In my first article, I argued that while data has become critical to business success, it needs to be of a sufficiently high quality for the benefits to be realised − something that research shows to be out of reach for most companies at present. I exposed the challenges in that piece; now I want to focus on how to achieve data health.
Data quality and trust in the data are clearly two sides of the same coin. Building trust in the data means the company needs to put the necessary processes in place to improve data quality; these are cleansing, matching, profiling, monitoring, enrichment, parsing and standardisation.
In parallel, the organisation must encourage its employees to review the data and share opinions on its relevance. This ongoing and iterative process builds trust in the data across the organisation.
Once the data is widely trusted, it becomes not only usable by individuals or specific business units, but it also provides a single, aligned view of the facts upon the basis of which the organisation can collaborate effectively.
There are many companies that are cracking the data-health conundrum, and indeed realising the benefits we all know that data can deliver.
When everybody in the organisation has the same body of data, and trusts it, it becomes possible to make business decisions quickly and confidently. A common store of trusted data also facilitates cross-functional teams to respond intelligently, and in a coordinated way, both to business opportunities and challenges.
It's important to stress that achieving data health and thus trust in the data is not a once-off event − as noted, it’s ongoing and iterative. Just as a healthy individual has to build healthy eating and regular exercise into their lifestyle, so the process of managing and improving data, and collectively assessing it, has to become part of the way the company operates, a part of its DNA.
To establish healthy data practices, roles and responsibilities must be clear, tracking and auditing must be extensive (with minimal friction), and regulations must be seamlessly integrated into core processes.
Good practice requires the following steps to be built into the fundamental fabric of data management:
Identification of risk factors: The best way to prepare for the future is to recognise areas of risk, before problems arise. Risks can be internal or external.
Prevention programmes: Good data hygiene requires good data practices and disciplines. Responsible labelling and documentation of data makes it easier to assess and control the intake of data, producing information that is easier to understand and harder to ignore.
Proactive inoculation: Machine learning can train systems to recognise bad data and suspect sources before they can take hold and contaminate programs, applications, or analytics.
Regular monitoring: The sooner a data health issue is detected, the better the chances of an effective intervention. Just like medical wearables help us track our health between annual checkups, companies should institute a practice of continuous data profiling in addition to assessments of all incoming data and regular batch checkups.
Protocols for continuous prognosis: Over time a doctor will tweak a prescription, providing more or less medication as the patient requires. We should adopt this philosophy with our data as well: the specifics of any intervention will continuously evolve and improve.
Efficient treatments: Any medical intervention involves a risk/benefit assessment: the clear advantages to the patient must be weighed against potential side effects. But that doesn’t mean you only move ahead when there is zero risk. Good data professionals know how to balance trade-offs between things like security and efficiency to the net benefit of the company and its customers.
Making a success of it
The Talend Data Health Barometer, which I quoted in my first article, yielded some pretty dire statistics. While almost all companies (99%) see data as crucial for success, 97% face challenges in using it. One-third say that not everyone in the company understands the data they work with, and 46% don’t feel their data is delivered at the speed and flexibility to make it useful to the business.
Unsurprisingly, the most recent Barometer recorded a 10-point drop in satisfaction in all five markers of data health: timeliness, accuracy, consistency, accessibility and completeness.
On the positive side, though, there are many companies that are cracking the data-health conundrum, and indeed realising the benefits we all know that data can deliver.
One such company is Globe Telecom. One of the largest suppliers of digital services in the Philippines, it faced the challenge of a saturated market with limited expansion opportunities. It realised the only way to grow was to nurture customer relationships and maximise lifelong value from each customer.
Like many companies, Globe Telecom had no problem getting data, but it used time-consuming and ineffectual manual processes to assess data quality. It found it hard to identify and resolve data challenges before they became problematic.
For this company, automating the process for assuring the quality of data as it flowed into the data lake was the solution. Doing so meant the data available to the company was as reliable as possible. In turn, better data quality built trust among the users.
The results were impressive: a 400% increase in trusted e-mail addresses, with a reduction of 30% in cost per lead. Even more important, the conversion rate increased by 13%.
Other visionary companies that are prepared to do the work are achieving similarly good results.
Wellness in humans is a multibillion-dollar industry because of the benefits it can potentially deliver. Similarly, healthy data can enable a company to continue thriving into the future on the solid foundation of a clear understanding of its customers and what they want. In both cases, one needs to take the steps necessary to ensure continued good health.
Share