My phone rings. It’s a call from a person I have never met. It’s a C-suite exec telling me his project team has major problems with their data. Judging from the sound of desperation in his voice, his reputation probably depends on the quick resolution of those problems. Many carefully-built careers have been scuttled by data disasters which often surprise the project team and their leadership. But decades of fielding frantic phone calls and subsequent project rescue experience have allowed me to compile this list of warning signs that your project might be destined to be doomed by data:
- Your System Integrator says, “Data is a business problem.” It is common for the System Integrators to marginalize their responsibility for the data. And we all understand that the data is owned by the business. But without the cooperation and helpfulness of the System Integrator, your data will have a very rough journey, and this friction could slow down or even derail your entire project.
After all, the only reason you are implementing your software is so that you can get data out of the new system. Therefore, it is essential that the focus on the data should be paramount within the project. Data should not be treated like a lowly stowaway to be unloaded roughly at the end of the trip. Data is everyone’s problem. - You have planned only one test data load before "Go Live". Data is fluid. It is the lifeblood of your business. It can also be messy, inconsistent, and deceptively complex. It takes multiple load cycles and testing with that data to ensure your data can be loaded correctly, and that the data is behaving as expected in the new system and in all reports, interfaces, and downstream systems.
If you have poorly timed or insufficient “Mock Loads” in your project plan, you’re headed for a rocky landing. - You don’t share Data Readiness Metrics and data-related issues as part of your regular status meetings. This is sailing blind without navigational aids. Even small issues not reported and addressed early can lead to major course corrections later.
The best practice: Have a knowledgeable Data Lead involved from the start of the project to manage and report on data-related activities. - People are creating “fake” data for use in a formal testing cycle. Using pretend data to test system functionality is like announcing you are ready to sail around the world because you read a book on sailing. Your real data has variability and challenges you cannot even imagine at this point.
The collection, filtration, cleansing, transformation and the actual loading of that data will be revealing, likely uncovering a host of issues — some of which may lead to software configuration or design changes.
System design changes in one area can impact other functionality within that system and potentially downstream systems. Bottom line – the earlier you find these data challenges, the more time you have within your project schedule to adjust for them. - You don’t have a written Data Validation Strategy. Data Validation is a process for determining that your data is complete and correct – as determined by the Business. It is a distinct set of planned, auditable activities that should start before functional testing (Iteration Test Cycles and User Acceptance Testing) and be performed again before "Go Live".
There is nothing worse than sequestering your entire business team for days of testing – and then watching everyone sitting idly because someone realized the data is not right and the planned test scripts cannot be performed.
A well-constructed formal Data Validation Plan is the “secret sauce” that engages the business in a methodical way to spot those unexpected data issues so the project team can take the appropriate action before it’s too late.
I wish you and your project team a safe journey. But be on the lookout for these warning signs, and if you see any, make course corrections quickly.
