Our client is one of the world’s leading electronics manufacturer with production in Vietnam reaching US$200 billion in 2020.
Our client designed a process to perform eInvoices verification to prevent payment discounts / bonus conditions bias. It required fully manual data checks related to 40,000 eInvoices versus data recorded by retailers on a monthly basis to confirm the discount and promotion calculated to customers.
NashTech identified resource issues that were affected by the volatility of data volume from no data on one day to 10,000 data points received on another. There was a contractual expectation of 98% data checks accuracy and a need to provide continuous service excellence.
Standardisation was applied before progressing with the technical design of an automatic validation solution pairing Microsoft Virtual Agent with UiPath robots in real time.
We delivered a phased automation over a two-month period to minimise any disruption to the operations and rolled out to the private cloud to ensure the solution is secure, easy to maintain and scalable.
The automation adoption followed agile methodology and a scrum framework. Post production release followed a hyper care period which ensured sustainable automation running and support.
We have designed and developed a highly flexible solution corresponding to a need of randomised data pools being validated against various promotion programmes rules. It’s a quality controller who steers automation on which pools to be checked and should it be done with end-to-end validation or only on some defined steps.
The engine of our automation solution is a robotic process automation software platform, which can be triggered by a quality controller filling in daily validation preferences to robotic operating log or via dialog with chatbot on Microsoft Teams. Resources supervising and auditors can access daily processing outputs in a real time.
The risk of missing any payment conditions incompliances has been reduced thanks to the reduction of errors. Now that manual efforts have been reapplied to focus on nonstandard scenarios the accuracy has also increased.
The key metrics have demonstrated a continuous improvement in monthly accuracy levels providing results greater than 99%. In ten months of running the bot, 67% of all validation activities were handled autonomously by the bot leaving the team more time to investigate nonstandard scenarios.
The time a bot takes to process and verify a single invoice has been reduced by 78%, from 90 seconds to 20 seconds. Time gained enables the operations team to spot audit inaccuracies faster and accelerate the customer experience by applying and communicating corrections in days rather than weeks.
