Manual Reviews can be performed in-house or by a third party vendor. Manual Reviews require staff members to perform manual checks on orders that were not low risk enough to be auto-accepted, and not high risk enough to be auto-declined. Manual Reviews are used with these “in-between” orders to determine if they are fraudulent and to try and convert the ones that aren’t.
In general, manual review is not a very good fraud-prevention technique. That is, you do not want to rely too heavily on manual review as it is time and resource intensive, does not scale well, and requires customers to wait for their orders to be processed. Manual review should be seen as a technique that is used to try and convert orders that are too risky to be automatically accepted, not something you rely on to catch fraud.
Keep in mind that the quality and effectiveness of manual reviews are directly proportional to the knowledge and experience of the review staff, as well as the tools and processes that they have established to perform manual reviews. This is an important for consideration for organizations that perform manual review in-house as they will also need to train and equip review agents with the necessary tools to successfully perform their job. Organizations with a high level of employee turnover may find difficulty maintaining consistency with overall manual review performance.
According to the 2013 Online Fraud Report and survey from CyberSource, merchants with online revenue below $5 million per year manually reviewed 42% of orders while merchants with online revenues exceeding $100 million per year reviewed just 11% of orders. Merchants with $5 to $25 million in online revenue and merchants with $25 to $100 million in online revenue manually reviewed 25% and 16% of orders, respectively.
This represents a big issue when it comes to scalability. If merchants are relying on manual review, what do they do when their business grows or when they have peaks? As a quick fix, or as a tool to look at for sales conversion, manual reviews are very effective. When it comes to fraud prevention, merchants need to be very careful how much they rely on them.
If merchants are going to rely on manual reviews, they need to use it to review the orders they were going to reject anyway. This allows merchants to have the possibility of converting an “insult” instead of using manual review to try to catch fraudulent orders.
Drawbacks of manual reviews include:
It does not scale well, the only way to grow is to add new staff.
The more people doing reviews the more unpredictable the results, as each will have varying levels of experience and will adopt their own styles for looking for fraud.
Merchants typically lack formal training for review agents and staff.
THE FRAUD PRACTICE
KEY NOTES
Alternative Solutions - Rules engine, fraud scoring, hot lists, warm lists, positive lists and consumer authentication
Building this In-House - Organizations should have a queue to manage the sorting and assignment of orders that are pending review. They will need to have staff on hand to perform the manual reviews and provide various tools and services for assessing order risk. Organizations will additionally need to make several determinations including when there will be staff on hand to perform manual reviews, what the optimal percentage of total orders to be reviewed is, what risk thresholds or signals will trigger review, and many others.
Estimated Cost - Building this in house from scratch will require significant investment including infrastructure and staff. Outsourcing manual review to a third party is typically charged on a per transaction basis.
Sample Venders - For the software interface for performing manual reviews, see Case Management Platform.
For vendors that will screen (automated and/or manaully review) orders and take on the riskso you do not have to review them, see Guaranteed Providers.
MANUAL REVIEW TECHNIQUE OVERVIEW
Manual review is a technique that can be performed in-house or may be outsourced to or managed by a third party vendor. In either case, staff members perform manual checks on orders to determine the authenticity of an identity and transaction to establish which orders are fraudulent. Typically the best (lowest risk) and worst (highest risk) transactions are automatically accepted or declined, and manual review is a means to try and convert all other order attempts where the merchant isn’t certain and requires a closer look.
Key considerations when implementing or buying this functionality include:
Is manual review something your organization will handle internally or will it be managed by a third party?
What is the average review time per transaction and how much time will it add to the purchase process?
What percentage of fraud do internal review agents or a third party vendor have to catch to make up the difference in their cost?
How many customers will be falsely rejected by the process (sales insults)?
What tools will be provided to the internal team or used by a third party service to conduct the manual reviews?
If outsourcing manual reviews, are you paying per review or only per accepted order?
If outsourcing manual reviews, are manual reviews performed 24/7?
If outsourcing manual reviews, who is liable for transactions that are reviewed and accepted but turn out to be fraudulent?
HOW DOES IT WORK?
For in-house manual reviews merchants set up a fraud-review team. Typically this team will work under the Call Center or the Finance Department. The organization should have a queue in place to manage orders while they are pending review, and this may include features for sorting and assigning transactions to review across review agents. The fraud agents or manual review team will be charged with reviewing all of the company’s orders in the queue to determine which ones are fraudulent and which to approve. The team should have tools they can use to look for suspicious activity. This can include link analysis tools, manual reverse lookup tools, and tools to look at activity by region, credit card number, or order size.
When using a third party service for manual reviews, an organization will send the orders they are not sure about to the vendor rather than an internal queue. The vendor will often incorporate both automated and manual checks to screen the order and assess the level of risk. Vendors will generally have an SLA in terms of review and response time and will indicate a response back to the organization on whether or not they should accept the order. If the vendor chooses to decline an order the merchant may perform additional checks to try and salvage it, but if the vendor offers any liability protection or guarantee against fraudulent orders this will not be available for orders they declined.
When the team finds a suspect order, whether an internal review staff or third party, they may simply reject the order or they may attempt additional follow ups to try and convert the order. Some of the common tools used in conducting a manual review are:
Performing reverse lookup authentication on customers name, address and/or phone number
Calling the bank to verify the shipping and billing information
Calling to verify the order data with the consumer or using a phone verification service (Note: you must authenticate a phone number before you verify it)
Checking historical purchase records to see if there is any other activity from this consumer, shipping address, credit card number, email, etc., etc. in the past
Requiring identity document verification to approve the order
Requiring the consumer to pass an out-of-wallet check or other strong authentication technique
HOW DO YOU USE THE RESULTS?
As a best practice, a very effective way to maximize sales is to set up automated processes to weed out the orders a merchant would not automatically accept, and then use reviews on what is remaining. This also allows a merchant to scale better, since they are only reviewing orders they were going to reject anyway. For a merchant that is only accepting or declining orders in an automated fashion today, sending the orders they would otherwise reject to a vendor who performs manual reviews provides an opportunity to convert these order attempts and increase sales.
Organizations will also want to track several metrics related to manual review to measure and assess review performance over time. There are metrics that measure the health and performance of manual review processes overall, and for in-house manual reviews organizations should additionally track performance by individual review agents. This includes metrics such as the percentage of transactions being sent to review, the review accept and decline rates (aggregate and per agent), the average time an order is in the review queue (or average response time for third party vendors), the average time spent per review (aggregate and per agent), and the percentage of reviewed accepted orders that turn that end up being fraudulent (aggregate and per agent).