We hear a lot about Digital Transformations. Serving customers, constituents, patients or members better and giving them a great experience is the goal. How can we talk of Digital Transformation without talking about Data?
Tom Redman calls himself "the Data Doc." He has been advising organizations on data quality for years. Doc also writes for the Harvard Business Review on the broader subject of data. He wrote an article last October in which he included a little diagram that I now use all the time. It boils down to what can happen if an organization has not done a solid job of its Data Strategy. Here are a few takeaways.
The Data Quality Obviously, right? Right. When clients think of this in terms of trying to boil the ocean, I remind them that "quality" is simply "how close to requirements are we getting?" So let's remember that it depends on the requirements - what are planning to do with the data? Did your data strategy cover the use cases? Is there a framework defined to prioritize them which balances business value and the cost of bringing them to life? How can we evaluate cost without having a solid grasp of people, processes and technology that makes it up? Not all data is equal, and the data governance design needs to take this into account. Not all data pipelines are identical either in terms of the data quality they require. Ignoring data quality or not covering it correctly in the data strategy can lead to difficulty (read unmet expectations and delays) and cost overruns.
The Organization and the Skills Organizing the work required and covering every aspect is essential. Structuring how IT and the lines of business work together are also important to keep delivering at the speed the business requires. New approaches such as DataOps and ModelOps can help in this respect and will require training of staff, and much support. Putting in place a good demand management process, a good portfolio management process and some governance around these are also success factors. Not addressing all these will certainly lead, as the Data Doc says, to little growth beyond what departments can achieve on their own.
Technology Technology has gotten pretty complex since the Hadoop, NoSQL and appliance ecosystems began letting us realize that there were alternatives to dealing with data beyond the traditional data warehouse offerings. We knew before that the data warehouse architecture, and its limitations, could not support all the use cases for which organizations needed to find solutions - to the increasing frustration of IT. So lines of businesses agreed to cut corners on stability and quality to get their needs fulfilled faster. In the meantime, IT went to conferences where Netflix was presenting how it was stringing together 300 Hadoop clusters, had a team of administrators hacking and adding new ones every week, and 100 Java coders to get some value out of it. It all seemed like quite a leap.
Thankfully, when there is a need, there are innovators that swoop in and solve the problems. The thing I tell clients to think about is that if cost is not an issue (I know, I know, but it helps to test limits), each data pipeline can be architected on its own - from a technology perspective. For instance, most organizations have the data warehouse pipeline and the data science (creation and operationalization of models) pipeline. While some of the data is shared between them and have common constructs, the needs of those who use them are very different, and it stands to reason that the technologies that support them are as well. If we bring future scalability as a requirement in the mix, it is even more critical to harmonize the technology with the business processes and their respective participating personas.
Conclusion There is a column in the diagram called Defense, which I am not ignoring but think it deserves a blog on its own; so I will leave that one alone for now. As you see, many moving parts need to be harmonized to make Data and Analytics successful; or seen another way, any number of these moving parts can cause problems. The insurance policy is a good Data & Analytics Strategy!