As a consultant, I have been a part of some very successful implementations of Data Analysis Projects. I have also been brought in to fix implementations which didn’t go so well. Being on a number of different projects, certain common components emerge. When these things are addressed early in the project, things go well. These items really cannot be put off to the end or ignored completely. These items are not related to the software, used in data analysis, as no matter what the tool selected, the solution will be incomplete if the following five areas are not addressed.
5 Components Every Project Should Include
Each of these items are important components of a successful Data Analytics Management practice. Problems with any of them can move a project from successful to a failed project.
Security is an obvious consideration which needs to be addressed up front. Data is a very valuable commodity and
Data is the New Gold
only people with appropriate access should be allowed to see it. What steps are going to be employed to ensure that happens? How much administration is going to be required to implement it? These questions need to be answered up front.
Business Continuity means the solution cannot be on the shoulders of one person, as that can be a risky situation. One person needs a break to go on vacation or not work, and needs a backup who is skilled and able to understand the system and run it alone. This can be a really big problem, especially for smaller organizations, who have only relied on one person. I have been brought in to assist companies who until very recently thought that they had a very successful Data Analytics Platform. The problem with it was that there was only one person who had the skill to maintain it, and that person quit.
Business Continuity can be a specific problem for Power BI users, as often times one user owns a report. Reports for an organization should never be owned by one person. All companies using Power BI should have the reports in a series of group workspaces, not belonging to any single person. Otherwise, if the person writing the report quits and their account is deleted, the reports are not then deleted as well.
Reliability is critical, because who cares about a data analysis project if no one believes the data is correct? In addition to accuracy, the system used needs to be reliable, containing data updates on a scheduled basis. How and when is the data going to be updated? How is that schedule communicated? The answers to these questions need to be addressed at the beginning of the project.
I remember working for one client who had over a 100 million dollar loss in a month on a visualization we created. I asked if the data was correct as that was a huge one month loss. I was assured that the data was not correct, but no one knew how to resolve the data issue. The reporting tool, whatever it happens to be, is not the place where data is fixed, it should reflect the contents source data. Where this rule is not followed, the reports are ignored as the data is suspect as no one knows why it should be believed as doesn’t match the source system. How is the source system data going to be fixed? This is often times a management issue as people need to be appropriately incentivize to fix things.
All data analysis needs Management Direction to set priorities. As there are only so many hours in a day, the important items need to be identified so that they can be addressed. What is important? Everything cannot be a number one priority as that means nothing is. In many data analytics projects, someone wants a dashboard. Great idea. What numbers show whether or not the company is currently successful? In most companies where I am helping them create data analysis project, the answer to what are the Key Performance Indicators [KPIs] is; no one has made up their mind yet. Management needs to provide the direction for the KPIs.
How are people going to get their hands on whatever it is that you just created? What are people looking for? Reports in their email? Visualizations on their phones? What do people want? Only if you ask the question do you know if you are providing the data in a way people want to consume them. In order to pick the most appropriate tool or design visualizations people are actually going to use, these questions need to be asked up front. Recently I worked for a client who had selected Tableau as there reporting solution, but they were getting rid of it. Why? The users wanted to do adhoc analysis in Excel, so they were using Tableau not to visualize their data or do ad-hoc analysis, but to select data for Excel Pivot Tables. A lot of money and time would have been saved if the question of how the users wanted to use the data was asked up front.
Hopefully all of your data analysis project include these components. In today’s environment, data is the new gold. This valuable commodity needs a system which is Reliable, Secure, important to Management, which can be distributed to continually provide value to the organization.
Data aficionado et SQL Raconteur