This looks strikingly like a problem of “irreducible complexity.” Only, just as the answer to the question “what use is half an eye” is “a lot more than no eye at all,” so there is an answer to this conundrum. Imperfect data is still data, regardless of its perceived quality.

Data quality, metrics, and tagging

In our opinion, there aren’t hard and fast rules regarding data quality. Assessment of the utility and completeness of data can be made on a more or less subjective basis, subject to some basic principles. The trick is to come up with some useful metrics that provide a benchmark and baseline for action, provided that a meaningful scale is created. Simply by creating a range—with 1 as the highest quality and 5 as the least—tagging will enable users to assign scores to data, according to their own criteria. A few examples will serve to illustrate the point.

Let’s consider a fiber that needs to terminate on a port in a specific cabinet or device. This is required for the delivery of a fiber service to a particular customer (a data point on its own) at a particular location (another data point). If we know where the cabinet or device is, we can give that data point a high score—say 1. If we know the fiber, but we don’t know on which cabinet or device it is terminated, we could give this data point a quality score of 5. If we do know which cabinet or device, we could assign a score of 1. Similarly, if we don’t know which port on the device, we may use a score of 3, and so on. This not only allows us to capture the data but also to generate a map of its quality, allowing it to be used immediately.

The same logic can be applied to a cell site and fiber rollout program. We may know where the cell has to go (1) to give optimum coverage, but we may not know where the nearest fiber duct is (5), which is essential for delivering the backhaul capacity required for 5G performance. We can use these to categorize tasks, understand the difficulty in completing them and complete the overall picture of network resources. Once we know where the fiber duct is, we can upgrade the quality score to 1.

In this way, even though we are using arbitrary values, we are assigning them logically. Completing this exercise gradually allows a baseline to be created. It allows all resources with a high-quality value (1) to be used immediately while allowing those with poorer quality values to be identified. This latter point is crucial because it means that a program to drive continuous quality enhancements can be initiated and followed, with clear targets in mind. However, it can be pursued when time and resources allow.

Such a program can be prioritized, too.  After the initial exercise, further rounds can be launched. It could be addressed, for example, to all data points with a value of 5, then to all with a value of 4 and so on, so that, iteratively, overall data quality can be enhanced, step by step from worst to best, or according to any preferred schedule.

By tagging data, it can be rendered immediately useful, even if caveats apply. Imperfect data can then be imported into a new, consolidated inventory system and made available to other processes, from the outset. This also lowers the barrier for new projects. If data quality and disaggregation have been a blocking point, resolving the issue with a clear, quality tagging process means that project delay can be eliminated and costs reduced.

Enabling an agile approach

It means that operators can adopt a more agile approach even during transformation projects. Because quality can be enhanced iteratively, teams can migrate towards a new, consolidated data model with a single inventory of network resources, working on the principle of “good enough is good enough,” so that they can immediately reap benefits. Of course, further benefits will be unlocked through time, but even half an eye lets a little light in. The resulting agility will underpin future efforts to capitalize on new-generation networks and services.

The questions of data availability, accessibility and quality are of fundamental importance. In order to deliver the performance and agility that 5G and the new network architecture promises, operators must be able to support dynamic service creation, orchestration, and delivery. They need to deliver to both customers and to support new business partnerships. In addition, operators are shaping up to support a new range of connected devices, which may be delivered directly or by third parties. In either case, they will become new resources to be considered within inventory management.

The foundation of this approach is network inventory: a consolidated record of all network resources and assets that is accessible to other solutions. The bricks and mortar of the inventory are the data that is imported, often from disparate sources and with variable quality. Many operators are challenged by imperfect and poor-quality data. But, by adding data quality metrics, the burden of poor-quality data can be eliminated. Migrating to a new, consolidated network inventory—an indispensable tool for future operations—need not be a challenge. The innovation of data quality tagging solves this problem, delivering clean data and providing the foundation of agile business and operational processes to support 5G and a new service model.