MCInnovative Design
WorkAboutLab

Open for thoughtful work

Matt - Principal Product Designer & Leader

uxmatthew@gmail.com

LinkedInGitHubDownload CV
OverviewNarrativeDecisionsOutcomesReflection
04·Consumer Accounting Platform·Shipped

How do you fix a product that scaled by helping users make mistakes?

Reconciliation was a systemic problem in QuickBooks, causing duplicate entries and incorrect top-line numbers. I led a multi-quarter initiative to overhaul how accounting worked; generating design principles, strategy, and a new UX.

Role: Principal Interaction DesignerDuration: 6 monthsTeam: PMs, Designer, Content Designers

Customers delighted

1,000+

QuickBooks data connections interface showing banking feed, duplicate alerts, and SmartCheck reconciliation tool.

Problem

Customers were unable to pull data into QuickBooks and reconcile it accurately. Users were frustrated by the differing approaches to the same actions and surprised by many errors in their top-level business numbers. Accountants were reporting that the data in QuickBooks often required tedious clean up of duplicates and unreconciled transactions. I was tapped to review and own changes to the entire experience.

Core tension

Our system was not quite as smart or simple as it appeared to users. Duplicates were quickly compounding. We urgently needed to put up guardrails, fix the root cause, and support reconciliation.

Overview of user frustrations and accounting errors in QuickBooks, including duplicate entries and unreconciled transactions.

Near-term focus

Recreating the core infrastructure and creating new tools would take significant time and serious design work to get right. In the meantime, we had a team ready to act fast and slow the bleeding.

Banking feed showing a friendly Heads Up alert inserted to slow users when they are about to duplicate a record.

Guardrail: Banking feed alert

It was well known that the majority of users had duplicate entries, due to the many different ways data was coming in. To prevent this problem from growing, we inserted a friendly alert to the Banking feed, to slow users when they were likely about to duplicate a record.

Contextual guidance

For users who engaged, we followed up with guidance pop-ups to show them how to proceed. We didn't like using the word 'match,' as it was clearly confusing users. We were later able to replace it throughout the product, but in the meantime, we worked to clarify it in context.

Follow-up guidance pop-up explaining how to proceed and clarifying the meaning of 'match' in the product.

Aligning on design principles

There were many internal experts with strong opinions about how all of this should actually work. I led a series of workshops to capture and align on design principles.

Design principles document and workshop outputs used to guide UI and content decisions for the data connections area of QuickBooks.

Principles and process

I worked closely with my Content Designer to synthesize hundreds of stakeholder notes, and created a series of design principles to guide the UI and content design. I created a presentation deck and shared it broadly with leaders and stakeholders. Illustrations were pulled from a repository of brand assets.

A concept to aim for

This process inspired strong ideas for improvement. I created a concept flow to visualize the ideal experience, where data would appear alongside its relevant context in the product, no longer siloed in a separate review area.

Concept illustration showing proposed flows where incoming data surfaces in context across the product rather than in a siloed area.

Proposed system vision

I also proposed a system where users would be walked through a series of simple questions and pointed to the correct next step. Users loved this simple direction in user testing sessions, and there were no concerns from stakeholders about the approach, but it would take some time to build the infrastructure. In the meantime, many users already had accounts full of duplicates and unreconciled transactions that needed addressing now.

SmartCheck

We invented a tool to solve for the last-minute tax-time crunch, taking users through all of their incomplete and likely duplicate records one at a time. Through many explorations and iterations on content and UI, we landed on a simple, focused experience we called SmartCheck.

SmartCheck tool showing a focused one-transaction-at-a-time interface for resolving duplicate and incomplete records at tax time.
Money-in and money-out tools showing transaction reconciliation workflow with visual relationship between transaction types.

Monthly workflow tools

I devised a new tool for reviewing and reconciling records, but in order to keep it simple and map to the existing mental models, I split it into two tools, one for 'money-in' transactions, and another for 'money-out.' For the first time, users had a chance to visually see the linkage between bank transactions and other record types in the product, and intuit how and why they needed to be reconciled.

Key decisions

The most important parts of this work were the decisions that shaped how the product would scale, how teams would collaborate, and where the interface should stay quiet or become explicit.

question we faced

Should the cleanup tools be merged into a single workflow or separate features?

Options we considered

  • Single workflow

    Lower scope, but doesn't match users separate monthly and annual routines.

  • Separate features

    Higher investment, but creates lasting habits and prevents the problem from returning.

What we chose and why

We shipped as separate features: SmartCheck for the immediate tax-time backlog, and dedicated money-in / money-out tools to integrate reconciliation into users' regular routine.

What we learned

Addressing the immediate pain earns trust. Addressing the root cause earns loyalty. When possible, ship both layers.

Outcomes

Reduced errors

25%

Customers delighted

1,000+

Framework impact

Multi-year

Design principles adopted

Cross-org

Reflection

  • Running alignment workshops first was the right call; it let us bypass entrenched opinions and move fast.
  • Should have pushed for the in-context data vision to proceed in parallel — the cleanup work made it easier to deprioritize.
  • Labels matter; Challenging conventions and driving new alignment was a lot of organizational work, but ultimately worth it.

Next case study

Agentic security review interface showing a Trust Center with an AI side panel for automated document analysis and checklist matching.

B2B Security & Compliance Platform

Agentic future of security reviews

Solo discovery and design that validated a new agentic tools and UI paradigm for security reviewers, estimated 1–2 hours saved per review.

Read next case studyView all