Intuit’s Machine Learning Helps You Save More Money


Managing money can be a stressful, time-consuming process — but for individuals around the world, there’s never been a more important time to do it effectively.

That’s because the global state of personal finance looks grim. Debt worldwide is ballooning, and for many that takes the form of private loans like credit card debt and student loan debt (particularly in the U.S.). According to one estimate, the 2018 global debt was equivalent to more than $86,000 per person — and more than double the average per capita income.

Another key global financial issue? Retirement. With the average lifespan at around 70 for males and 75 for females, people need more money to retire — money that many don’t have. In many countries, including Japan, that means putting off retirement to 70 or 75. In other countries, absent pension programs, many people don’t have the money to retire. In the U.K., 22% of people don’t expect to be able to afford retirement. In the U.S., 15% of people don’t have retirement savings at all.

Intuit Wired

Courtesy of Intuit

While the solution to some of these challenges extend beyond money management, getting as much as possible out of existing assets and other financial activities should be high-priority. Intuit, the financial software giant behind QuickBooks and TurboTax, has long worked to create tools that make financial management easier, less time-consuming — and more beneficial for their customers.

“For a lot of people in the U.S., their tax refund is the largest check that they get in the mail,” says Ashok Srivastava, a senior vice president and chief data officer at Intuit. “Our aim is to help our users find more deductions.”

Recently, Intuit has declared its strategy to become an AI-driven expert platform, using AI technologies to help its customers find more money with less work and more confidence. As a part of this strategy, Srivastava and his team innovate new and unique ways to incorporate machine learning, natural language processing, and knowledge engineering throughout Intuit’s products. While the in-house data science team focuses on model creation and testing, Intuit implemented Amazon SageMaker in 2017, which cut the time it took to deploy a machine learning model from months to weeks. With this internal upgrade, Intuit has been able to generate a series of custom machine learning models — paired with other machine learning services like Amazon Textract — to create better, more personalized financial management services to help users improve their financial health.

Machine Learning Helps Maximize Your Tax Refund

Taxes are often associated with stress and hassle, but doing them correctly is paramount. Tax return averages in the U.S. fluctuate each year, but often fall in the range of $1,000 and $3,000.

To make sure users are getting the largest possible tax refund, Intuit incorporates machine learning throughout the TurboTax experience to help users more efficiently file their taxes. For example, a custom machine learning model built in SageMaker is aimed at helping users decide between standard and itemized deductions. The algorithm is trained using the 80,000 pages U.S. tax code to recommend the deduction based on an individual’s specific set of circumstances — across certain types of asset portfolios, to individuals living abroad that owe U.S. taxes — all while keeping up-to-date with the latest revisions to the tax code. TurboTax’s custom model quickly adjusts to these changes and guarantees users are making the best elections each year.

Beyond optimizing for the highest-possible return, TurboTax uses machine learning to shorten the filing process, which takes an average of 13 hours. With Intuit’s computer vision capabilities supported by Amazon Textract, entering information from tax forms like W2s or 1099s takes seconds. Rather than a user having to enter form fields manually, the service scans pictures of the forms and digitizes them. Then, using contextual data from TurboTax’s existing database of tax codes and compliance forms, Amazon Textract verifies accuracy and identifies any anomalies or missing data for the user.
And what if TurboTax users need assistance filling out their taxes? There’s a model for that, too. Since tax season has particularly high peaks of traffic patterns for customer service, Intuit built forecasting columns to predict call centers’ demand on a particular day or time to make sure customer service agents are staffed to manage the demand.

Make The Most Out of The Money You Make

Maximizing returns — and just as importantly, managing regular expenses — also comes from tracking expenses year-round. For the average consumer, Intuit’s Mint can monitor transaction activity for personal bank accounts, credit cards, and loan balances. Machine learning automatically categorizes this information to help users see where they’re spending, track their budget, and identify saving opportunities.

For businesses, accurately categorizing transactions is even more critical. Intuit’s QuickBooks has long helped millions of small- to medium-sized businesses manage their finances with its easy-to-use bookkeeping service. But by adding custom machine learning models to its workflow, QuickBooks created processes that protect users from common mistakes. Similar to Mint, a custom machine learning model in QuickBooks automatically categorizes transactions into “personal” and “business” expenses to ensure a business’s taxable income is correct — avoiding a painful audit process and maximizing deductions.

In 82 percent of the cases where a business fails within its first five years, the root cause is poor cash flow management. Beyond the time (and money) saved by automatically tracking expenses, QuickBooks also helps business owners avoid this common pitfall with cash flow forecasting. The model studies a business’s spending habits and using predictive intelligence can project finances into the future, helping business owners determine whether they are on track to afford rent or payroll at the end of each month and course-correct, if needed.

Running a successful business is particularly complicated for the people who work for themselves as part of the growing gig economy. While the rise of freelance work has increased the most in the U.S., the U.K., Brazil, and Pakistan all saw close to 50 percent growth in 2019. For freelancers, filing taxes and manage expenses requires extra time and attention. Often, contract work isn’t automatically taxed, requiring more care in filing to avoid audits and to maximize deductions. Expense tracking is more difficult, too, since gig workers’ income and spending tend to fluctuate. Personal versus business expenses are more blurred. Intuit’s machine learning models help with all of these issues: In TurboTax, for example, the model automatically categorizes personal and business expenses to help users appropriately file claims come Tax Day.

“With the rise of the gig economy, we started seeing a larger demand for personalization across all of our products,” Srivastava says. “People want to see personalized recommendations so that they can make the best financial decisions for themselves and their unique financial situations.”

In fact, machine learning enables Intuit’s products to seamlessly respond to trends — from changes in the tax code to changes in the way people work. Its success to date drives continued exploration into different applications for Intuit products.

“We are on a multi-year journey that started more than ten years ago,” Srivastava says. “We’re making significant investments across natural language processing, document understanding, time series forecasting, personalization and recommendation systems. But at the heart of it, our goal is to use machine learning and AI in order to help customers make the best financial decisions that they can to improve their financial futures.”

Coinbase uses ML to create a secure cryptocurrency exchange

Learn more »

Capital One uses ML to better protect customers from fraud

Learn more »

Zendesk helps companies provide faster customer service using ML

Learn more »

T-Mobile humanizes customer service using ML

Learn more »