5 key challenges when tuning AML transaction monitoring software

Having a correctly tuned AML solution is essential for any organization dealing with high volumes of financial transactions. With compliance risks being higher nowadays, regulators are constantly inspecting AML transaction monitoring systems to ensure adherence to international regulations.

One of the main issues faced is the high number of false positives which can be caused by an incorrectly tuned transaction monitoring system. Then there is the other side of the coin – if an organization performs extensive tuning on its AML software, it runs the risk of reducing the effectiveness of the system. If caught by authorities this lack of effective monitoring could also result in fines.

Model and data alignment

The first step for any tuning exercise should be to understand the objectives of the model. Is it risk minimisation or detection of suspicious behaviour? The second step would be to identify the data feed that will be used for the model – meaning the type of data the model will be using to monitor transactions. The data type needs to satisfy the requirements of the model. For example, it is no use having a model which is able to monitor online bank deposits if the data that will be provided to the model will not include such information.

Therefore, organizations and institutions need to establish their main objectives when it comes to compliance so that their mitigation expectations are met via the model and data type they will be using. With these criteria established, it will be easier to tune the software according to their specific needs.

5 common tuning errors

  • Data and model conflict: when elements in the data field are misused and misread by the model, errors are made resulting in false positives. For example, if a ‘country’ field is left empty and filled in as ‘NA’ which stands for not applicable, the model may misinterpret it as being Namibia. Therefore, for the model, all transactions with an empty ‘country’ field will have originated from Namibia.

  • Structuring patterns: criminals use the structuring process to deposit illegal funds into accounts by breaking down the large amount of funds into a number of smaller transactions. These transactions are created in a way which fall below the reporting threshold in an attempt to avoid detection. However, honest businessmen also make use of this tactic on a frequent basis and therefore such transactions can be wrongly flagged. Therefore, the software needs to be tuned to catch suspicious activity via customer profiles and the pattern of transactions.

  • Velocity patterns: velocity patterns are noticed when a customer receives a deposit in their account and then rapidly withdraws the funds a few minutes later. This can be a sign of suspicious activity, although it can also represent a genuine client withdrawing money to pay their bills. Tuning the software to recognise certain patterns is essential for removing false positives.

  • Exclusion lists: these exclusions or ‘whitelists’ are created by organizations to exempt certain customers from being monitored by the AML software. This is usually done to minimise false positives given the customers’ reliable track record of legitimate transactions and honourable business practices. If, however, these lists are not reviewed periodically and patterns studied for abnormal behaviour, this limited amount of tuning could lead the organization into hot water with the relevant authorities. Reason being that without performing risk assessments on whitelisted clients, illegitimate transactions by usually reliable people can go under the radar, resulting in no flags being raised.

  • Suppressions: organizations sometimes tune their software to suppress alerts based on particular criteria in order to minimise false positives. They have to ensure however, that the system is properly tuned so that illegitimate transaction alerts are not suppressed in the future. If such correct tuning is not in place, an analyst may end up classifying an alert as a false positive given that they are unaware of the subsequent alerts which the software had suppressed.

The Solution

ComplyRadar utilises a full risk-based approach to eliminate disruption to genuine customers, detect potential criminal behaviour, and demonstrate full ongoing compliance. It sends you notifications on the transactions that matter and enables you to automatically apply a full-pattern analysis to instantly see suspicious transactions in real time. Get your free, no-obligation demo of ComplyRadar today. Visit https://www.comply-radar.com/ for more information.

Top 5 measures for AML transaction monitoring systems

Active or real time monitoring of financial transactions should be the order of the day for institutions and businesses handling high volumes of deposits and withdrawals. This capability is essential in mitigating risks, thereby improving the monitoring and detection process. By effectively and actively monitoring transactions, organizations are in a better position to meet their AML obligations without impacting their day-to-day operational processes.

Impediments to real time transaction monitoring

  • Data quality and mapping: consolidation of transactions on different platforms is perhaps the most challenging aspect of AML. This is because the data needs to be coherent for monitoring purposes. Data quality needs to be taken into account, especially since sometimes, similar batches of data are grouped together, making it difficult to monitor holistically.

  • Lack of feedback on SARs: an organization expects that whenever a suspicious activity report (SAR) is filed, that it receives feedback from the competent authorities about the investigation and the outcome. However, sometimes, this feedback is not up to standard or even non-existent to allow the self-improvement of the organization’s monitoring capabilities. This leads to improvement efforts being deemed subjective as people within the organization itself have to come to their own conclusion on what is working and what is not.

  • Scenario alerting techniques: if an organization keeps moving the goalposts as to what is considered risky and what is not, quality of monitoring will be diminished. Non-transparent analytical methods such as neural networks add another complex layer in relation to the unpicking and explanations for decisions.

5 measures for accurate transaction monitoring

  • SAR disclosure rates: by categorising the files of cases being submitted, understanding the reason behind the majority of submitted cases, and acknowledging the areas of business affected – an organization can extract valuable quantitative information which can be investigated.

  • False positive ratio: comparing the overall changes in this measure provides an indication of the monitoring system’s performance. When at high-level, the ratio provides the best indications of any mismatch between detection processes and overall risk.

  • Alert volumes: any changes in alerting volumes should be investigated. Some potential issues could be underlying within the transaction monitoring system itself. Some examples include scenario changes, rule changes, setting up of new scenarios, introduction of new business lines, and the introduction of new types of transactions.

  • Operational costs: costs also need to be monitored in two aspects. The first aspect is the size and capability of the investigative team which could also be handling day-to-day system tasks. Secondly, the supporting infrastructure around the team needs to be constantly assessed as well. In this case, operations growth needs to be taken into account.

  • Number of monitoring rules: as the number of scenarios or rules increases, so does the strain on system management processes which in turn could lead to a higher number of cases. An increased number of rules could be due to changes to existing compliance regulations, lack of consolidation in account segmentation, and new products or business lines being added.

The Solution

ComplyRadar helps you address AML transaction monitoring requirements by automatically identifying suspicious behaviour in real time or on a scheduled basis, while minimising false positives. Get in touch with us today to book your free, no-obligation demo. Visit https://www.comply-radar.com/ for more information.