Abstract
ConnectRN is a clinical staffing company that uses a phone app to direct clinicians to shifts, while managing those shifts via a desktop webapp. Over the course of my time there, I completed several projects within the internal-only accounting tool.
I could not help but notice the generally poor state of the platform's UX, especially within the invoicing tools.
After a few conversations about the situation with our finance team and our COO, I was asked to address our clunky invoicing process.
Invoicing Archeology
I conducted multiple interviews with every member of our finance team, and documented the end-to-end invoicing process in Figjam.
Diagram of several key parts of invoicing. Red and yellow boxes indicate difficult or risky steps in the process.
A small section of the invoicing flow. I devoted a lot of time to defining terms.
I found multiple glaring issues in our Invoicing tool.
Invoice assembly was confusing.
Assembling an invoice meant comparing multiple different sets of spreadsheets outside the platform. Icon inconsistencies made button function unclear.
Two sets of icons, both used for adding / removing line items from an invoice in different tabs.
Invoices couldn't be saved.
There was no "invoice object" per se within our database, so there was no way to save or retrieve invoices in the accounting platform. Exported spreadsheets and PDFs were the only artifacts our accountant could use.
Invoice controls - note that there is no 'save' button. 'Load' here refers to loading items into an invoice, instead of loading a previous invoice.
Invoice editing was not trackable.
Since there was no saved invoice beyond whatever got exported, any edits in response to client disputes were done manually in Adobe Acrobat. We had no way to track these edits in aggregate - a major issue when those edits are generally to lower the dollar value of each invoice.
Two Solutions
With my research complete, I drafted two potential solutions for our COO to review.
1: Integrate with NetSuite
We could simply remove the invoicing tools, and send our data to NetSuite. Doing this would be a low development effort, and give us access to NetSuite's bevy of financial tools like automated invoicing and dispute tracking.
Our finance team was already using the software for multiple tasks, so adoption would be fairly seamless.
A feature roadmap shows the relative simplicity of developing and adopting the 'Integrate with Netsuite' solution.
2: Fix our Tools
We could improve the functionality of our existing invoicing tools. Doing this would let us keep more data inside our CRN database, but both I and the finance team felt this was less practical.
This path would involve a slow rollout of each feature, and many many workflow changes for accounting as we kept updating the Accounting platform.
A feature roadmap shows the relative difficulty of developing and adopting the 'Fix our Tools' solution.
Our C-Suite was very optimistic about long-term business growth, and so asked us to pursue the 'Fix our Tools' solution. While I did push back on this, the COO and CEO were very keen to invest in the CRN accounting platform. There were even talks of building something we could resell as its own product.
COMMUNICATION
Each prototype was presented at our weekly finance syncs for approval from our accountants, finance manager, and occasionally payroll or other team members. More finalized versions were then shown to design review and funding teams.
An upgraded invoice screen with improved design system usage and save/load functions.
A new feature - adding credits to invoice line items in response to client complaints.
Presenting these prototypes in design review and funding meetings was an interesting challenge. I did my best to reduce jargon and control the level of detail in discussion, to prevent lengthy meeting derailments.
Example slide from a funding meeting, explaining a new process for saving invoices.
Just in Case
A consistent point of feedback from the Finance team was the desire to "just use NetSuite". I agreed with them, but the C-Suite was adamant that the long-term play was the ideal route.
I decided to build out the "Integrate with NetSuite" solution in Figma, as it was a pretty small design lift. It seemed wise to have this second option ready.
An early draft of the "Integrate with Netsuite" solution.
The Budget Shrinks
As the invoicing designs entered funding and got scheduled for development, the larger business hit major issues. Changes in COVID regulations and the healthcare market meant that the business' runway was shrinking dramatically.
In response to this pressure, the C-Suite asked us to deliver invoicing improvements much more rapidly. At this point I was able to re-pitch the quicker "Integrate with Netsuite" solution, and this time we got approval to proceed immediately.
Rapid Implementation
I quickly consulted with the product and engineering teams to define the authority of each platform in the integration. This foundation would guide any future work.
It made sense for NetSuite to own all invoice details (makeup, value, credits), as the invoices were assembled there. However, ConnectRN would retain ownership of all timesheet details (length, pay rate).
This relationship would allow us to send a timesheet into NetSuite before it was even ready for billing. After my earlier discussions with finance, I knew this was useful for them to accurately calculate quarterly finances.
An ownership diagram lays out which platform will own which data points.
Engineering teams had been shuffled around due to multiple departures and a round of layoffs. Dan Lavoie, my new EM, was a fantastic help in getting the solution shipped within two weeks.
We focused on speed - a single button that would mark a wide selection of timesheets as "billed", and export them in a spreadsheet for use in NetSuite. After a brief test period, we were soon managing all invoices through NetSuite.
The final, 'bare bones' implementation of the NetSuite Connection
Future proofing
While supporting the engineering team, I also explored the further steps of this integration. Delivering the items to NetSuite was great, but we were still relying on an editable spreadsheet to carry information. That spreadsheet would be vulnerable to tampering or hacking - which was unacceptable given the financial information inside.
In addition, the transfer was one-way. As we added credits and refunds in NetSuite (or if a bug occurred), the CRN and NS databases would come out of alignment, confounding our internal auditors.
To resolve this, I planned for a direct integration that could send data to NetSuite without the additional export step.
An upgraded "Connection Wizard" that allows the user to rapidly sync the ConnectRN and NetSuite databases.
A direct integration would also give us a chance to check for database discrepancies. Brandon Vitabile, a business health specialist, had been raising this issue in our financial syncs. I designed a system to evaluate conflicts and allow for resolution.
A user reviews all conflicts between the two databases, with options on how to align them together.
The impact
The basic version of the "Integrate with Netsuite" solution immediately cut over a day of invoicing labor out of our accountant's schedule.
That represented about $25k yearly saved, with about a month of total design and engineering work. Our accountant was able to use that time for more important accounting tasks, and the invoicing process was now much more resilient to data drift.
This 90% increase in "time to invoice" speed was also well received by our clients, and generally increased our perception as a professional, well-organized partner.
FINAL THOUGHTS
As a designer, this project presented many interesting lessons about working within large organizations. How you pitch an idea can have enormous effects on the implementation schedule.
I am very glad that I had the foresight to prepare our backup plan, given that the first plan proved to be too large to execute.
However, in future I will involve my external stakeholders more directly in decision-making meetings - to prevent the sort of detour that occurred with this project.
