How do we make financial reconciliation more visual?

My Responsibility

Led an engineer and designer to redesign the reconiciliation interface. I was responsible for program management high-fidelity prototyping and user-testing.

Duration

6 weeks

Category

User Experience Design, Interface Design, User Studies

The Impact

A Solution That
Helped Users
And Sales People

120K

In Sales Pipeline

80%

Est. dec in Processing Time

The Problem

A Complex
Reconciliation
Process &
Unintuitive
Visualizations

A Complex Reconciliation Process & Unintuitive Visualizations

Reconciling settlement reports from food delivery marketplaces with bank statements is complex. They needed to reconcile with their banks and the marketplace. The then existing design did not help users understand the 2 reconciliation statuses of each transaction. More importantly, it didn't let see actionable insights. I created a design that made it easier for users to visualise different stages of transaction reconciliation in a comprehensible manner.

Issues With Existing Design

Lack of specific details
in pie diagrams

Visuals did not show
relationships between
different data sets

Buttons did not let users
complete necessary actions
with ease

Exploratory Research

We Needed to
Learn Business
Needs …

Focus Understanding how accountants and finance managers reconcile transactions and settlements, what tools do they use and what are they looking for in the complete dataset.

Methods - Interviews

We talked to the people who used the previous dashboard, and the development team of the previous dashboard to understand their context. Additionally we also spoke with heads of finance and audit at larger firms to learn about best practices.

Insights

  • Details were missing At its core, any reconciliation tool needed to clearly highlight potential errors. Currently the visualisation merely showed what % of transactions were not reconciled. It didn't let users see transaction specific details. For this they needed to download an excel file

  • Confusing nomenclature Accountants weren't familiar with the terms of the visualization module. Partly because the initial development team was not familiar with the right terminology. But, this reconciliation use case was also new and didn't have established terminology.

  • Alignment with fundamental accounting principles

    • Traceability Visualizations can help make the flow of transactions or the matching of data between systems transparent.

    • Verifiability Visual tools help you quickly double-check that things are correct.

    • Actionability Highlighting anomalies or discrepancies visually or incorporating collaborative features allows users to take corrective action based on what they see.

    • Completeness Visualizations can help identify gaps by showing missing or incomplete data. This can be particularly useful when reconciling multiple data sources

Sample Questions

How Are You Using
Reconciliation Analytics?

How Are You Using
Reconciliation Analytics?

Is there any part of the process
that feels time-consuming?

Is there any part of the process that feels
time-consuming?

Focus: Understanding how accountants and finance managers reconcile transactions and settlements, what tools do they use and what are they looking for in the complete dataset.

The research was conducted in two parts:

  1. Secondary research: In this part which we studied how data visualisation works, how datasets similar to ours are being visualised and how we could do it more efficiently. We also did a competitive survey to find if any product that showcases a similar idea through a dashboard. But we could not find anything that was free and open to access.

  1. Primary research: In this part we talked to the people who used the previous dashboard, some potential new users who had expressed needs regarding reconciliation services and the development team of the previous dashboard.

Focus: Understanding how accountants and finance managers reconcile transactions and settlements, what tools do they use and what are they looking for in the complete dataset.

The research was conducted in two parts:

  1. Secondary research: In this part which we studied how data visualisation works, how datasets similar to ours are being visualised and how we could do it more efficiently. We also did a competitive survey to find if any product that showcases a similar idea through a dashboard. But we could not find anything that was free and open to access.

  1. Primary research: In this part we talked to the people who used the previous dashboard, some potential new users who had expressed needs regarding reconciliation services and the development team of the previous dashboard.

Secondary Research

… And Borrow
Patterns

… And Borrow Patterns

Focus In this part which we studied how data visualisation works, how datasets similar to ours are being visualised and how we could do it more efficiently.The research was conducted in two parts:

Methods - Online Research & Current User Interviews

  1. Secondary research: We also did a competitive survey to find if any product that showcases a similar idea through a dashboard.

  1. Primary research: In this part we talked to the people who used the previous dashboard, and the development team of the previous dashboard to understand their context.

Findings

To map this type of data either expandable pie charts or funnel charts or Sankey diagrams can be used. This validated the initial idea of funnel charts since funnel charts are Sankey diagrams for simple, linear flows. We found a good example of funnel charts on Atlassian Marketplace by Kanbananalytics. This became the main reference of the interface for rest of the project.

Image source: https://datavizproject.com/

Visualization Options I considered

Ideation - Visualization

I Went
with Reconciliation
Funnels using
Sankey Diagrams

I Went with Reconciliation Funnels using Sankey Diagrams

Paper Prototyping

I created paper explorations of what a sankey chart and expandable pie-charts would look like in situations that needed complex reconcilitation visusalization and tested it among internal users.

Results

Ultimately, we found that Sankey Diagrams' rectangular and vertically expandable shapes made better use of precious horizontal real-estate than pie-charts complex reconciliation scenarios.

Design Reasoning

Sankey Diagrams Were less cluttered
and Showed reconciliation flow more intuitively

Funnel Charts Were less cluttered and Showed reconciliation flow more intuitively

Vertically expandable sankey diagrams were real-estate efficient and comprehensible

Vertically expandable sankey diagrams were real-estate efficient and comprehensible

Vertically expandable sankey diagrams were real-estate efficient and comprehensible

Multiple Pie diagrams weren't space efficient

Multiple Pie charts weren't space efficient

Ideation - Dashboard

A Dashboard That
Made The Visuals
Actionable

A Dashboard That Made The Visuals Actionable

Then we started to brainstorm on what features were needed for the dashboard.

  1. Filters

    1. Timeframe: calendar, to choose specific date or time range

    2. Channel: payment, delivery and e-commerce channels, auto chosen as per the merchant

    3. Methods: applies for payment methods only, will not be there in delivery and e-commerce cases

    4. Sub-methods: applies for payment methods only, will not be there in delivery and e-commerce

  2. Amount-Count switch: A switch to change the mode of data between transaction amount and transaction count

  3. Visualisation: A clickable funnel chart with columns for orders, transactions, settlements and mismatches

  4. Table: A table showing a few entries from the selected dataset, all recent transactions. Also has a download button.

Usability Testing

Did Our ProtoType
Align With
The Principles Learnt
During Exploratory
Research?

Did Our Prototype Align With The Principles Learnt During Exploratory Research?

After a clear enough idea was formed, I started the final layout and prototyping of the dashboard. It went through multiple iterations after reviews from product teams. Once this was done we conducted Usability Testing. Our Questions were aligned with the four principles - Traceability, Verifiability, Actionability & Completeness.

Traceability

Can Users See How
Transactions
Flow stage-to-stage ?

Verifiability

Can Users Double-click into
specific transactions to
check specific details ?

Can Users Look Into Transactions To Check Details ?

Actionability

Can Users Take The
Necessary Actions Based
On What They See ?

Can Users Take Actions Conveniently ?

Completeness

Does The Visualization
Show The Full Set of
Transaction across Stages ?

Does The Visualization
Show All Transactions across Stages ?

Usability Insights

Users Needed
Easier Access
to information

Users Needed Easier Access to information

After a clear enough idea was formed, I started the final layout and prototyping of the dashboard. It went through multiple iterations after reviews from product teams. Once this was done we conducted Usability Testing. Our Questions were aligned with the four principles - Traceability, Verifiability, Actionability & Completeness.

Traceability

Nomenclature of
Sankey Diagram varied
across company segments

Nomenclature of
Sankey Diagram varied across company segments

Verifiability

Users did not want to
download
excels to see greater detail

Users did not want to download excels to see greater detail

Actionability

Users needed help
substantiating concerns
with banks and
marketplaces

Users needed help
substantiating concerns with banks and marketplaces

Completeness

Does The Visualization
Show The Full Set of
Transaction across Stages?

Users Weren't 100% Sure They Saw All
Available
Data

Key Changes

The key insights that we found from the testing sessions and the proposed solutions for them are mention below.

We could not incorporate the assigning feature. After proposing these changes our project halted and the product development took over.

Clickability of boxes not apparent to users

Solution: Changing the color saturation of the box and increasing text size in the hover state.

Changes in table not apparent to user

Solution: Changing the colour of the table header according and adding intermediate loading animations before updating the table.

Greater transaction level detail required by user. More actionable features needed as well like assigning entries to accountants.

Solution: Display an overlay that shows mismatch details according to internal and external records.

Terminologies used were unfamiliar to all interviewees. These varied across every single interviewee.

Solution: Display a tooltip explaining the term on 'i' icon. Also give an option to edit the terminology.

Senior interviewees needed someway to import this chart to a powerpoint or other ways of presentation.

Solution: Export option current graphic on-screen to transparent .png, .svg, or .jpeg. Share to Email/Slack option.

Initial Design

I made the first version which is shown below. It had a few issues like wrong UI elements, bad hierarchy and emphasis using layouts.

Final Design

The issue with categories of the data and connections between them was fixed in this version. I also added more information in figures as it was important for the end users. Also fixed the issues with other UI elements.

Reflection

What Went Well

Doing More
With Less

Resources were constrained.
So, I learnt to make compromises that
did not impact the integrity of the
research and designs.

Resources were constrained.
So, I learnt to make compromises that did not impact the integrity of the research and designs.

Balancing Business
& User Needs

PortOne needed traction. I needed to
ensure that we met our target users'
critical requirements. But beyond that
the team needed to focus on sales.

Considering this was a startup, I learnt to make compromises that did not impact the integrity of the user research and designs

Reflection

What I Would
Change

What I Would Change

Dig Deeper Into
Technical Scenarios

Backend integrations were being custom
built. Eventually, they needed to be
productized to facilitate a
product-led-growth motion.

Backend integrations were custom built. They needed to be productized to facilitate self-onboarding.

More Collaborative
Features

We weren't able to build the report
download feature. This would have been
the terminal action for users. And would
have closed the loop for users.

We weren't able to build the report download feature. This would have been the terminal action for users.