The Innovation Tax: How Unresolved Data Problems Compound Against Every Strategic Bet

When focusing on one initiative, you tend to miss issues which happen beyond the initiative. Data is collected and definitions are articulated, and the team closest to the process is able to move things along. With respect to issues in the surrounding environment, workarounds become process norms. They are not brought to attention as they do not impede progress. They become the way work is accomplished in the particular environment.

After being involved with more than one initiative, you start to recognize certain things. A digital access program relies on consistency of patient ID across sites. A care management program calls for the development of longitudinal views of patient data, incorporating varying definitions of the data. A finance program relies on agreement between clinical data and the corresponding code. Each of these programs is considered a workstream, and invariably, each workstream encounters the same problem. Each workstream solves the problem using whatever means necessary to reach the next goal.

It is not unusual for each team to do what is necessary to reach their individual goal. The real challenge is identifying the work that shows each team solves the same problem that exists in multiple workstreams, using different workarounds that they are not even aware of.

The first signs of the issue show up later when the required outputs need to be integrated across the portfolio. Project timelines begin to stretch. Instead of focusing on the planned work, resources start to be allocated for reconciliation. Project plans spend weeks of effort to determine what the data actually means prior to its use. Structural costs are rarely documented. Instead, they are folded into project timelines and labeled as a project complexity. In reality, they represent the burden of unresolved structural issues carried from one project initiative to another.

This all takes time and no one booked time for this. There are no dashboards that show this time lost. This time lost will not show up in a status report. You will find it in the gap between the projection and the actual output. It will eventually be the expected cost of doing business, instead of being recognized for what it actually is.

I have seen cases where two different teams have built different data pipelines to reach the same goal, not by choice, but by a lack of visibility into what the other team was doing at the data layer. Duplication was only discovered after both were in production. By this point, the cost of reconciliation was greater than the cost of maintaining both.

This type of problem will not ever become a visible problem without looking at the portfolio as a whole. Individual initiatives are continuing to deliver. Teams are continuing to progress. The friction is so spread out that no individual stream carries so much friction that it becomes a clear problem. It is only when you take a step back and recognize that the same structural issues are being solved repeatedly that the problem becomes clear.

What you are describing is not really progress. It is progress with more and more overhead as more initiatives are created. This is because the issues have never been addressed. Each new program adds more friction. Over time the cumulative effect of this work limits what the organization can actually do, even when the strategy and resources are in place.

Again, you have an example illustrating that a component of a system that performs well in one area of the system may not perform well in another area. A model, dashboard, reporting framework (whatever you want to call it) that was built and tested in one system starts to behave differently in a new system where the underlying data was generated from different assumptions. Although the output may appear to be valid, the confidence in the output erodes, and the effort required to maintain confidence in the output increases with every new system in which the capability is implemented.

This is the same issue that is becoming more apparent as organizations begin to implement AI in a more extensive manner. A model that is trained on data from one facility is exposed to variation that the model was never intended to deal with when the model is implemented in a system that was developed through a merger or acquisition. The problem is not the model. The problem is the data that the model is reliant on, and that data is a reflection of many years of discrete business practices that have never been reconciled. AI does not create this problem, it inherits this problem and makes it more difficult to overlook, because the impacts of inconsistency are more severe and happen more quickly when the decisions being guided by the model are related to clinical or operational practices.

This problem is particularly tricky to address because it is inherently difficult to measure. For instance, the impact of unaddressed data inconsistency across a portfolio cannot be quantified in a single, discrete cost item. The impact manifests as additional work and effort and as the increasing disparity between the aspiration and capacity of the organization. Those who have experienced such a situation tend to refer to the phenomenon of data inconsistency as unquantifiable drag and friction.

Over time, this type of work becomes the new organizational reality in terms of what is actually doable. This is exacerbated by the widening of the gap between what the strategies and initiatives of the organization can achieve in practice and what the organization is able to achieve. This gap cannot be resolved by addressing one single problem. Rather, it is newer and progressively tougher work, carried over from earlier unresolved work and from the effort required to address the original work.

The paradox is that most organizational initiatives create more drag and friction. This is the case because, from an organizational perspective, most of the initiatives create more work to be done and increase the energy required to achieve the objectives, without making it evident where the work bottlenecks. The greatest organizational drag and friction occurs where the greatest work is done to address what is not visible.

Christopher Hutchins Founder and CEO, Hutchins Data Strategy Consultants