5 Manual Fraud Review Best Practices to Identify Fraud Without Causing False Positive
Let’s say you have a customer who submits an online order. You run the purchase through a barrage of fraud screening tools… and the transaction raises a few red flags.
Should you reject the purchase? Or risk getting hit with a chargeback later if it turns out to be unauthorized?
Performing a manual review is your most reliable option for determining if a transaction is fraudulent or not. But why would you need to review an order if your system has been largely automated? Better yet, how do you scale this process to make it feasible with your day-to-day operations?
Read on to find out.
- Dropshipping Scams: 10 Tips to Keep Your Business Safe
- What are Velocity Checks? How Do They Stop Fraud Attacks?
- ECI Indicators: How to Understand 3DS Response Codes
- Account Takeover Protection: Best Practices for Businesses
- How Digital Risk Protection Protocols Keep You Safe
- 10 Spells to Ward Away eCommerce Scammers This Halloween
What is a Manual Fraud Review?
- Manual Fraud Review
A manual fraud review is a fraud prevention tactic that consists of a human being reviewing incoming transaction data to determine a specific response or course of action. Manual fraud reviews are an essential part of any fraud prevention strategy, as automated technologies are in their infancy and so far incapable of nuanced thought or selection.
[noun]/man • yə • (wə)l • frôd • rə • vyo͞o/
Once upon a time, nearly all commercial fraud reviews were performed manually. This was before technology driven by artificial intelligence and machine learning, meaning fraud prevention relied heavily on human staff to verify orders and process transactions.
However, even though AI systems now exist in force, this doesn’t mean that your business should rely on them completely… at least, not yet.
Typically, manual reviews come into play after an automated system has flagged a transaction as potentially fraudulent. In other words, once an algorithm has gone over the details of a transaction, the system will rate the transaction based on its level of risk and advise a response. When the system is unable to make a determination, presumably because certain details are unclear, the transaction will be flagged for the human oversight provided by manual review.
How Do Manual Fraud Reviews Work?
Manual fraud reviews are most often conducted by trained analysts or data experts with access to pending transaction logs. This person will examine the available data from a logical perspective, then decide the best course of action to take.
This process usually involves several steps, including:
- Filing and processing incoming data
- Inspecting said data for accuracy
- Comparing incoming data to historical data
- Verifying details
- Verifying users
- Contacting customers (if necessary)
- Managing any disputes or chargebacks that may have resulted from previous decisioning
- Considering the details of each case and explanation carefully, in order to more accurately diagnose and resolve the situation
The analyst reviews the transaction in question in the context of all these indicators outlined above. Then, they make a final decision as to whether the transaction should be accepted or rejected.
The results of this process vary from case to case. It all depends on the severity of the issue or the reason the transaction was flagged. Again, this process is made much easier with machine learning as a preliminary review tactic.
Why Manual Fraud Review is Still Necessary
AI-driven algorithms can conduct a lot of processes much more quickly than a human being can. But, while machine learning and AI have come a long way in the past few years, a machine can only react to the data that’s fed into it and guess at a correct answer. It can’t engage in critical thinking or reasoning the way a human does.
Machine learning is best used for large tracts of data that it would take hours to manually sort. The AI is capable of receiving and processing data at a rate that a human being simply couldn’t match. Additionally, machine learning algorithms are capable of working 24/7 without further investment.
That said, manual fraud review is generally better able to detect nuance in incoming data, communicate directly with customers and banks, and make logical decisions from complex information. While machine learning has a lot of potential, for now, manual fraud reviews hold the edge. Using both in tandem is likely to achieve the best results.
|Manual Fraud Review||Machine Learning|
|Requires time investment||Works in real-time|
|Requires staffing||Investment and maintenance costs|
|Fewer false positives/negatives||Frequency of false positives/negatives|
|Capable of complex decisioning||May decline/block users without logical reason|
|Improved customer service||Incapable of customer service|
|Not available 24/7||Scalable; handle multiple jobs at once|
|Requires scheduling||Works 24/7|
Advantages of Manual Fraud Review
AI fraud detection tools and manual reviews complement one another. Each offers advantages that the other simply can’t provide.
For instance, manual review holds an advantage over an automated programs in the following ways:
Based on a survey of over 400 merchants, the report presents a comprehensive, cross-vertical look at the current state of chargebacks and chargeback management.Access the FREE Report
Machines might be “smart” enough to match the results generated by human insight someday. For example, machine learning may “teach” computers how to factor in human emotions and misconceptions, or to differentiate between gray areas and legitimate fraud risks. That said, humans will still hold a monopoly on the reading and translation of human thoughts, actions, and emotions for the foreseeable future.
Drawbacks of Manual Fraud Review
Manual fraud reviews are not a perfect solution on their own. Yes — it’s incredibly necessary to have qualified staff on hand to sort through and make decisions regarding questionable data. You will still need to plan ahead regardless, though.
There are several issues that can arise during the review process that can be challenging for you to manage. Some of the most common include:
These are all questions you will need to answer before deciding how best to handle the manual review process. Once you have the answers, you can decide how to implement solutions that work best for your business.
Machine Intelligence & Human Insight: A Winning Combination
We want to drive this point home: fraud screening is not a question of “human intelligence vs. artificial intelligence.”
The best path is to pair manual review with AI-driven automation. In this way, you can maximize the benefits of both solutions while avoiding the pitfalls of relying on one method over the other.
An example of how best to do this would be to use machine learning to pre-process the bulk of your incoming orders. You can then set your parameters to move all flagged items into a secondary folder for manual review.
The majority of your transactions would be either approved or denied automatically. The most complex cases, however, could be set aside for manual human oversight. This would leave your team with only a small number of items for investigation and decisioning.
5 Steps to Optimize Manual Fraud Reviews
The best solution is to implement automated tools, backed by human oversight and expertise. To get the most out of both methods, though, you need to plan for externalities and unpredictable situations.
Below, we’ve outlined five best practices you can adopt to streamline your manual fraud review process and prevent fraud losses while avoiding false positives:
Step 1 | Track Approvals & DeclinesTrack Approvals & Declines
First, you need to know how many flagged transactions you’re processing each month, as well as your number of approvals and denials. Once you have these figures available, you can calculate your review-to-decline rate, which will tell you the total number of declines your business processed in a year.
Segmenting results by individual processes — whether an order was subjected to manual fraud review or automated process — should help you identify which mechanisms are most effective at fighting fraud.
A review-to-decline rate of 20-50% is optimal. A review-to-decline rate below 10% hints that your processes can and should be largely automated. In contrast, a decline rate in excess of 50% suggests that you’re rejecting a lot of valid purchases.
Step 2 | Assemble a Team
Once you have an idea of how to balance manual fraud reviews and automation, you can better apportion tasks and data through the appropriate channels. If most of your reviews are handled through an automated system, you probably won’t need to outsource to a third-party fraud analyst.
On the other hand, if your business has a high rate of fraud and chargebacks, despite having automated systems in place, you will probably need help. Whether you hire internal staff to manage this process, or outsource to a third party, will probably depend on your budget.
Step 3 | Calculate Costs
Fraud review, whether automated or manual, requires a healthy budget. Automated processes tend to feature upfront costs associated with software and various upgraded hardware. In contrast, manual reviews mean more labor. There’s literally no way to get around that, even if you’re a sole proprietor and are managing this process on your own.
According to a recent survey from the Merchant Risk Council, the average merchant spends between $2 and $5 per individual fraud review. These costs can be much higher once you account for payroll taxes and other variable expenses. As a rule of thumb: if staffing costs add up to more than $5 per fraud review, it’s probably worth considering outsourcing your fraud review processes.
Step 4 | Consider Your Chargeback Ratio
Whatever you do: don’t forget your chargeback ratio! As we mentioned earlier, dissatisfied customers tend to contact their banks when they fail to understand something or a situation doesn’t go their way. The number of chargebacks you receive is intrinsically linked to the way in which you handle:
- Enforcement of terms of service
- Customer service (including manual fraud reviews)
- Order rejections and cancelations
- Billing and shipping issues
In order to get the most out of your manual review process, considering your chargeback ratio alongside your review-to-decline rate will be a wise practice.
Step 5 | Build a Multi-Layered Strategy
In order for manual fraud review to work as it should, it can’t be the only strategy you have in place to detect and eliminate fraud. An effective multi-layered fraud management strategy combines:
- Machine learning
- Advanced fraud tools (AVS, CVV, 3DS)
- Manual review
- Chargeback representment (when appropriate)
As we state time and again on this blog: the best and most effective anti-fraud strategies are multi-faceted. It doesn’t matter if you optimize every single facet of the manual fraud review process; if you fail to adopt the right supplementary tools and practices, you won’t see success.
Building Out a Multilayer Strategy for Fraud Review & Prevention
Manual fraud review has an important role to play in ensuring data integrity and avoiding criminal fraud. It’s also crucial for data integrity and helping identify post-transaction threats like friendly fraud, return fraud, and cyber shoplifting.
The role manual fraud reviews play is dependent on intelligent fraud and chargeback source detection, as well as a comprehensive chargeback management system. Only a reliable, end-to-end fraud detection strategy can:
- Look beyond reason codes to find the true sources of chargebacks
- Be more proactive about future disputes
- Identify new revenue opportunities
- Reduce fees, overhead, and other costs
- Eliminate false positives and accept more transactions
This is the only way to see true revenue recovery and sustainable growth. Ready to learn more? Click here and get started today.
What is a manual order review?
A manual fraud review is a fraud prevention tactic that consists of a human being reviewing incoming fraud data to determine a specific response or course of action. Manual fraud reviews are an essential part of any fraud prevention strategy, as automated technologies are in their infancy and so far incapable of nuanced thought or selection.
How long does a manual review take?
Most manual fraud reviews can take place in just a few minutes. However, depending on the complexity or severity of the incoming issue, a thorough review and investigation may take several days in extreme cases.
How does a manual fraud review work?
Manual reviews are typically conducted by trained analysts or data experts with access to pending transaction logs. This person will examine the available data from a logical perspective and then decide the best course of action to take.
How do you use the results from a manual fraud review?
Typically, manual reviews come into play after an automated system has flagged a transaction as potentially fraudulent. In other words, once an algorithm has gone over the details of a transaction, the system will rate the transaction based on its level of risk and advise a response. When the system is unable to make a determination, presumably because certain details are unclear, the transaction will be flagged for human oversight.
Why are manual reviews important in fraud detection?
Simply put: because they’re more effective.
Despite the breadth and scope of emerging technologies, manual fraud review will continue to hold an advantage for the foreseeable future. A day may come in which you no longer need to double-check a computer’s work, but today is not that day.
Machine learning and AI have come a long way in the past few years, but a machine is largely incapable of seeing and diagnosing gray areas, particularly those that revolve around human decisions.