem, you will just waste time and energy responding to symptoms.
Setting yourself up for success
Organizations can improve the success of their business intelligence projects in many ways. It is important to begin with the right mindset. Remain flexible and open. Understand there are no silver bullets.
Here are some other ways to improve the chances of your project succeeding:
- Data consistency: Ensure that there is no missing data in your project. If there is one way to sabotage an intelligence project it is to overlook this step. For example, we had a client that had missing timestamp data. This made it difficult to map trends over time. As a result, we were limited in how much of the information we could use. Another example was when we discovered there was a field that was a number for some customers and also turned out to be a text label for others. Before the project could continue the difference had to be reconciled.
- Stay flexible: Things change and projects can change along the way. You can have restarts and still be successful. It is important to note that there may be opportunities that come from making changes. At NBS we prefer technologies that allow schema-on-read. This approach has been highly favored for big data applications. It makes it easier to solve problems such as the number and text field mismatch mentioned in point 1. This is because the fixes can be made after data has been captured rather than before. Some examples include AWS Athena, Google BigQuery, Presto and Hive.
- Tread carefully: There are many SaaS and cloud-based solutions on the market, which makes it easy to get started. Unfortunately, once you have started you may discover problems as the incompatibilities between solutions become apparent. For example, some teams may not know how many integrations a solution is leveraging. This makes it difficult to perform changes in a planned way.
- Know your constraints: Every organization has its limitations. Knowing what those are and working within them are key to creating a successful business intelligence project. While this includes budgetary constraints, it is also important to be aware of any other constraints such as technical constraints, capacity constraints or performance constraints. For example, how long will it take to train existing users on a new solution? Do you need to store 10GB of data or 1,000GB? Do your users need data fresh every 24 hours or every 10 seconds?
- Create a roadmap: Business intelligence projects need to have a clearly-defined plan forward. One that has phases integrated into the strategy. BI should always be a step-by-step process. Inside this roadmap, there are some tools or platforms identified as the best ones for the organization. Some specific goals to consider as part of a roadmap include how many reports in the system? How many users? How many data sources? What legacy systems are expected to be replaced? It is also important to have a clear timeline. For example, ‘We are targeting the top 3 reports on <legacy system X> to be replaced in the next 3 months, with a view to complete replacement of <X> within the next 12-18 months.’
- Understand your data users: Define what data your users need. How much information do they require? What way does it need to be delivered to them? If you are serving marketing users, have you confirmed your metrics definitions with them? If the data appears to be in error, who is the domain expert that will assist in determining the course of action to correct it? Once you have an understanding of your users, you will be better able to define the scope of the project.
- Goals: Along with your project strategy, develop your key performance indicators (KPIs), service-level agreements or operational metrics. These should be measurable goals that you want to achieve with the project. Once they have been set, ensure that you review the project often to see if you are reaching the goals or if something needs to be adjusted. NBS performs ongoing monitoring to ensure these goals are being met.
Being flexible and open are two of the most important things that companies need when they embark on a business intelligence project. Be ready to change if the project calls for it and able to adapt and restart if needed.
At NBS Consulting, we know what it takes to help develop business intelligence projects that work. We stand alongside our clients throughout the project process so they never feel overwhelmed or abandoned. Contact us at email@example.com and find out how we can help you build an intelligence project that succeeds.