Background
Prior to each month close, the accounting team had to perform a reconciliation process that reconciles invoices received from Dr. offices, clinics and hospitals that needed to be paid to obtain patients' medical charts to the actual payments made to those entities by check or credit card. This process was being done manually for many years by comparing rows and columns on a spreadsheet to downloaded bank statements and credit card statements.
​
If, in any given month, an invoice received that was paid could not be matched between the invoicing system and the accounting system, that payment would need to be written off. Due to data issues within the invoicing system, along with human error in spreadsheet comparisons, the average write off on chart payments was approximately $100,000 per year.
Thinking Through the Problem
The first thing I needed to do was to figure out the root cause of the issue(s) that led to the inability to reconcile invoices to payments. In order to do this, I needed to fully understand the invoice data from one system and the payment data located in a different system. This reconciliation process was very difficult because nothing directly tied invoice data to payments made. At the time I began the analysis I needed to understand how the accountants were actually reconciling these. So, the first thing I did was to conduct several interviews with each stakeholder.
-
Conducted three 1 hour interviews using screenshare to have each stakeholder walk though their process step-by-step
-
Documented each workflow visually in its' current state
-
Asked many questions about what would cause an invoice and payment to match and what wouldn't cause it to match
-
Discussed some edge cases that would cause an invoice and payment to never be able to be reconciled
Requirements Process
After discussing the issues with the stakeholders, it became clear a new solution was needed. Once I fully understood the current state, I then reached back out to each stakeholder and performed a requirements gathering session. When I began this process, I was told any solution was nearly impossible for a couple of reasons:
​
-
The process is just too complex
-
They could not think of anything that would assist them to make reconciliation easier, quicker and more accurate
​
Using my requirement gathering skills, I asked they forget about the current state, and asked "if we could start from scratch, what would this process need to accomplish?" Once I removed the built-in negative bias from their mindset, I was able to get actual requirements.
​
-
The system must have the ability to show invoices and chart payments
-
The user must be able to compare these and match where possible
-
The user must be able to have visibility where a match could not be made that could possibly be matched the next month
-
Write off's must decrease significantly
-
The process cannot take more than a week total
​
The Solution
My years of experience taught me that this problem was simply a data issue. What I needed to do was to have both the invoice data and payment data together. Then using the rule sets to reconcile, as stated by the accountants, most of the issues would be resolved.
​
Since one of the requirements was the ability to visualize what needed to be reconciled, I decided a new role-based application made sense. It would allow anyone who needed to do the reconciliation process to simply log into an application, and complete this work prior to closing out the month. The solution was comprised of the following:
​
-
Creating a new accounting database
-
Migrating invoice data into this new database every evening
-
Downloading both bank statement and credit card statement payment data into this new database
-
Creating interactive mockups for the new application
-
Writing stories that developers would code for to apply the rules on what causes a invoice to be matched to a payment
​
While development was in process, I regularly would engage the stakeholders on the progress being made. I would demonstrate how the application would work, explain the data, and would make sure to get their buy-in along the way.
​​
The Result
The complex, process-heavy and human error issues were resolved. The first month in Production, the accounting team was able to reconcile 95% of invoices and payments. 90% of these were the result of an auto-match feature that would simply auto match the transactions based on the rule sets that took about 3 seconds. The other 5% were reconciled manually by comparing what was left over. The un-reconciled transactions were the result of a payment made at the end of the month that didn't hit the bank/credit card statement in time prior to month end. These would be auto-matched the next month.
​
I was able to reduce the annual write-offs from $100k down to about $60 per month.