In this special five-part podcast series, I have visited with Phil Fry, VP, Go To Market at Verint, which is the sponsor of this podcast series. In this podcast series, we consider how Verint is changing the future of financial compliance by challenging the accept wisdom through capture, control, sustainability & oversight. I found this process as useful to think through a wide range and assortment of compliance issues for any compliance field: anticorruption compliance; trade compliance; AML compliance or any other type of compliance. Today in this concluding Part 5, we tie it all together, through a discussion of oversight of you the entire process. (The Verint process is so innovative, I have cross-posted the entire series on Innovation in Compliance this week as well.)
Oversight in the Verint process is not the traditional compliance definition of ongoing monitoring or auditing but rather how to bring together data and analyses from across an organization. Companies have mountains of information available to them, there are two key problems that continually bedevil the compliance professional. First is that the data is siloed and therefore inaccessible but equally importantly it is useless without understanding. Part of the later problems that that the compliance function is largely populated by lawyers who have no professional training regarding transactional data.
However it is also a problem of exercising proper compliance control. That is, your is dependent ability to see the whole picture, to review data gathered from a variety of disparate sources. Fry stated this is a discussion “about how we gather, combine and analyze that information, created by various operational systems, in order to enhance the ability to keep the operation running smoothly, to identify and address potential issues early on and respond to compliance requirements effectively. To do this we need to link data from the communications platforms into the trade and market information.”
Ideally such a protocol allow you to adopt a more proactive approach. Prevention is always the goal rather than simply detection and then remediation. However to do so, Fry believes one must “not only have to consider the data but understand the intent and the exact conditions when that data was created.” Here the visualization of the data is as important as the data itself. There has been a huge focus over the last few years around dashboarding, to the point that we are saturated with charts and graphs that no one has the time to review. He believes the “delivery of truly focused Business Intelligence is critical to success and has a vital role to play in providing real insights.”
While this might not appear easy, it can be a straight-forward exercise. Fry believes, “the key is the ability to cut through all the noise created by the mountains of data, and home-in on areas of concern, of potential or real non-compliance, that require attention.” This requires a consolidate operational data from a variety of sources and analyzing it within a framework that understands the data, can classify it appropriately and identify anomalies or red flags. That way it can then guide compliance teams to the specific activities, transactions or individuals that give cause for concern. Fry did caution that “given the sometimes highly technical nature of conversations about financial transactions, merely capturing or integrating into a regular recording is not enough anymore. The fact that voice conversations are inherently unstructured, carried out over two handsets at the same time, often in a noisy environment and include multiple languages can provide a real challenge.”
It can be particularly challenging in the financial trading environment, where such systems are built to handle the structured data found in the trade, order management systems and written communications, and these are used to good effect. Yet the insights they provide might take on a different complexion if looked at alongside the unstructured voice communications. This means that understanding the intent within voice data is key to preventing analysis tools creating a plethora of false positives that create additional work rather than efficiency. Better oversight comes from being able to quickly and intuitively see where these differing factors cross paths, affect one another, or reveal a truth that is not immediately apparent from the individual data.
By bringing together key operational and performance data into a single framework, one can verify operations and build queries and analyses that augment the value of the information, offer truly actionable intelligence and enhance control.
We concluded by considering speech analytics which is not widely utilized as the other technology but is still evolving. Fry said that, “it’s a sophisticated task to be able to automatically recognize where the significant information crops up during a conversation, and to mark and tag it so that it can be found again, viewed and analyzed. When a conversation consists of words and phrases that have little or no meaning in everyday conversation, it’s doubly difficult. Mainstream speech analytics tools typically only recognize, transcribe and tag “standard conversations”, they were not designed or built with trading in mind.”
What is required is technology that not only creates much higher quality recordings, making transcription of the language more accurate, but can also be programed to recognize the “tribal language” used in financial services and even within a specific company or group of traders. Obviously the same is true for any other industry where certain buzz words are used to describe bribery and corruption, trade sanction evasions or money-laundering conversations. Fry believes that “ Once you are able to produce these highly accurate transcripts and tagging, speech analytics can take a whole new place in the oversight regime. Conversations, interactions and trades can be analyzed for the inclusion of key words and phrases, however obscure and these insights can illuminate the investigation and reconstruction of trades, together with other data from other sources.” The final step is to pull all this data together but when you do so, it starts to become much easier to identify anomalies or patterns that have a compliance implication and know precisely where to start in order to put them right.
I hope you have enjoyed this five-part series with Phil Fry of Verint. For more information on Verint, check out their website here.