Excel will not tell you when a number is wrong.
It recalculates, it never complains, and it looks exactly the same on the day it starts lying to you. The workbook that prices your jobs was built by someone who has left. This is the job that finds out what it is getting wrong.
| A | B | C | D | E | |
|---|---|---|---|---|---|
| 3 | Job | Client | Sale | Cost | Margin |
| 4 | J-1041 | Ashcroft | 4,200 | 2,860 | 1,340 |
| 5 | J-1042 | Weald & Co | 1,880 | 1,145 | 735 |
| 6 | J-1043 | Ashcroft | 6,400 | 4,010 | 2,390 |
| 7 | J-1044 | Halden | 2,750 | 1,600 | 1,4001 |
| 8 | J-1045 | Weald & Co | 3,120 | 2,090 | 1,030 |
| 9 | J-1046 | Pryor Ltd | 980 | #N/A2 | – |
| 10 | J-1047 | Ashcroft | 5,600 | 3,340 | 2,260 |
| 11 | J-1043 | Ashcroft | 6,4003 | 4,010 | 2,390 |
| 12 | J-1048 | Halden | 1,450 | 905 | 545 |
| 13 | Total | 9 jobs | 32,780 | 19,960 | 12,090 |
Worked example on a demonstration workbook · no client data appears on this site
Three faults, and only one of them looks like an error.
The broken lookup announces itself. The other two sit there looking like perfectly ordinary numbers, and those are the ones that cost money for months.
Someone typed over the formula. Every other row recalculates. This one never will. Measured
Visible, at least. But the total below it silently treats the gap as nothing, so the margin reads better than it is. Measured
J-1043 is on rows 6 and 11. Real revenue, counted twice, feeding every report built on this sheet. Measured
Every pound in the report carries a label saying where it came from.
It is easy to write "this is costing you £40,000 a year" and impossible to check. So every figure is tagged with how I got it. If a finding does not fit one of these three, it gets no number.
I found the error and did the arithmetic.
A fault in your actual file, with the difference computed. Open the workbook and check it yourself in a minute.
You told me the time it takes.
Your hours on a task, priced at what that time costs. It rests on your estimate, so it is labelled that way.
What one named failure would cost.
One named thing going wrong, costed. Never a prediction, never added to the measured total.
What the £300 actually buys.
Two to three hours on your file, then a written report you own. Always paid. A free audit is a sales call. You pay for this one, so it can tell you not to buy anything.
The real one, with the tab called "old" and the colour code only you understand. Do not tidy it first. The mess is the evidence.
Formulas traced, hardcoded overrides found, lookups tested, duplicates matched, and every place the workbook depends on one person remembering something.
Findings ranked by what they cost, each tagged Measured, Observed or Scenario, each with the cell reference so you can check my working.
Fix it yourself with the report in hand, or have me rebuild it. Commission within 60 days and the £300 comes off the price.
And if the file turns out to be sound, you still pay. The two to three hours happened either way, and you get the report either way. Put plainly: if I find nothing, you have saved £300 and you should not hire me.
This is the report itself. Read it before you buy one.
A worked example of the full £300 audit, run end to end on an invented joinery firm. Four workbooks, three findings that matter, and every figure carrying the cell reference so the working can be checked. Nothing is summarised or held back for a paid version, because there is no paid version of this document that looks any different.
- 4 workbooks reviewed
- 8 risks, each cell-referenced
- 3 findings that matter
Five pages. No client data appears in it, and it prints to A4 exactly as it appears on screen.
Updated: monthly, first working day
WHERE THE NUMBERS COME FROM Data tab: pasted from the accounts export
Calcs tab: do not type in the grey columns
IF SOMETHING LOOKS WRONG Check row count on Data vs Report first
Duplicates: filter column A for repeats
Every workbook I rebuild leaves with a README tab.
The thing that actually breaks a spreadsheet is not a formula. It is that the person who built it knew things they never wrote down, and then they left.
The audit report goes to you. This goes in the file, and stays there. The first tab explains the workbook in plain English: what it is for, where each number comes from, what never to type over, and what to check first when it looks wrong. Your next hire picks it up without ringing anybody.
Included as standard · not an extra line on the quote
Where I will tell you not to bother.
Three ways this ends without you paying me again. That is the point of charging for the audit.
- “The sheet is fine.”
- It happens. If the workbook is sound, the report says so and you have paid £300 to stop worrying about it. That is a legitimate outcome, not a failed job.
- “This should not be a spreadsheet.”
- Some things have outgrown Excel and need a proper database. I will say so, and the audit gives you the evidence to justify the spend to whoever signs it off.
- “Fix these three yourself.”
- If the fixes are simple, the report tells you exactly which cells to change and you keep your money. I would rather have the referral than the rebuild.
Where this usually leads.
Send me the spreadsheet you're least sure about.
Twenty minutes on a call first, so you know what you are buying. Then £300, fixed, and you find out what it has been getting wrong.
Book an intro call£300 fixed · credited in full against a rebuild within 60 days