by Anthony West Mar 13, 2018
Anthony West, CTO, Actiance, explains how the archiving of voice messages (via work phone, personal phone, Skype, etc.) is becoming more difficult, and why companies, particularly ones in regulated industries, need to deal with the increasing volume of vmails.
While it’s true that advances in computing are creating mountains of data that companies must deal with, it’s also true that most “data” is, in fact, being generated the old-fashioned way—by human vocal chords. The days of ignoring this fundamental data source are over, especially if your organization is in a highly regulated industry. For example, under the Markets in Financial Instruments Directive (MiFID II), which went into effect at the beginning of this year, financial services firms doing business in the EU must now gather all communications related to a trade upon the request of a regulator . This includes not just structured transactional data from the trades themselves, but also unstructured communications data such as email, IM, social media and, you guessed it, voice communications. And the latter may be tripping up organizations the most. Let’s talk about a few of the factors that make voice data difficult to manage, and how each requires a special archiving approach.
Difficult to capture
While many banks, investment firms, and other regulated businesses are already archiving landline communications, they’re now required to do the same for mobile. However, most organizations don’t have tools configured to capture this kind of data. This may be particularly problematic as BYOD becomes increasingly the norm—how do you archive a conversation that takes place on a personal device?
Be sure, therefore, that your archiving system includes not just a strategy for capturing conversations on mobile devices, but also personal devices. One way to address BYOD is to offer a separate corporate number on employee devices for calls/texts that’s carrier-agnostic and can be deployed at scale without any change to the device, SIM or existing mobile network. Beyond that, however, companies need to come to grips with the reality that regulators don’t care so much about the platform that the conversation takes place on, as they do about the conversation itself. So make sure that your archiving system captures conversations across a range of systems, including MS Teams, Cisco Spark and anywhere else they’re happening.
Difficult to store
MiFID II’s mandate to store voice data covers a host of individuals that didn’t fall under previous mandates and could lead to a 16-fold increase in the number of conversations that must be captured (and stored for up to seven years). Furthermore, digital voice recordings can be 1,000 times larger per element than email, with a minute of recording requiring about an MB of storage. Further complicating the matter, a great deal of metadata must be captured and indexed, such as ID, duration, phone number and employee number. Put it all together, and you’ve got a storage problem of “big data” proportions.
One of the most effective ways to deal with issues of scale in these instances is to leverage the cloud for archiving, which enables you to scale out your storage by leveraging the resources of a major provider (as opposed to buying massive amounts of hardware to house petabytes of additional data). Regardless of where you choose to house your data, be sure that your system is prepared to handle massive volumes coming at a breakneck speed .
Difficult to understand
Furthermore, even when voice data is adequately captured and stored, it’s not always easy to decipher its meaning . Audio could be in an unfamiliar language, inaudible or too cryptic to make sense of. Here’s where AI becomes essential. By integrating archiving with cognitive engines and applications which can be orchestrated to reveal multivariate insights, companies can unlock data from a wide range of audio files that might otherwise provide little if any information. AI can be particularly useful in transcribing and translating evidentiary audio, and machine learning can uncover patterns and clues in conversations that might otherwise go completely undetected by human beings.
Difficult to contextualize
In many conversations, there’s quite a bit of information outside of the conversation itself that must be pieced together for it to make sense , or to avoid being misconstrued. Without context, we can miss important clues, or draw wildly inaccurate conclusions. This may be great if you’re a sitcom writer (anyone remembers Three’s Company?), but it’s not so great if you’re scrambling to gather info for the FCA.
As you adjust your data infrastructure to meet MiFID II voice compliance, be sure your strategy includes archiving the data in a platform that provides contextual awareness—i.e., the ability to connect voice conversations with related communications that might take place over social media, text, email or other mediums.
Voice capture creates a challenge and an opportunity
Put simply, companies need centralized control over their voice conversations which:
- Provides a full view of all conversations happening in all channels
- Automatically identifies areas of risk
- Scales to massive demands for ingestion and storage
- Offers the full context of conversations that may be spread across multiple channels
This presents both a challenge and an opportunity. To be sure, mandates for voice capture require technical capabilities that thousands of organizations aren’t ready for. But these challenges are surmountable, and as companies implement the necessary policies and technologies, they’re creating a system that will position them to derive tremendous intelligence from the information that is exchanged by their employees and customers through the world’s largest medium for data conveyance: the human voice.