Case 23
Numbers Sanity Check
Scenario
You are about to send a cost-saving estimate to the CFO. It was put together quickly by a colleague and combines several spreadsheets.
Why this framework
Uses full PINPOINT because the consequence is higher: adds a professional stance, examples, self-checks, your own review before use, and a plan to tweak and test.
Example context
Summary (cost-saving estimate for moving invoice processing to a shared service centre): Annual staff cost saving £312,000; annual software cost increase £48,000; annual error rework saving £27,500; one-off migration cost £90,000; net first-year saving £201,500; payback period 5.4 months. Source spreadsheet 1 (staff costs): Current team of 11 FTE in finance operations, total loaded cost £486,000. Proposed team of 6 FTE, total loaded cost £198,000. Source spreadsheet 2 (software and migration): Current licence £22,000 a year. Proposed platform £71,000 a year. Migration one-off: £62,000 consultancy plus £28,000 internal time. Source spreadsheet 3 (errors): 1,840 invoice errors a year, average rework 0.75 hours at £24 an hour. Expected reduction 60%.
Prompt
Purpose: Check that the cost-saving estimate is sound before I send it to the CFO.
Context: Use the attached summary and the three source spreadsheets. The source spreadsheets are the final word over the summary.
Narrow the ask: Recalculate each figure in the summary from the sources. Give a table: figure, summary value, your value, match or mismatch, and likely cause of any mismatch. Then list assumptions the estimate depends on. Stop and ask if a figure cannot be traced to a source.
Persona and role: Act as a meticulous financial analyst. Assume nothing is right until it is traced.
Example of good: "Annual saving: summary 420k, recalculated 380k; the summary counts January twice."
Check your work: For each material claim, label it as [Verified Source], [Inference], [Assumption], or [Unverified]. Cite the source where available. If an unverified or assumed point could materially change the conclusion, flag it for review rather than guessing. Show every calculation step.
Before I use it: I will check the mismatches myself and fix the summary before sending.
Materials (pasted here in place of the attachments):
Summary (cost-saving estimate for moving invoice processing to a shared service centre): Annual staff cost saving £312,000; annual software cost increase £48,000; annual error rework saving £27,500; one-off migration cost £90,000; net first-year saving £201,500; payback period 5.4 months.
Source spreadsheet 1 (staff costs): Current team of 11 FTE in finance operations, total loaded cost £486,000. Proposed team of 6 FTE, total loaded cost £198,000.
Source spreadsheet 2 (software and migration): Current licence £22,000 a year. Proposed platform £71,000 a year. Migration one-off: £62,000 consultancy plus £28,000 internal time.
Source spreadsheet 3 (errors): 1,840 invoice errors a year, average rework 0.75 hours at £24 an hour. Expected reduction 60%.Example output from ChatGPT
Cost-Saving Estimate Validation Not ready for CFO submission
[Verified Source] The summary overstates the first-year saving by £32,628 (19.3% relative to the recalculated value). Four of the six reported figures are incorrect. The migration cost is correct, while the payback period uses an inappropriate calculation basis and incorrect inputs.
Corrected first-year saving £168,872
£32,628 below summary
Annual recurring saving £258,872
Before migration costs
Corrected simple payback 4.2 mo
Assuming even monthly savings
- Reconciliation against source spreadsheets
All amounts in GBP. The three pasted source extracts are treated as the authoritative data.
| Figure | Summary value | Recalculated value | Result | Likely cause |
|---|---|---|---|---|
| Annual staff saving | £312,000 | £288,000 | Mismatch | [Inference] £24,000 overstatement; source of difference unknown |
| Annual software increase | £48,000 | £49,000 | Mismatch | [Inference] £1,000 omission or input error |
| Annual error rework saving | £27,500 | £19,872 | Mismatch | [Inference] Different error assumptions or arithmetic error |
| One-off migration cost | £90,000 | £90,000 | Match | None |
| Net first-year saving | £201,500 | £168,872 | Mismatch | Incorrect underlying savings and costs |
| Payback period | 5.4 months | 4.2 months | Mismatch | [Inference] Summary appears to use first-year net saving rather than recurring saving |
Traceability: All six figures can be recalculated from the supplied sources, so there is no missing source figure requiring an immediate stop. However, the precise origin of the three incorrect component values and the summary's payback methodology are unverified and require review.
- Full calculation trail
A. Staff savings - Source 1
[Verified Source]
Current cost = 11 FTE, £486,000
Proposed cost = 6 FTE, £198,000
Annual saving = £486,000 − £198,000 = £288,000
Variance = £312,000 − £288,000 = £24,000 overstatement
B. Software increase - Source 2
[Verified Source]
Proposed platform = £71,000
Current licence = £22,000
Annual increase = £71,000 − £22,000 = £49,000
Variance = £49,000 − £48,000 = £1,000 understatement
C. Error rework savings - Source 3
[Verified Source for inputs; Assumption for the forecast reduction]
Annual errors = 1,840
Rework hours = 1,840 × 0.75 = 1,380 hours
Current rework cost = 1,380 × £24 = £33,120
Expected reduction = 60%
Annual saving = £33,120 × 60% = £19,872
Variance = £27,500 − £19,872 = £7,628 overstatement
D. Migration cost - Source 2
[Verified Source]
Consultancy = £62,000
Internal time = £28,000
Total = £62,000 + £28,000 = £90,000
E. Net savings - Sources 1–3
[Verified Source for arithmetic]
Staff saving = £288,000
Less software increase = −£49,000
Plus error saving = +£19,872
Annual recurring saving = £288,000 − £49,000 + £19,872 = £258,872
First-year net saving = £258,872 − £90,000 = £168,872
Summary variance = £201,500 − £168,872 = £32,628
F. Simple payback - Sources 1–3
[Verified Source for arithmetic; Assumption for timing]
Annual recurring saving = £258,872
Monthly recurring saving = £258,872 ÷ 12 = £21,572.67
Payback = £90,000 ÷ £21,572.67 = 4.17 months, rounded to 4.2 months
[Inference] The summary's 5.4 months appears to come from £90,000 ÷ £201,500 × 12 = 5.36 months. That uses the net first-year saving, which already deducts the migration cost, instead of the recurring saving.
- Assumptions requiring validation
| Assumption | Status | CFO review requirement |
|---|---|---|
| All five eliminated FTE costs are genuinely avoidable | [Assumption] | Confirm actual reductions, redeployment and timing |
| Proposed £198,000 staff cost includes all shared-service charges | [Unverified] | Confirm no additional allocation or overhead costs |
| 60% error reduction is achievable | [Assumption] | Obtain pilot evidence or operational validation |
| Every avoided rework hour translates into financial saving | [Assumption] | Confirm whether savings are cash-releasing or productivity gains |
| Software fees remain constant for the year | [Assumption] | Confirm contract terms and implementation dates |
| No further migration, training or parallel-running expenses | [Unverified] | Validate the complete transition budget |
| Full savings begin immediately and accrue evenly | [Assumption] | Obtain implementation milestones and monthly cash flows |
- Recommended CFO position
[Verified Source] The arithmetic supports £168,872 in first-year net savings and £258,872 in annual recurring savings.
[Inference] The business case remains financially attractive on the stated assumptions, but the 4.2-month payback is only an illustrative simple payback, not a validated cash-flow forecast.
Hold the summary until Finance confirms the avoidability of staffing costs, validates the 60% error reduction, and supplies the transition timetable.
Correction to the opening: five of the six summary figures mismatch, not four. Only the one-off migration cost matches.
