Your Finance Systems Run on One Person's Memory
It starts with one broken report. A staff accountant fixes it, and suddenly they are the report person. A year later they administer the ERP, maintain the reporting server, own the SharePoint site and its workflows, and babysit a virtual machine where a scheduled automation runs macros that move payroll data between systems. There is no backup. The only documentation is what they wrote themselves, scattered across a few folders and their inbox. Their title still says accounts payable.
Almost every mid-sized company has a version of this person. They are usually the most valuable employee in the department and the single largest operational risk in it. Nobody planned it that way. It accumulated one helpful fix at a time.
Why do accountants end up running IT systems?
Because finance owns the outcome. When a report is wrong the night before close, or a manual reconciliation eats two days every month, waiting for a ticket in the central IT queue is not an option. The person closest to the problem solves it with the tools at hand: a report designer, a low-code automation platform, a macro, a scheduled task. It works, so the next request comes to them too.
This is not unusual or reckless. It is how most technology outside the core platform gets built. Gartner has reported that 41 percent of employees acquired, modified, or created technology outside of IT's visibility in 2022, and predicted that figure would reach 75 percent by 2027. Finance is one of the departments where that business-built technology carries the most weight, because it touches payroll, payables, and the numbers leadership reports on.
What happens when the only person who knows is out?
Most discussions of knowledge loss focus on resignations. In finance, the more frequent failure is mundane: the owner is on vacation during close, out sick on payroll day, or simply in a meeting when the automation throws an error. Someone else has to figure out which server the report runs on, what the macro expects as input, which credentials the scheduled job uses, and why the export file needs to be renamed before the bank will accept it.
The answers usually do exist somewhere. They are in a setup note in a OneDrive folder, a vendor support thread in Outlook, a Teams message from the night the automation was first configured, a comment in the script itself, and a change request filed with IT two years ago. But a backup who did not write any of it does not know where to look or what to search for. Deloitte has noted that complex spreadsheet macros are often understood only by the people who built or regularly run them. The same is true of everything built around them.
What does key-person risk cost a finance team?
The obvious cost is a missed deadline: a late payroll file, a delayed close, a report that leadership waits on. The less obvious cost is time. In a survey of treasury professionals at more than 100 large US and European companies, treasury software provider Kyriba found teams were losing roughly 4,800 hours a year to spreadsheet work, and called out the risk of a single person owning the master file as an operational exposure on its own.
There is also an audit cost. Controls around financial services and financial reporting assume that processes are documented, repeatable, and reviewable by someone other than the person who performs them. When the only explanation of how payroll data moves is in one employee's memory, every audit request becomes an interview, and every finding about segregation of duties is harder to answer. This is why compliance and audit preparation so often becomes a scramble to reconstruct how a process actually works.
How do you reduce dependence on one person for finance systems?
The standard advice is to document everything and cross-train a backup. Both are right, and both tend to stall. The owner is too busy running the systems to write a manual, and the backup never has a quiet week to learn. What usually exists instead is a trail of partial documentation spread across every tool the owner has touched.
That trail is more useful than it looks if it can be searched as a whole. Retrieval-augmented generation, or RAG, indexes content where it already lives, retrieves the passages that match the meaning of a question rather than its exact wording, and produces an answer that cites its sources. A backup can ask "what do I do if the payroll export fails" and get an answer assembled from the setup note, the vendor thread, and the old change request, with a link to each. The owner still matters. They just stop being the only index.
How does RetrieveIT help?
RetrieveIT connects to the places this knowledge already sits, including SharePoint, OneDrive, Confluence, Outlook, Microsoft Teams, Google Drive, and Jira, and makes them searchable together from one box. Semantic search matches plain-language questions to the right document even when the wording is different. Every answer cites the source system, document, author, and date, so a backup can verify a procedure before running it.
Permissions from each source are enforced at query time, so a finance workspace can include the runbooks and vendor threads the team needs without exposing payroll data to people who should not see it. New content syncs within minutes, so the note the owner writes today is searchable tomorrow. It also gives IT teams a way to understand business-built systems before they formally take them over, instead of starting from a blank page.
The accountant who became the systems owner did the company a favor. The way to repay it is to make sure that what they know can be found when they are not in the room.
Stop depending on one person's memory
Connect SharePoint, OneDrive, Outlook, and Teams to RetrieveIT and give your backup one place to find how things work, free for 14 days. No credit card required.
Get Started Free