A mobile app that helps self-employed individuals automatically manage their finances. Hurdlr tracks your mileage, expenses, income and provides you with real-time tax estimates so you’re never surprised come tax time.
One new feature, Deduction Finder, tackled a problem for business owners of being overwhelmed with their numerous expenses that loaded in after linking their bank account.
We found that Hurdlr users often missed deductions that could have easily been included on their tax return, but sometimes the amount of expenses to track can become daunting. Small business owners wear multiple hats and are often spread too thin. This can occasionally cause them to not tag every single expense as Business/Personal. That being said, the question I asked was:
How can we help users automatically identify which expenses are VERY LIKELY to be business deductions?
Utilizing machine learning, finding deductions that are most probably a business expense has now been automated. The incredible engineers I’ve worked with utilized machine learning to automatically find expenses that are MOST LIKELY to be business expenses.
This allows our users to exponentially drop the amount of time needed to go through all their expenses and determine for themselves.
In lieu of user research, I relied on my CTO’s subject matter expertise to come up with a bunch of hypothesis’ to do the initial design work (to just get it out there fast) and then leaned heavily on user acceptance testing after you shipped to validate the hypothesis. We did this via Beta testers recruited by our Customer Success team.
As a product, Hurdlr has several primary users to tailor the user experience for; getting to know who we’re designing for is especailly important. For the Deduction Finder specifically, we wanted to target Small Business Owners, Real Estate Agents and Rideshare Drivers.
After a few iterations, we had a flow we were initially happy with, so we decided to test this with users.
2 weeks later, we came back receiving good results with use, but we found that a significant amount of users skipped the part where they had to review the "NEEDS REVIEW" deductions (ironically). So, we removed that step entirely and let them go into their "NOT LIKELY" deductions instead. The decision saw a ~22% increase in users completing the entire Deduction Finder.
This feature helped our users filter through thousands of more expenses in a fraction of the time and headache. Link to prototype.
If you’d like to see more of the Deduction Finder UI and Prototype, please don’t hesitate to contact me.