Wall Street & Technology is part of the Informa Tech Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 8860726.


12:49 PM
Connect Directly

Here's Why Tracking Voice Patterns Can Improve Compliance

In addition to standard compliance stipulations surrounding voice recording, flagging emotional sentiment and keywords of internal communications can unlock significant value in captured information.

How much oversight is too much? Given the regulatory pressures, the answer may be there is no limit. More is better. Always more.


Mark Whiteman is co-founder of Citicom Solutions, a UK-based managed solutions firm that recently launched Assure, a non-disruptive technology to capture and store internal voice and data for trade reconstruction and analytics. In an interview, he says mining data for red flags is just as much about regulation as it is about branding.

For regulatory purposes, most organizations are already capturing information from mobile, e-mail, IM, and voice, but Whiteman says organizations around the globe are eager to proactively use the data to show more transparency.

"We've been around for a number of years," he says, "and the industry is acutely aware that everywhere in the world they are not the most popular. Meanwhile, regulators are closing in more and more on requirements... They want to solve the branding issue and become more transparent in order to prove to the markets they are doing the right things."

"The motives are different, and they tend to be different for different people in the organization," he observes. "The global head of compliance, that person is looking to prove regulatory compliance, and the CEO is looking for good press so they can shout about all the stuff they are being criticized for and put it right."

Emotional Trends

Basic compliance requires the capture of voice on mobile and office phones, but this data is too often stored without any critical analysis for compliance. In fact, what you tend to find in compliance teams is a frustrating waste of man hours: Rooms of people hired to listen to calls, and they tend to do it in reaction to a complaint or investigation.

What Citicom's Assure offers is real-time analysis of all data elements that “listens” to all internal data and flags unusual activity, keywords, inappropriate terms and emotional sentiment. These patterns are then presented in a visual format to compliance teams, COOs, HR reps or other contacts to then act on.

"On voice, there are certain things you can analyze that you can start to say emotion is taking place," explains Whiteman. "Shouting, people talking over each other, use of language, and a number of other key things." Yes, sometimes people have bad days, but if all the emotional pings are charted over time, they start to tell interesting stories.

"We've plotted the historical data against Enron, and you can see the emotional trending peak and troughs. When you look at certain events you can see the actual peaks are where these instances are happening. It's fair to say it's another piece of evidence a corporation should look at in an evaluation."

Unusual Language

Of course, voice recognition is nothing new, it's been around for almost 20 years, but the application for regulatory responses within financial services is new, and use cases are still starting to emerge.

For example, on a particular day the software picked up on the unusually frequent mention of Martin Luther King. It was highlighted as a phrase out of context, and those monitoring the company were alerted. As it turns out, it was Martin Luther King, Jr. Day, and colleagues were asking each other about their plans. The system was not wrong, but at the same token it did pick up "Jesus Christ" in excess at Enron days before the bankruptcy. Far from being a holy holiday, you saw people sending messages home along the lines of "Dump the stock, Jesus Christ we're going to lose a lot of money." The first instance here is nothing, the second holds a lot of value.

Whiteman recalls when mobile voice recording began in UK the market saw a significant shift in behavior. "First thing, a lot of people went out and bought a private mobile. We also saw a huge reduction in the use of mobile voice. And there was a huge increase on IM, even though they knew they were being recorded there too. The other thing is, when things start to get recorded, you start to see people playing strange games."

Such games include speaking in foreign languages. In a particular case, half way through a recording of a trade the two speakers broke into Polish. Use of foreign languages to conduct business was outlawed, and employees had signed an agreement saying they wouldn't. The technology alerted their compliance department and the traders were disciplined.

This also include the use of pseudonyms like "Chinese restaurant" as code for an unauthorized trade ("time to get out of the Chinese restaurant"). On its own, the words are quite innocent, but when used in excess of 50 times in the last week, it becomes suspicious. "The software says you might want to look at this," explains Whiteman. "You can then listen to calls and say this doesn't quite look right."

"All we can do is provide the dashboard," says Whiteman. "It highlights out of context, unusual language. Not swearing, but unusual stuff."

In Defense of Extreme Surveillance

You could argue that employees change behavior once they know they're monitored, which could impact the effectiveness of the monitoring system, and could alter company culture. But, ask yourself, if their office culture is promoting something unpalatable to the office, is change a bad thing?

Consider the LIBOR scandal in the UK. It suggested the culture at Barclay's was all wrong, and essentially characterized by the attitude of don't worry about regulations just do the deal. Had this kind of monitoring been in place, they may have not been so lighthearted in their dealings. "I think [monitoring] will shape culture of banks," Whiteman says, "and maybe it needs to be changed in some instances."

What's Next - Visual Signals

Under Dodd Frank, video, like voice, must be captured and stored for trade reconstructions. Whiteman says the big question today is how to analyze body language from video in an equally valuable way as voice. For example, how do you detect someone saying aloud "I wouldn't do the trade" while simultaneously giving a thumbs up - wink wink.

"It's big because people are doing more video conferencing," he explains. "It places a huge reliance on organizations to capture and reconstruct data in a valuable way."

Citicom's Assure has begun its expansion in the US with a US-based tier one bank. The bank is committed to a global long-term roll out following a proof of concept in the US and UK. Becca Lipman is Senior Editor for Wall Street & Technology. She writes in-depth news articles with a focus on big data and compliance in the capital markets. She regularly meets with information technology leaders and innovators and writes about cloud computing, datacenters, ... View Full Bio

Register for Wall Street & Technology Newsletters
Stressed Out by Compliance, Reputational Damage & Fines?
Stressed Out by Compliance, Reputational Damage & Fines?
Financial services executives are living in a "regulatory pressure cooker." Here's how executives are preparing for the new compliance requirements.