Regular rule-set reviews and rule adjustments can reduce the review rate, decline rate, or charge-back rate. They allow agents to use data analysis to make the rule-set more efficient and effective. This article describes the proper setup of a Datamart report to find inefficient review rules in need of modification. Users unfamiliar with Datamart and creating reports should read Create a Datamart Report before continuing.
Step 1: Set Up a Datamart Report
Transaction Date (Monthly) = Current and Previous Months: Looking at only the last two months will give the most up to date results so transactions marked for review by previous rulesets or rules that no longer exist are filtered out. If there is insufficient data expand the filter by an additional month. First, filter the report to display on transactions that were marked for review by the rules agent using the following filters:
Initial RIS Reply = Review: If the initial RIS Reply only includes review this will automatically filter out transactions that were approved or declined by the rules engine, so only transactions marked for review will remain.
Review Count = 1: Review Count = 1 filters out transactions that triggered more than one review rule This isolates transactions to a single rule so any overlap between rules is eliminated. This is important in determining that agents are decisioning the orders based on a single rule rather than a combination of rules or other factors.
Decision = excludes Not Available or Timed Out: This will filter out transactions that were never reviewed by an agent or transactions that have not yet been reviewed. This leaves only transactions that were already approved or declined by an agent. Like Review Count this is important in isolating down to only transactions that were actually reviewed by agents.
Next, lay out the report to display the Rule Description, Decision, and Transaction Counts:
Step 2: Analyze
In the example above the first rule was approved 97% of the time with 221 transactions marked for review being approved out of 227, while only 6 were declined. An approval percentage above 80% means agents are reviewing a high number of apparently good orders, and that a rule may need to be adjusted. Right click on the rule and select “Keep Only….” to view only that single rule.
The rule should be adjusted to flag fewer transactions. Rules can be adjusted in a variety of ways but the easiest and usually most effective is adding a score condition to the rule or by adjusting any other variables already present in the rule. Simply adding a score condition or adjusting the variables can help separate the good transactions from the bad ones.
For example, adding “Original Score by 10s” to the rows field shows that of the 134 transactions with a score less than 39 marked for review 131 were approved. Adding a condition to this rule where Score >39 would eliminate 134 manual reviews.
On the other hand, rules that are frequently declined should be changed to auto-decline. In the following example, the two highlighted rules are being declined quite often, more than 94% and 88% of the time, respectively.
Both rules would be good opportunities for change from review to decline rules.
In addition, breaking out the transaction by score or another criteria allows for the creation of a new decline rule without declining some of the orders that were approved by an agent.
As a priority, focus on rules with a high number of transactions, even if the percentage of approved to declined transactions is close to 50%. Small adjustments to frequently triggered rules will have a larger impact than big changes to rules that rarely trigger.
Article is closed for comments.