Can you outline why your team embarked on this project and the problem that needed to be solved?
According to PwC’s Global Economic Crime and Fraud Survey in 2018, procurement fraud is one of the most common frauds across industries. As such, DBS Group Procurement & Services (GPS) recognizes the need for constant vigilance and the importance of enterprise wide anti-fraud controls. Year after year, we challenge ourselves to revisit and enhance our anti-fraud controls. With the migration to our new Procure-To-Pay system (P2P) completed in 2018, spend management is on the uptrend and more procurement transactions are now electronically captured through P2P. As our procurement data accumulates, we started to analyze them for anomalies, albeit manually. This led to the creation of rulesets for automation of transaction surveillance so that monitoring of transactions can be carried out more regularly.
In the past, our procurement operations team spent 170-man hours (or about 0.1 Full-Time Equivalent) monitoring transactions using spreadsheets, scrutinizing anomalies to prevent purchase order splitting and creating a variety of reports for management and stakeholders’ review. Through the automation of transaction surveillance, we are now able to achieve 81% reduction of employee toil and to redeploy manpower to new projects.
How were things done originally and what was the inspiration to innovate the process?
We realized that staff may not be fully aware of the requirements of the Group Procurement Policy, Standard and the procurement process which resulted in the following challenges:
a.) No contracts signed for procured goods and services
b.) Maverick spend (non-PO backed invoices or outside of P2P – no spend control)
c.) Same staff sourcing and approving
To compound on these problems, there were two additional challenges: (i) in the invoicing or payment phases, duplicated invoices detection was not systematic; (ii) there were risks of data input errors in the P2P.
Embracing DBS’ digital and sustainable culture; and with the blessing of our Regional Head of Procurement (Donna Trowbridge), we decided to build this capability internally ourselves. GPS Risk Team got excited and attended trainings and self learnt Qlikview, PowerBI, R & Python. We selected Python due to its flexibility and portability. As we gained success and momentum in the use Python, we provided showcases in our Agile sprints to key stakeholders. These drew interests in our capability which we can deploy to other use cases such as automation of key processing areas.
What KPIs did you use to measure success for this project? (For example: performance, customer satisfaction, revenue, sales or relevant financial gains?)
i.) Through time and motion study on employee toil, we measured staff productivity (employee hours) saves.
ii.) Through performing regular user surveys (3 x ‘U’s: Useful, Usable and Used), we sought to enhance the employee journey and experience
How you plan to ensure that the new model remains relevant and adapts to the future needs of the market?
In partnership with our Group Compliance control function, we keep close tabs with the latest fraud trends globally. We will continuously add on and refine the rules into our data model. Future roadmap covers the expansion to include account payable and expense data. Over time, we will correlate the procurement end to end data to more effectively surface outliers for Case Management with stakeholders. Through such outreach efforts, we progressively educate our internal procurement users and continuously seek opportunities to reduce our fraud risk exposure.
What advice do you have for those who may want to implement this innovative approach in their own organizations?
i.) Be data – driven. Adopt the 4A approach for data management: Ask -> Acquire -> Analyze -> Act
ii.) Be bold to experiment, make mistakes, iterate/pivot
iii.) Engage your stakeholders in this employee journey as early as possible (listen and incorporate their feedback)
How did your team assess the risks/potential for your third party management strategy?
Several options were evaluated for transactions surveillance:
Option A: Partnering with DBS Technology Support unit to build in-house – this process helped clarify our requirements but was going to cost us several hundreds of thousands SGD to build a platform from scratch to meet 100% of our specifications
Option B: Looking outwards to buy an established procurement analytics tool (Tool A) from a market leader –this tool can meet 50% of our requirements but will cost us almost double of Option A
Option C: Exploring adding a module to our existing P2P system – which was cheaper than Option A & B but was only able to meet 10-20% of our requirements
Option D: Do-It-Ourselves (DIO) & building our own procurement analytics capability– using available open-source analytic tool to build rulesets to identify outlying procurement transactions and trigger notifications to procurement users and approvers and work with them on remediations (Case Management)