Spreadsheets vs Accounting Software: When a Seller Actually Outgrows Excel

A seller outgrows a spreadsheet at the point where three things become true at once: more than one person needs to touch the file, inventory cost layers have to be tracked rather than estimated, and a mistake made in March would not be caught until the books are closed in December. Order volume is the wrong trigger. Plenty of sellers run 500 orders a month in Excel without difficulty. The trigger is whether an error can survive undetected long enough to reach a tax return.

That framing matters because the usual advice, migrate once you hit some revenue number, gets the causation backwards.

What spreadsheets are genuinely better at

Start here, because the case for migration gets oversold.

A spreadsheet answers a question nobody anticipated, in about four minutes, with no vendor roadmap involved. Accounting software answers the questions its designers thought of. For scenario modeling, one-off analysis, negotiating with a supplier over landed cost, or testing what happens to margin if freight rises 12 percent, a spreadsheet is faster and better and will remain so.

Spreadsheets also cost nothing, run offline, are portable to any successor, and cannot be discontinued. That last one is not theoretical. When the bookkeeping provider Bench abruptly shut down on December 27, 2024, customers lost access to their accounting and tax documents at the start of tax season, as TechCrunch reported at the time. An acquisition by Employer.com was announced three days later, on December 30, 2024, and TechCrunch put the affected customer count at roughly 12,000. Bankruptcy filings in Canada the following month showed about $2.8 million in cash against $65.4 million in liabilities for a company that had raised $113 million. A spreadsheet on your own drive does not do that to you. Anyone weighing a platform migration should read the ConnectBooks write-up of what that episode meant for sellers caught in it, because the exit plan matters as much as the feature list.

What spreadsheets are measurably bad at

The error research is unusually well documented, and it is worse than most people expect.

Raymond Panko of the University of Hawaii has catalogued spreadsheet error studies for three decades. Across the five field audits he rates as having the strongest methodology, 91 percent of the 55 spreadsheets examined contained important errors. Across a broader set of nine studies publishing their methodology, 84 percent of 163 spreadsheets contained errors. Panko’s own assessment is that these figures probably understate the problem.

Two findings matter more than the headline. First, people are bad at finding the errors: across nine experiments with more than a thousand participants, inspectors caught roughly 60 percent of known errors, and no participant found all of them. Second, people do not know they are making errors. In one of Panko’s studies, developers estimated a median 10 percent chance that they had made a mistake. In fact 86 percent of them had.

The consequences scale with the stakes. The European Spreadsheet Risks Interest Group maintains a public archive of documented cases. One entry from January 2024 records Norway’s sovereign wealth fund losing roughly 980 million kroner on a benchmark calculation error caused by an incorrectly entered date. Another records Marks and Spencer correcting a quarterly trading statement the same day it was issued: the 7am version reported group sales up 1.3 percent, and the corrected version hours later showed sales had fallen 0.4 percent.

The three conditions that actually force the switch

Condition 1: more than one person edits the file

A spreadsheet maintained by one person has one mental model. Add a second editor and you add a second set of assumptions about what column H means, whether row 400 is included in the total, and whether last month’s tab was ever updated. Version control in spreadsheets is social, not technical, and social controls fail quietly.

Condition 2: inventory cost has to be tracked rather than estimated

This is the real dividing line for product sellers. A spreadsheet can hold an average cost per SKU. It cannot easily hold cost layers, meaning the actual price paid for each specific batch of units still sitting in the warehouse.

The distinction has tax consequences. The IRS lists specific identification, FIFO, and LIFO as the methods for identifying inventory cost in Publication 538, and requires that inventory practices stay consistent from year to year. A spreadsheet that quietly shifts from one approximation to another between January and September is not applying a consistent method, and nobody will notice until an examiner asks how closing inventory was derived.

Condition 3: an error can survive until year end

Software catches certain mistakes structurally. Double entry bookkeeping does not let a transaction post to one side only. Bank reconciliation forces every deposit to match something. Negative inventory quantities surface as warnings. None of these are intelligence, they are just constraints, and constraints are what spreadsheets lack.

If a seller closes their books monthly and reconciles every account, an Excel error gets caught within 30 days and the exposure is small. If the books get assembled once a year in March for the prior year, a formula error introduced in month two has eleven months to compound.

What migration actually costs

Three things, and only one of them is the subscription.

The first is the opening balance. Software inherits whatever inventory value, liability balance, and retained earnings figure you hand it. Migrating with a wrong opening inventory number means every subsequent period is wrong in the same direction, permanently.

The second is learning the abstraction. A spreadsheet shows you every number. Accounting software shows you reports built from numbers you cannot see without knowing where to click. Sellers frequently report feeling less in control for the first quarter after switching, which is accurate rather than irrational.

The third is vendor risk, which is the Bench lesson. Ask any prospective platform how data comes out, in what format, and whether historical periods travel with it.

The honest recommendation

Keep both. This is not a compromise position, it is what most competent finance operations already do.

Software owns the ledger, because the ledger needs constraints, an audit trail, and the ability for two people to work in it without destroying each other’s work. Spreadsheets own analysis, because analysis needs flexibility that a ledger cannot provide. The failure mode is not using Excel. The failure mode is using Excel as the ledger.

If you are trying to decide right now, run one test. Take last month’s closing inventory value from your spreadsheet and recalculate it from purchase invoices without looking at the original formula. If the two numbers agree, you have more runway than you think. If they do not, you found out cheaply, and the Small Business Administration’s guidance on managing business finances is a reasonable place to start on what a real set of books requires.