An Excel waterfall built for 3 investors works fine. The formulas are readable, the allocations tie out, and QA takes minutes. At 10 or more investors, the same model starts producing problems that are hard to see and harder to fix.
Rounding Errors Scale With Investor Count
Pro-rata allocation across 3 investors produces rounding differences measured in fractions of a cent. At 15 investors, those fractions add up. If your model rounds each allocation independently using standard rounding, the sum of individual allocations can differ from the total distribution amount. That difference might be $0.02 or $0.15, but it means the model doesn't tie out.
You find the discrepancy during QA. Now you have to manually adjust cells to force it to reconcile. You don't document how or why. Next quarter, a new team member opens the file, doesn't see your notes, and starts fresh. The model is now inconsistent from distribution to distribution.
Formula Complexity Grows Nonlinearly
A 3-investor waterfall model might have 50 formulas. A 15-investor model built the same way has hundreds, many of them nested references to other cells. One misplaced absolute reference ($B$5 vs B$5) and the model silently calculates wrong numbers for every investor added after that point.
No error message. No warning. The spreadsheet balances. Distributions are paid. But the allocations to other investors are all calculated against the wrong tier, or the wrong capital position. These errors go undetected for quarters until reconciliation reveals the inconsistency.
QA Time Increases Faster Than You Expect
Checking a 3-investor waterfall takes 15 minutes. You verify each investor's capital position, trace how it was allocated, and confirm the tier structure that applied. Checking 15 investors against the same waterfall structure takes much longer because you're verifying each investor's capital position, tier allocation, and pro-rata share independently.
Most GPs report that QA time grows faster than linearly with investor count. A simple model that took 15 minutes for 3 investors takes 1.5 hours for 10 and 4+ hours for 20. At some point, QA becomes so expensive that you skip it. That's when errors enter the system.
As investor count increases, QA time grows faster than linearly. By 20 investors, the process becomes a days-long slog.
The Spreadsheet Doesn't Announce When It's Broken
The spreadsheet doesn't announce when it's outgrown its structure. It just starts producing numbers that are slightly off, and you discover it when something doesn't tie out. Or worse, you don't discover it until an audit or an LP calls with a question.
By then, you've paid distributions based on a model that silently made errors. You can't go back and fix all historical distributions. You can't confidently defend the calculations. You're explaining rounding logic and formula mistakes instead of running the fund.
Why Excel Still Feels Fine
The waterfall still calculates. Numbers flow through. Investors get paid. Everything feels functional until the day QA takes 5 hours instead of 15 minutes, or an auditor asks about the allocation logic, or an LP questions their distribution and you can't explain it quickly.
At that point, the spreadsheet isn't broken, it's just revealed that it always was. It never scaled past the simple cases it was built for.