Success story

Automating tax processes for an international bank

By Kevin Maroney

A multinational financial services organization turned to Keyrus for assistance in improving their data processes as well as how they present their analysis. 

Prior to this engagement, real-time tax provision forecasting required multiple teams and manual manipulation, which meant there was a high risk for human error and breaks in the audit trail. The different teams were responsible for different data (forecasts from the FP&A, actuals from finance, tax rates and discrete items from tax). No one person understood the full process from source to PowerPoint, so providing data quality assurance was impossible. 

As mentioned, the end user was receiving the final product of the tax forecast outlook in a PowerPoint. The CFO and other senior management wanted to understand how the updated forecasts were deviating from the plan and, more importantly, what was driving that deviation. 

This required significant manual time and effort in creating the visuals in Excel, manually analyzing and checking to see the largest variances, and then highlighting those in an ad-hoc section of the PowerPoint. This was long, tedious, manual, error-prone work, and it was expected to be done monthly.

Solution

Keyrus created an automated, repeatable process using Alteryx to ingest the data from the different teams and create a data source that the new Tableau dashboard would connect to directly. 

Keyrus recommended an incremental approach. The ideal state would be to go directly against the source, but due to security regulations, this was not doable at the given time. The compromise was to create a process that required a user to drop raw extracts into a landing zone and simply run the Alteryx process as is. 

This process did not need to be adjusted each month because it was built in a generic and configurable fashion. The process archived raw files and implemented data validation checks to ensure no data reached the dashboards unless it met all requirements. The process provided the tax team an easy way to adjust tax rates to predict tax amounts with changing rates. This entire data process was built using Alteryx.

The next main initiative was to sunset the historical way that they were providing this report to the CFO. Using Tableau, Keyrus created a dashboard that highlighted tax variances from update to update. Using the tools native drill-down ability, the dashboards allowed users to easily find variances and drill into the detail to understand the drivers of these differences (were these differences tied to NIBT changes or were these due to different discrete tax adjustments?). This removed the need to format and manually build this report each month (and even the existence of the report all together). 

We also revamped what was once an Excel report on the derivative holdings to a Tableau dashboard. Simply being able to roll up, filter, and drill down at different levels of detail (ForEx vs. Interest Rate Swaps, only products with over 100M Notional, products maturing in the next month, etc.) greatly increased efficiency for end users.

Keyrus implemented these tools as well as a full set of documentation and best practices around them. Skilling the internal team and making sure they were able to support the solution was a top priority that was accomplished as part of the engagement. 

Results 

The initial process was highly manual, meaning there were many questions about where metrics came from and how to assess the data quality. We were able to consolidate this tiered process into two simple steps: place files in the landing zone and run the Alteryx workflow. The resulting process is more efficient, scalable, and has a much lower risk for human error. The company saves two weeks per month of manual effort and eliminated the average five monthly mistakes that slowed down reporting. The c-level can also drill into the numbers to better understand forecasting changes without relying on the operations team to take time to explain results. Overall, the organization has significantly more confidence in their data.