Arnold

Inspiration

The audit industry has always been notorious for its long working hours, tight schedules, manpower shortages and signature peak periods. Arnold is here to make a change.

What it does

Arnold serves as a booking platform for audit clients, where they are able to schedule audit engagement periods. Through machine learning and surge pricing, Arnold is able to optimize the appropriate audit fees for each unique client based on factors as the company's size and level of audit complexity.

Transparent Dynamic Pricing

Based on surge pricing similar to Grab, a surcharge would be imposed when clients book peak periods. Penalties are also imposed on last-minute changes in audit periods.

In the short-term, this incentivizes clients to shift audit periods to less-peak dates, reducing manpower shortages in accounting firms.

In the long-term, clients are encouraged to change their financial year-end to align with less-peak periods.

Audit Fee Optimization via Machine Learning

Utilization of machine learning to analyse features such as company size, audit risk, audit complexity, etc to optimize and predict audit fees.

Automation of audit fee pricing removes subjectivity and improves transparency of certain audit planning processes, which helps manage clients' expectations.

Automation of Audit Scheduling

Managing clients’ expectations through hassle-free booking of audit engagements. This provides a seamless user experience from audit planning to the start of the audit engagement.

How we built it

Model Overview

The core objective of our machine learning model is to determine, on each available engagement date, the probability that a client will accept an engagement proposal of a variable price. In other words, we want to accurately estimate the engagement demand curve at various prices. The difficulty of applying the conventional demand-supply model to this scenario lies in the lack of data available: audit companies are not incentivized to lower prices to find out if clients are open to a change in financial quarter or engagement date, and vice versa. Without client valuation data available, the bayesian-optimal mechanism is suitable to determine the optimal pricing for a particular day.

The bayesian-optimal mechanism predicts optimal prices most accurately if the provided probability distribution (i.e. demand curve in this context) is as accurate to the client’s behavior as possible. In order to achieve this, we employ a deep learning model to generate the curve based on real-world inputs; after every engagement, we collect client-specific data such as company size & revenue, company sector, and various risk assessment metrics against the client’s action (client’s action refers to whether the client has accepted or rejected the proposed optimal price for that particular date). We also aggregate data from the behaviors of companies with similar profiles, and input these information as hyperparameters to the deep learning model.

Challenges we ran into

The core objective of our machine learning model is to determine the probability that a client will accept an engagement proposal at a certain price on a certain date. In other words, we want to accurately estimate the engagement demand curve at various prices. The difficulty of this method lies in the lack of data available: audit companies are not incentivized to lower prices to find out if clients are open to a change in financial quarter or engagement date, and vice versa.

Accomplishments that we're proud of

For some of us, this is our first hackathon, and we are proud to be able to apply our accounting knowledge to better the industry.

What we learned

Through this hackathon, we learned that the accountancy industry faces several unique challenges that require diverse skills to solve.

What's next for Arnold

Global expansion to ride on the wave of technological boom in accounting services.

Creation of a marketplace with transparent audit pricing for audit companies and corporate clients.

Integrated internal talent management system to further enhance and optimize manpower allocation, increasing efficiency and productivity.

The one-stop solution for fully-digital audit planning.