How to Get Rid of Your Data Debt
- febrero 24, 2020
Corporations amass vast amounts of data in their infrastructures, some useful and some not. If your data is not aligned with your business strategy, it is considered useless data creating what we call “data debt” and is the result of not having a detailed, written plan for data, including its collection, maintenance, storage, and disposal. Like therapists who work with hoarders on the A&E series to help them overcome their obsession with collecting things, I work with a team of data experts at NTT DATA To help organizations sort through their data.
Here are five steps I recommend for tackling data debt:
- Assess whether you have data debt. Not all data is debt. You must first understand how your data is helping implement your organization’s overall goals — including what data you have and how it’s collected, who uses it, how they use it and what they do as a result of using the data. This last assessment is essential. If you collect, organize and make data available to users who never really use it, then you are wasting valuable time and effort. Every data source you have serves some purpose. Articulate that purpose and how it contributes to your strategy. If you find no real use for a data source, you have found your data debt and can work to remove it.
For example, your retail company has been collecting information about your customers’ store preferences for years. But you recently made it possible for customers to shop online and pick up their delivery from any store. The store preference data you have collected via fields, records, or entire data sets may not be the information you need for understanding where a customer would like to pick up their delivery. This data may be considered data debt. What do you do with that data now? Is it still contributing to your overall strategy and goals? Depending on that answer, you may have to alter the data or delete it from your system.
- Evaluate the data debt you have identified. Once you have identified records, fields, or entire datasets of data debt, categorize each data element into one of four areas using the standard two-by-two matrix of costs (time and money) versus complexity (difficulty to solve). Those items that are relatively inexpensive and easy to solve could be your first target for the data debt clean-up effort. For example, if you identified a data source that is no longer relevant for marketing but could be combined easily with another source to provide sales with additional customer information, consider this your first project and show value immediately for reducing data debt.
For the more complex issues such as data needed for analytical purposes, consider developing a process that moves data continuously from active to inactive to archived to deleted. While evaluating your data debt, consider putting a price tag on the data that you will be eliminating from active use within your systems. How much space does this data consume on your database? What does that space cost to procure, set up, and manage? How many people touch this data but don’t use it? For example, do accounts receivable clerks have to wade through countless lists of companies that are inactive? Do sales reps have to scroll through page upon page of contacts to find those that are current with a company? How much does this activity cost your organization in productivity loss?
- Plan your work and work your plan. At its best, getting rid of data debt is a complex project within an organization and deserves corporate-wide buy-in and support. For efforts such as these, start with a well-articulated, detailed plan that is highly socialized with executives in finance, operations, sales and marketing. Build a team of both data specialists and those who understand how the business uses the data. Once you know the plan, have buy-in and the right resources, you can tackle the project one small bite at a time — by department. Manage each project with rigor, constant scrutiny and attention to detail. Make sure that everyone involved and affected by the data clean-up is aware of your progress and any hiccups that you may encounter. Constant communication is the key to success.
- Set a goal and measure your progress toward it. When you have established your plan, you will need to set a goal for your data debt elimination or consolidation. For example, how many of the projects identified in Step 2 will you undertake in this quarter, this year, the next two years? How will you know when the project is complete? When productivity improves? When database sizes are reduced? Whatever measurement you select for each project, make sure that the goal is communicated widely, and all involved know what you are trying to achieve. Establishing an executive steering committee to oversee progress is a good practice. This committee can act as an amplifier within the organization, giving strength and validity to your efforts. It also helps to involve this steering committee to finance projects because data debt projects often cross fiscal boundaries such as quarters and years.
- Keep going. Data debt projects are time-bound, meaning they have a start date and an end date. But managing data is more than a set of projects. The final step to any data clean-up project is to document and gain agreement from all teams as to how you will keep the data clean going forward. The effort of getting rid of data debt can be for naught if you don’t have a plan for continuous management of data to ensure that debt does not accumulate again.
While therapists can help hoarders clean their houses, if the hoarders do not change their attitude, the data mess can creep back. Similarly, organizations need to change their attitudes towards their data to remain debt-free. At each of the five steps above, a best practice is to document the process that should be used to keep the data debt minimized. With this documentation in hand, you have established a workable, enterprise process for data management that should see little to no data debt throughout the years.
Learn how NTT DATA can optimize your business with data insights and AI.