Spreadsheets are popular in finance for good reason. They let teams analyse data and build calculations quickly, and they cover gaps in system functionality when the business can't wait for central systems to catch up.
The problem is that spreadsheets grow. A file starts as a simple calculation, then gets copied, shared and built into a monthly or regulatory process. Before long, the spreadsheet is running the process instead of supporting it. At that point, it has become an end-user computing (EUC) tool, sitting outside the organisation's core technology and inside its control environment.
The spreadsheet error problem
People build and maintain spreadsheets, and people make mistakes. Research widely cited by spreadsheet specialists, including Raymond R. Panko of the University of Hawaii, found that around 88% of business spreadsheets contain errors. Even carefully built spreadsheets have been found to have errors in at least 1% of formula cells. [1]
Errors are only part of the risk, though. When a process depends on one person knowing how a set of spreadsheets fits together, the bigger danger is what happens when that person is no longer there.
An insurance market participant running on more than 100 spreadsheets
A UK firm submitting quarterly returns to its insurance market had grown many times over in twenty years, and its reporting process had grown with it. The return was produced by hand in Excel, which was used to source, check and process the data. As volumes and complexity increased, the process came to involve more than 100 spreadsheets. [2]
Only one person understood how it all fitted together, and that person was no longer available to run it.
Other problems followed. New business couldn't easily be added to the reporting flows, so extra spreadsheets were bolted on. Overrides, defaults and business rules were scattered across the files, which meant sign-off involved a long time spent working out and explaining the numbers. Eventually, the process was using up all the time available before the submission deadline. [2]
NextWave moved the process from Excel to Alteryx in phases, so benefits could be delivered and spreadsheets retired step by step. A NextWave engineer worked through the process in the order agreed with the client, understanding, documenting and rebuilding each part. At every stage, the logic was reviewed with subject matter experts (SMEs) to remove redundant steps and switch to better data sources. Each spreadsheet was only retired once its replacement had been signed off, and data validation rules were added at the start of each flow so problems were caught early. [2]
The results:
- Run time fell from weeks to minutes, and poor-quality source files could be rejected straight away.
- The business could keep growing without missing submission windows.
- New business could be added to the reporting cycle quickly.
- The time needed for analysis and sign-off dropped significantly. [2]
Speeding up the work with AI
The slowest part of a project like this is working out what each spreadsheet actually does. Formulas, overrides and business rules are spread across tabs and linked files, and someone has to trace them all.
NextWave's EUC elimination accelerator, delivered on Alteryx by NextWave's AI experts, is designed to speed up that step. Large language model (LLM) agents read Excel models, parse the formulas and embedded business rules, and turn them into standardised logic. That logic is translated automatically into Alteryx workflows, which are checked against the outputs of the original spreadsheets before being deployed to a governed platform.
Finance SMEs still confirm the rules and sign off the results, but it takes far less time to get there.
Building processes you can trust: the VURA approach
Replacing spreadsheets is only worthwhile if the business can trust what replaces them. That depends on two things.
First, the people who understand the data and the business best need to stay involved. Finance teams should help build and validate workflows, so the logic reflects how the business actually operates. Alteryx acts as the business logic layer that makes this possible.
Second, the processes themselves should be visible, understandable, repeatable and auditable. Together, these principles form VURA:
Visible
- Data lineage: Data can be traced automatically from inputs to outputs, so anyone can answer the question, "Where did this number come from?"
- Data control: Inputs and outputs are managed centrally, with need-to-know access, instead of sitting with individual users.
Understandable
- Clear business logic: Rules sit in a visual workflow that finance experts can follow and validate, instead of being hidden across tabs.
- Supportability: A support analyst or central Centre of Excellence (CoE) can resolve issues without needing the original creator.
Repeatable
- Consistent results: The same inputs and steps give the same results every time, without relying on one person.
- Fewer manual errors: Cut-and-paste mistakes and incorrect data movement are taken out of the process.
- Automated data quality checks: Issues are picked up automatically instead of by hand.
Auditable
- Audit trail: There's a record of every change to the process, who owns it, and what ran, when, and with which inputs and outputs.
- Change control: Nothing goes live without an agreed approval and testing process.
What happens when spreadsheets stop running the process?
Spreadsheets will always have a place in finance for analysis, modelling and one-off work. The issue is when a workbook, or in this case more than 100 of them, becomes the system behind a recurring regulatory process.
This firm went from a return that only one person knew how to produce to a process that runs in minutes, scales with the business and can be explained at sign-off.
NextWave helps finance teams find where EUCs are creating risk, decide which processes to automate first and build Alteryx solutions the business can maintain itself, using our AI-enabled accelerator to move faster. Get in touch to see how Alteryx One makes this possible and to explore our proven playbook for rapid success.
Sources
[1] Stuart Leung, Forbes, "Sorry, Your Spreadsheet Has Errors (Almost 90% Do)", 13 September 2014.
[2] NextWave Consulting, "Quarterly Monitoring Returns automation", September 2023. Client: UK insurance market part
November 9, 2026