Data has a lifecycle beyond acquisition. In that regard, it’s no different to products, services and customers; it’s something that needs to be fully managed by the organization. Ignoring this task — and the end of the lifecycle in particular — means the organization will miss out on opportunities for learning and even growth.
While the acquisition of new data and managing it receives a lot of attention, we hear much less about the end of the data lifecycle — data offboarding. Customers aren’t loyal, they leave. Products and services aren’t sold forever. When they reach the end of their lifecycles, what happens to that data?
In his book “Ends: Why we overlook endings for humans, products, services and digital. And why we shouldn’t,” Joe McLeod confronts the challenge of designing with endings in mind. We urgently need to apply those insights here and overcome the temptation to hoard data.
The problem with data hoarding
We hear a lot of talk about how, why and when organizations collect data. Rightly so: customers expect integrity and laws demand compliance; no company wants (or can afford) to be put through the wringer for failing on either grounds.
Customers, products and services all generate and consume data. Organizations hoping to be “data-enabled” or “data-driven” will need to ensure that this data (and especially personal data) is well maintained while it’s useful to the company, and managed in line with the requirements of the law and best practice. Data must be cleaned, kept up to date and so on because it has value which can be maximized when effectively managed. Conversely, data becomes less able to deliver insights if it's ‘messy’ and chaotic.
Organizations typically understand when they got their data and what data they have. Most organizations even know where it has come from. Legislation and good practice also emphasize the importance of understanding who is responsible and accountable for that data.
However, when it comes to decisions about offboarding, organizations need to understand why they have the data they have. Metrics might tell you how often particular data have been used, but that's generally only part of the story — and perhaps not the most important part. A piece of data could be used only very infrequently, but it may well provide valuable insights that aren’t available through other routes. Weighing up ‘frequency of use’ with ‘value when used’ is something to be done on a business-by-business basis. It's well worth spending time and effort on.
What often follows from the notion that a single piece of data might be infrequently used but of great value, is described by Caroline Carruthers and Peter Jackson in their book “The Chief Data Officer’s Playbook” as “data hoarding” — retaining data just in case it might be useful at some point in the future.
While such a move might seem careful or even thoughtful, in reality, data hoarding can be a sign that a firm has no data strategy or vision beyond acquisition, that it doesn’t understand its data enough to be able to identify what is – or what might be – useful, and that, in the worst cases, it doesn’t even understand what business it is in.
Regulatory considerations notwithstanding, there is a strong business case for actively considering whether or not to retain data — and, just as importantly, to have a plan in place about when and how to offboard it. To put a real world value onto this idea, it’s worth noting that researchers have calculated that poor quality data costs companies $12.9m annually.
The way to redress the balance between data hoarding and retaining data that really does have potential to be useful is to address a few straightforward questions:
What product or service does the data support?
Does the data support ongoing or ended relationships?
If the data doesn’t support a product, service or current relationship, what business purpose or objective is served by keeping it and do we have permission to keep it?
From an AI perspective, old and improperly maintained data will maintain the biases and inconsistencies plaguing the business if it’s used to train new models. If data really is an asset, it needs to be maintained like any other — otherwise it will fall into disrepair and decay.
Closure is an opportunity to use data more effectively
In “Ends”, MacLeod argues that closure creates an opportunity. In the case of customers, it allows the org to ask for their reasons for leaving, and then use that information to help improve customer retention policies, to celebrate their relationship, or just create a positive departure experience such as Spotify’s “Goodbye for now :( playlist”. McLeod argues that a positive closure building trust and transparency, not only creates a path for a customer to return, but also improves the organizations relationships with their other customers. Data, suitably anonymised and aggregated, from the customer’s interactions with the organization up to and including the departure is endlessly useful for refining current and future offerings.
The same approach can be applied if a company discontinues a product or service. The data relating to the product or service itself can be archived (where it can be analyzed for future planning exercises, but won’t impinge on the utility of data about live services), or it can be deleted depending on company policy and legal requirements.
Begin with the end of the data lifecycle in mind
The second of Stephen R Covey’s 7 Habits of Highly Effective People is “Begin with the end in mind”; McLeod’s Ends gives us an approach for doing that effectively. In the context of letting go of data, this means understanding that offboarding is a legitimate and important part of the data lifecycle. To do this properly, organizations need to develop a framework that allows them to see what good data looks like — only then can they determine when its utility is ending. In a post-GDPR world, thinking about endings becomes critical in understanding what data should be kept or discarded.
Taking into consideration the offboarding and end-of-life experience of customers, products and services has a profound impact on data and data management. Being mindful, having a clear data strategy that is aware that data has a limited useful life and understanding there will come a point when the data needs to be purged should get any firm on the right road to effective data offboarding.
The result might be less data — but that data will be better quality and ultimately provide more useful insights.
Disclaimer: The statements and opinions expressed in this article are those of the author(s) and do not necessarily reflect the positions of Thoughtworks.
