tekion Corp 2024

Reduced time taken by accounting clerks in reconciling bank transactions and respective accounting entries

New bank reconciliation interface showing a side-by-side comparison of Bank Statement and GL Postings with selected transactions for reconciliation.
Team
Product Designer (Myself)
Lead Product Designer
Product Manager
My Role
Research
Conceptualization
Design
Dev handoff
Duration
6 Months
Background & problem

Still Stuck in Manual Mode While Business Needs Evolve?

Bank Reconciliation serves as a critical daily process where dealerships match their bank transactions with account entries in their ERP/accounting software, specifically Tekion’s ARC.
While our Bank Reconciliation application has effectively handled manual operations for our current user base, our expansion to larger enterprise clients necessitates automation of key tasks and the development of advanced features to meet growing demands.
Old bank reconciliation interface displaying a PDF bank statement from HDFC Bank on the left and a list of accounting entries on the right.
Old reconciliation process—Bank statement on the left; GL Posting on the right.
Users reported that it takes 2-4 hours to reconcile ~100-200 transactions.
Painpoint 01
Users struggle with complex scenarios like partial payments, split transactions, and transactions with timing differences.
Painpoint 02
Users lack visibility into the mapping between individual bank transactions and their corresponding accounting entries.
Painpoint 03
Focus Area

Let's first focus on establishing the essential foundation.

BAI2 file format was launched in 2001 (one year before I was born) and yet our product was not utilising the features of this file format in 2024, presenting us with significant untapped potential for enhancement and innovation.
We’re living in the age of ML and AI—at the very least, the easy stuff should be automated by now.
Illustration of a white document icon labeled '.bai2' centered on a bright orange to yellow gradient background. The icon has a folded corner, symbolizing a file, and represents the BAI2 file format commonly used for electronic bank statements.
Introduce .bai2 file format
White sparkle icon with two diamond-shaped stars on a vibrant orange to yellow gradient background, symbolizing artificial intelligence and automation.
Automate some of the reconciliation
Challenge 01

Solving the biggest challenge

Banking reconciliation presents four distinct scenarios: reconciling a single bank transaction to one accounting entry; reconciling a single bank transaction to multiple accounting entries; reconciling multiple bank transactions to a single accounting entry; and reconciling multiple bank transactions with multiple accounting entries.
Developing a solution that elegantly handles all these complex matching patterns represented one of our most significant technical challenges.
Two tables side by side showing the mapping of bank transactions to accounting entries, connected by an arrow. The left table lists bank transactions such as 'ACH Deposit - ABC Auto Finance' with amounts, while the right table shows corresponding accounting entries like 'Ford F-150 sale to John Smith' with matching amounts. The $25,000 entry is highlighted in both tables.
One-to-One Matching Scenario
Two tables side by side showing how bank transactions map to accounting entries using gray arrows. The left table lists various bank transactions with corresponding amounts, such as 'POS Credit - City Bank' for $22,500. The right table breaks this amount into accounting entries like 'Honda Civic sale to Mary Johnson' for $20,000, 'Extended warranty' for $1,500, and 'Window Tinting service' for $1,000. Matching rows are highlighted in beige.
One-to-Many Matching Scenario
Two tables side-by-side comparison of bank transactions and accounting entries with gray arrows connecting related items. The left table includes a $15,000 ACH deposit from Metro Credit Union and a $3,000 wire transfer from David Williams. These are mapped to a single $18,000 accounting entry on the right labeled 'Toyota Corolla sale to David Williams.' Matching rows are highlighted in beige.
Many-to-One Matching Scenario
Matching visualization between bank transactions and accounting entries. Seven bank transactions are listed on the left, including entries like a $25,000 ACH deposit from ABC Auto Finance and a $1,200 wire transfer from Robert Brown. On the right, corresponding accounting entries include a $25,000 Ford F-150 sale and a $2,000 vehicle deposit for Robert Brown. Multiple gray arrows connect bank transactions to entries such as 'Toyota Corolla sale to David Williams' and 'VSC refund,' with related rows highlighted in beige.
Many-to-Many Matching Scenario

Initial Approach : Splitting & Consolidating Transactions

During our competitor analysis, we observed that QuickBooks offers a feature that comes close to addressing our problem. With a few thoughtful adjustments, we believe we can offer an even better solution.
Specifically, by giving users the ability to split a single entry into multiple entries or consolidate multiple entries into a single one, we thought we can significantly streamline the reconciliation process and better meet user needs.
Bank Reconciliation interface displaying a split view. On the left, a PDF bank statement from HDFC Bank is shown with transaction details like dates, withdrawal amounts, and deposit amounts. A $621,177.31 EFT entry is selected with options to either 'Split Amount' or 'Consolidate Postings.' Below, a breakdown of transactions appears with colored dots indicating their reconciliation status and a difference amount highlighted in red: $114,699.07.
However, in practice, it introduced additional complexity and made the reconciliation workflow more cumbersome, rather than simplifying it.

Complexity in Splitting & Consolidation

When it came to splitting entries, the workflow was relatively straightforward: the user could select an accounting entry, input the split amounts, reference numbers, and transaction types — a simple and intuitive process.
However, consolidating entries introduced several challenges. Key questions emerged, such as:
  • How should users select multiple entries for consolidation?
  • What should the affordance and signifiers look like to make the consolidation action clear and discoverable?
  • How would the system handle conflicts or mismatches between the selected entries?
These considerations made the consolidation workflow significantly more complex compared to splitting, requiring more careful UX design to avoid user confusion and workflow breakdown.
Challenge 02

Distinguishing Between Reconciliation Statuses

We also needed to display the reconciliation status for each entry — whether it was auto-reconciled, manually reconciled, or yet to be reconciled.
During discussions, the PM raised an additional point:
"Areyy! If a user can un-reconcile an already reconciled entry, shouldn't we also show an 'Unreconciled' status?"
As more statuses kept getting added, the system was at risk of becoming too complex and overwhelming for users, making it harder to maintain clarity and simplicity within the workflow.
An 'Un-Reconciled' status didn't make sense in the current design, as reconciliation is simply a check or uncheck action.
The Bigger Picture

Modernizing Bank Reconciliation with BAI2 Integration

By introducing BAI2 file support, we enabled automatic reconciliation through intelligent matching of transaction amounts and types. This also opened the possibility to display bank statements directly within the UI by parsing the imported data.
I revamped the reconciliation interface with an intuitive two-column layout, enabling users to directly compare bank statements with accounting entries within the system, replacing the outdated manual process of uploading and cross-referencing PDFs.
Getting buy-in from the PMs and securing additional development bandwidth was a challenge, but it was essential to deliver the improved experience.
Bank Reconciliation interface displaying a split view. On the left, a PDF bank statement from HDFC Bank is shown with transaction details like dates, withdrawal amounts, and deposit amounts. A $621,177.31 EFT entry is selected with options to either 'Split Amount' or 'Consolidate Postings.' Below, a breakdown of transactions appears with colored dots indicating their reconciliation status and a difference amount highlighted in red: $114,699.07.
Old Reconciliation Page
New bank reconciliation interface showing a side-by-side comparison of Bank Statement and GL Postings with selected transactions for reconciliation.
New Reconciliation Page

But how does this solve our biggest challenge?

This new approach removes the need for users to split or consolidate entries during reconciliation. Instead, they can intuitively select the relevant bank transactions on the left and directly match them with corresponding accounting entries on the right.
Previously, reconciliation was treated as a simple toggle — checking or unchecking entries to mark them as reconciled. While lightweight, the old design lacked accuracy. With the new interaction, reconciliation becomes a deliberate, aligning with the way accounting should work: precise and intentional.
New bank reconciliation interface showing a side-by-side comparison of Bank Statement and GL Postings with selected transactions for reconciliation.
Reconcile once Statement Amount & GL Posting Amount matches

Solving the Reconciliation Status Confusion

I explored a range of interaction patterns — from using tabbed layouts to experimenting with visual indicators — all with the goal of reducing friction and making reconciliation statuses easier to scan and act on.
Each direction aimed to balance clarity with efficiency, ensuring users could stay focused without any unnecessary clicks or confusion.
Tabs meant more clicks than necessary. We saw room for a smarter solution.
The indicator for ‘Manually Reconciled’ was straightforward — a simple check mark inside a circle. But designing a clear, non-intrusive visual cue for ‘Auto Reconciled’ was a pain in the a**. Still, after rounds of iteration, we cracked it.
After speaking with multiple subject matter experts, we realized one key insight — users don’t really care about what’s unreconciled. Their focus is on what’s matched and what still needs action.
Manually Reconciled vs. Auto Reconciled (Various Iterations)
Small Tweaks, Big Experiences

Improving 1% of every aspect

I didn’t stop at solving just one part of the problem. I took a step back and evaluated the entire bank reconciliation experience — from selecting an account, uploading and processing a BAI2 file, to displaying only the most essential information during reconciliation, and finally, presenting users with a clear, detailed reconciliation summary.

Prerequisites of Bank Reconciliation

Old Workflow
New Workflow

Bank Reconciliation Header

Nothing more than the essentials
Nothing more than the essentials

Reconciliation Summary (Download/Print)

Old PDF with little to no information
Improved PDF with detailed matching information

Reconciliation Details

We took this opportunity to build a Reconciliation Details page
Signing off

Waiting for the report Card (metrics)

The revamp launched less than a month ago and we're currently monitoring early performance metrics. Key features implemented to address user pain points:
Users reported that it takes 2-4 hours to reconcile ~100-200 transactions.
Painpoint 01
Users struggle with complex scenarios like partial payments, split transactions, and transactions with timing differences.
Painpoint 02
Users lack visibility into the mapping between individual bank transactions and their corresponding accounting entries.
Painpoint 03
With the capability of auto reconciling more than 50% percent of the transactions, we believe that the time taken will be reduced to ~1-2 hours.
Solution 01
With the ability of selecting multiple transactions or entries to reconcile, this solves the complex matching scenarios.
Solution 02
With reconciliation details page, users get to know exactly which bank transactions are mapped with which accounting entires.
Solution 03

key takeaways

This project was jam-packed with a lot of work, explorations and discussions with stakeholders. It feels almost impossible to boil down my learning, but if I had to, here they are!
You won't have the complete requirements upfront. You'll need to collaborate with the Product Managers to define and improve them throughout the project.
Learning 01
You have to work closely with developers to understand technical constraints early and iterate on designs based on implementation realities.
Learning 02
You should not be constrained by existing design system components - you can create custom solutions when user needs require it.
Learning 03