One statistic leaps out when discussing data governance -- we create more information every few days now than was created from the beginning of time to 2003.
But information governance is not just about dealing with the volume of data -- it involves effectively and efficiently managing the growing amount of unstructured data (e-mail, social media, telephone conversations), especially in the face of an increasingly strict global regulatory environment.
Against that wider backdrop, two issues immediately arise: current information governance is in its infancy, and current document management systems are cumbersome at best and unusable at worst.
Despite being in its infancy, information governance will quickly gain visibility within business. The inefficiencies caused by overwhelming data volumes are becoming unmanageable and costly. In fact, gaining control over data volumes can provide significant savings that companies cannot afford to ignore. So much so that information governance is set to change the corporate world over the next 10 years, to the same degree that corporate governance changed corporations over the last 10.
It seems that despite its importance there is little understanding about what information governance is, but the goal of information governance is simple -- to make corporate data easily available to employees who need it to advance a business. The opposite is also true, as information governance also helps identify data that doesn’t advance a business, and destroys it in compliance with law and company policy.

Categorisation

Incumbent document management systems are barely keeping up with the issue and are clunky to use.
Yet, if a company does not know what documents and data exist within its infrastructure – including hosted and cloud infrastructure -- it cannot interrogate and utilise the data effectively to make smarter decisions. Being able effectively to map data is a crucial element in information governance.
To manage data, companies must understand where it is and its context. Documents need to be categorised, because if not they cannot be managed. One US lawyer comments: 'Until they make [document management systems] as easy and as intuitive as saving to your hard drive, people will save to their hard drive.'
But categorisation isn’t without challenges.
First there is the choice between self-categorisation and system categorisation. Self-categorisation requires employees to self-declare what kind of document they create, which can be an ineffective process. The issue here is that the human psyche -- whenever a barrier is put in front of a task ordinarily done without barriers, people become disengaged. The other issue with self-categorisation is that it fails to allow for human error and much less intentional mis-categorisation. Mis-categorisation can be done even with the noblest of intentions, such as keeping documents ‘handy’, for instance.
Similarly automatic categorisation is not without flaws because without human interaction, the system can become static and it can’t improve. Therefore, a new approach to the categorisation workflow is required. This should be a hybrid solution that combines powerful automatic categorisation with human involvement -- one that requires minimum effort from the user, but can provide suggestions for which documents fit which categories. By bringing together human judgement with machine-based categorisation, the categorisation engine can learn from human confirmations of its suggestions. In effect, this could be another use case for the e-discovery technology known as predictive coding.
Agile technology such as this would even allow a user to set thresholds for human confirmation. If the system is more than 85 per cent certain that a particular document fits a particular category, the process would be automatic. For documents less certain, the process would require a confirmation of the choices the system suggests. For documents too difficult to quantify, the user would be required to choose the category.

Ensuring compliance

Only once documents have been categorised can policies be applied to support the second aspect of information governance; document retention. Document retention is susceptible to a number of different factors. This could be affected by external regulatory or compliance obligations, under which documents must be retained for a certain period, or a retention policy could be set internally by the company.  In many instances companies may be exposed to both external and internal retention policies and both have to be enforced accordingly.
Under information governance policies, documents can be given a date to self-delete based on the retention policy it falls under. This enables companies to avoid the issue of immortal data, the opposite of governance.
To tame the 'wild west' of unstructured data and enforce information governance, companies need to adopt  technologies that enable them to understand where their data is, the context of that data and be able to apply policies to manage data retention and deletion. However, you cannot rely on human or machine based processes alone.
Companies require a solution that provides a better workflow and machine-based learning to help drive efficiency.
The legal sector is leading the way, but other industries are likely to follow suit as they struggle to keep on top of the data they hold. The US Department Energy Office of Energy Efficiency and Renewable Energy is one example of a business that realised a few years ago that traditional 'print and file' policies had become an inadequate means of records management; it now uses a multi-faceted electronic records management system. Within three months the automated records programme accurately categorised 87 per cent of records -- saving time, cost and resource. The goal had been to reach 70 per cent.

Howard Sklar is senior counsel at Recommind