What running the business on spreadsheets actually costs: a worked calculation.

The short answer

Spreadsheets are free. Running a business on them is not. The cost is paid in payroll — hours spent re-keying, reconciling, and repairing — and in the occasional error that reaches a customer. Neither line appears on any invoice, which is why this bill is almost never added up.

This page adds it up. The model below is deliberately simple, every assumption is stated so you can change it, and the two worked examples (a 12-person services firm and a 60-person distributor) use conservative inputs. If your numbers are smaller, good — you probably don’t need to change anything yet. If they’re larger, you now know what the fix is worth.

Only two external figures are used: the U.S. Bureau of Labor Statistics’ benefits-to-wages ratio and published spreadsheet-error audit rates. Both are cited in Sources. Everything else is a labeled assumption.

The model

Spreadsheet debt has two parts. The first is predictable and large; the second is unpredictable and occasionally enormous.

Part 1 · Labor

Annual labor cost = people × hours per week of manual work × loaded hourly cost × 48 weeks.

• People: everyone who re-keys, reconciles, or builds reports by hand. Include the owner — owner hours are the most expensive in the building and the most often left out.
• Hours per week: measure it. Ask each person to keep a tally for one week of time spent moving data between a spreadsheet and anything else, or fixing something that was wrong. People underestimate; a one-week tally is more honest than a guess.
• Loaded hourly cost: hourly wage ÷ 0.70. The BLS March 2026 Employer Costs for Employee Compensation release puts benefits at 30.1% of total private-industry compensation (wages 69.9%), so a $25/hour wage costs the employer about $35.75/hour. That excludes recruiting, equipment, and management time, so it is a floor.
• 48 weeks: assumes four weeks of leave and holidays.

Part 2 · Errors

Annual error cost = errors that reach the outside world per year × average cost per error.

The audit literature gives a base rate: in Dr. Raymond Panko’s review of field audits (EuSpRIG, 2000), rigorous audits found errors in at least 86% of operational spreadsheets, with cell error rates of about 1% to 2.5% where measured. Those are error rates for the spreadsheet, not for what escapes it; most errors are caught or are harmless. So the model does not apply the cell error rate to your revenue — that would be dramatic and wrong. Instead it asks a plainer question: how many times last year did a wrong number reach a quote, an invoice, an order, or a payroll run, and what did each one cost to fix, credit, re-ship, or apologize for? Most owners can answer that from memory. Count it, price it, and label it an estimate.

Worked example A · 12-person services firm

A commercial cleaning, landscaping, or specialty-trades company: one owner, two office staff, nine in the field. Quotes in a spreadsheet, schedule in another, invoicing in accounting software, customer list in a third place. Assumptions:

Errors (assumed): six escaped errors a year — two mis-quoted jobs at ~$800 each in margin given away, three invoicing errors at ~$150 each in time and credits, one missed scheduled job at ~$1,200 in re-work and goodwill. Total: $3,250.

Spreadsheet debt, example A: about $48,000 a year, of which $44,700 is labor. Cut the manual hours by half with a connected quoting/scheduling/invoicing setup and the saving is roughly $22,000 a year — against tools in this class that cost $2,500–$4,500 a year at list for a team this size. If the owner alone gets five hours a week back, that is the single largest line.

Worked example B · 60-person distributor

A regional wholesale distributor: sales team, customer service, warehouse, purchasing, and accounting, running an aging ERP surrounded by spreadsheets for pricing, backorders, commissions, and the weekly management report. Assumptions:

Errors (assumed): twelve escaped errors a year — four mis-priced orders at ~$2,000 each in margin, four backorder or stock-out misses at ~$1,500 each in expedite fees and lost lines, two commission disputes at ~$1,000 each in reconciliation time, two shipping errors at ~$2,500 each in freight and credits. Total: $21,000.

Spreadsheet debt, example B: about $279,000 a year. That is the cost of the gap around the ERP, not the ERP itself. It is also, not coincidentally, the size of budget people are afraid to ask for when they propose fixing it — even though the fix is often integration and reporting work, not a new ERP. What a replacement ERP actually costs belongs next to this number, not instead of it.

Do this for your own business

1. List the handoffs. Every place a number leaves one system and is typed into another by a person: quote → job, order → invoice, timesheet → payroll, inventory count → purchase order, everything → the weekly report.
2. Tally for one week. Each person keeps a running count of minutes spent on those handoffs and on fixing wrong numbers. Don’t estimate; tally.
3. Load the wage. Hourly wage ÷ 0.70. For salaried staff, salary ÷ 2,080 ÷ 0.70. For the owner, use what you’d pay someone to do your job.
4. Multiply by 48. That is Part 1.
5. Count last year’s escaped errors. The ones you remember are the expensive ones. Price each honestly. That is Part 2.
6. Write the total down and date it. Now every proposed fix — a shared sheet with an owner, a $50/month tool, a CRM, an integration, an ERP — can be judged against a number instead of a feeling.

If you want a faster first read, the free Systems Sprawl Diagnostic takes three minutes and scores the pattern; the signs you’ve outgrown spreadsheets describes what the debt looks like from the inside.

The three objections, answered

This calculation gets the same three pushbacks whenever it is done inside a company. They are worth answering before the number is written down, because each one is partly right.

• “Those hours aren’t really lost — people are paid anyway.” True, and irrelevant to the decision. The question is not whether the payroll is spent but what it buys. Six hours a week of re-keying is six hours not spent on collections, quoting, or customers. If there is genuinely nothing better for those hours to do, the business has a different problem than spreadsheets.
• “Our spreadsheets don’t have errors.” The audit literature says otherwise, and it says it about spreadsheets built by professionals who were sure of the same thing. Panko’s point is that spreadsheet error rates match human error rates on any complex task — a few percent of cells — and that nobody, including the author, reliably spots their own. The escaped errors you remember are the visible tip; the ones absorbed silently into a quote or a margin are the rest.
• “The new system will just create different manual work.” Often true, and the strongest of the three. A badly chosen or badly implemented system replaces re-keying with workarounds. That is why the calculation should be redone after any change, with the same one-week tally, and why the fix should be judged on the delta, not on the demo.

A note on the owner’s hours

In example A, the owner’s five hours a week are 46% of the labor total. That proportion holds in most owner-led firms under about 25 people: the owner is the integration layer, the person who knows which spreadsheet is current and fixes the number before it goes out. Two consequences follow. First, the loaded rate to use for the owner is not their draw but the cost of the work they are not doing — usually sales, hiring, or simply being replaceable. Second, spreadsheet debt is the reason many owners cannot take a week off; the systems only reconcile when they are in the building. That is a cost the model cannot price, and it is frequently the one that finally motivates the fix.

What to fix first

The tally will show that manual work is not evenly spread. One or two handoffs usually account for most of the hours — commonly quote-to-job, order-to-invoice, or everything-to-the-weekly-report. Fix those first, and fix them at the cheapest level that removes the re-keying:

1. One sheet, one owner. If three people maintain three versions of the customer list, the cheapest fix is one shared list with one person allowed to edit it. Cost: zero. This alone often removes a third of the reconciliation hours.
2. A purpose-built tool for the single worst handoff. Quoting, scheduling, or invoicing software in the $50–$300/month range usually removes the biggest line in a small firm’s tally, and can be run month-to-month while you confirm it does.
3. Integration before replacement. In example B, most of the debt is in the gaps around the ERP. Connecting pricing, backorders, and reporting to the system that already holds the data is usually a fraction of the cost of replacing the system, and it is reversible.
4. Replacement, when the tally says so. When the connected-and-patched version still leaves six-figure debt on the table, a new core system is justified by arithmetic, not by a sales cycle. Take that arithmetic into the vendor conversation.

What the number does not tell you

Three cautions, because a calculation this simple can be misused.

• It is a floor. It leaves out the decisions made on wrong numbers, the hires delayed because nobody could see the workload, the sales missed because the quote took three days. Those are real and larger; they are also unmeasurable, so they’re left out on purpose.
• Recovered hours are not automatically recovered dollars. Cutting six hours of re-keying a week doesn’t cut a paycheck. It only pays off if those hours go somewhere useful — more quotes, faster collections, an owner who can leave at five. Have that plan before buying the fix.
• The cure can cost more than the disease. A $40,000 platform to remove $8,000 of debt is a bad trade, and it happens constantly because the $8,000 was never calculated. Do this math first; it is the cheapest thing on this page.

Where this fits in an assessment

This calculation is the first thing a Business Systems Assessment quantifies. We map where the manual work is, time it with the people doing it, price it at loaded cost, and set it against the cost of each fix — so the roadmap is ordered by return. The assessment’s pricing is published; the arithmetic on this page is free to use whether or not you ever call us.

Sources

• U.S. Bureau of Labor Statistics, Employer Costs for Employee Compensation — March 2026 (released June 12, 2026), bls.gov/news.release/ecec.nr0.htm — official; accessed 2026-08-27. Private industry: wages and salaries 69.9% of compensation ($32.60/hr), benefits 30.1% ($14.01/hr). Used for the ÷ 0.70 loading factor.
• Panko, R. R., “Spreadsheet Errors: What We Know. What We Think We Can Do.” Proceedings of the European Spreadsheet Risks Interest Group (EuSpRIG), Greenwich, July 2000; arxiv.org/abs/0802.3457 — academic; accessed 2026-08-27. Rigorous field audits found errors in at least 86% of spreadsheets audited; measured cell error rates of 1.1%, 2.2%, and 2.5%.
• All wage, hours, and error-count figures in the worked examples are labeled assumptions chosen to be conservative for the business described; replace them with your own tally.