R2Z Implications
| Initial Data Exploration and Analysis | Deeper Data Exploration and Analysis | Setting Benchmarks for Merchant Performance | Limitations |
|---|
Funnel plots and line of best fit graphs are basic tools of exploratory data analysis, but have not been widely used in analyzing online retail transactions. Their abilities to compare the performance among different merchants in a group in an unbiased manner makes them useful in setting performance benchmark for each merchant.
On the other hand, logistic fit graphs allow us to understand the correlation between independent variables and key dependent variables, though causality is not determined. Decision tree analysis can then be used to identify optimal cut-off values for controllable independent variables of a transaction that would improve each merchant’s approved-to-rejected transaction ratio.
Thus, this study showcases both reactive and proactive approaches that our sponsor’s merchants may adopt to improve their performance.
In summary, we have discovered two general trends:
• As the number of transactions increase, the number of approved transactions increase.
• As the monetary value of a transaction increased, the approved-to-rejected transaction ratio decreases.
Our analysis provides insights on how to reduce risk borne by payment processors. For over performing merchants, our sponsor can invest more resources to retain them and acquire similar merchants in the future. For underperforming merchants, they can investigate and set benchmarks to improve their individual performances, and recommend cut-off values for transactions to improve approved-to-rejected transaction ratios.
In future, other metrics may also be considered in evaluating merchant performance – e.g. revenue per merchant. Our sponsor can expand on their dataset to gather more data information that provide a better picture on the underlying factors which affect approved/rejected transaction ratio.
