Data quality & internal tools

Numbers your team can close the month on.

I audit your internal tool against your own source files, fix what doesn’t match, and add controls so bad data can’t get back in.

coverage check · your workbook vs the appIllustration
ProgrammeYour workbookThe appStatus
A · Jan12,480.0012,480.00 Match
B · Jan3,215.603,215.60 Match
C · Year15,901.0018,900.00 Traced
Cause: an amount counted in January but left out of the workbook’s annual total. The app is right; the workbook is flagged.
D · Feb7,030.257,030.26 Rounding
4 lines · 3 match · 1 explained Nothing left unexplained
Illustration. Made-up figures, showing how every difference is traced to a cause.

The engagement behind the numbers

A payroll allocation platform, built to replace Excel.

An international human rights NGO, through Relief Applications. I built the app to replace the client’s Excel workbooks, then proved it against them.

1/222The only material gap in the coverage totals, and it was an error in the client’s own workbook, not the app.
219/222coverage totals matched to the cent
87%of employees’ 2025 payroll to the cent
8people paid but missing from the roster, found and flagged
~7%payroll inflation prevented by stopping a re-import
The full log (13 more) +
  • about 200false budget overruns avoided by catching a stale source file before deploy.
  • 3,138 linesand 1,342 grid columns checked read-only in production after a rounding fix. No discrepancies.
  • 17defects found and fixed by checking the app end to end against the client’s own workbooks.
  • to the centthe December 2025 allocation matrix, reproduced by the app after the fixes.
  • 178 of 178lines where hours, euros and cost agree with each other and convert back exactly.
  • a testthe coverage check now runs as an automated test in the codebase, not a one-off script, so a regression fails the build.
  • hoursdiffer from the client’s workbook on purpose: the app rounds each line down against the allocated cost, the workbook rounds to the nearest hour against the full employer cost. Under the workbook’s formula about 98% of 1,177 hour cells would match, and the rest trace to specifics in the workbook itself.
  • all 30 staffchecked, not a sample, when the app’s monthly hours came out an hour away from the client’s workbook. The cause (rounding order and cost base) was found and fixed.
  • 5 pull requestsin one day, for five bugs reported by client staff, each checked against live data before it was called done.
  • 6 layersof the app traced to find the contradictory instructions that had stalled the client’s month-close, and unblock it.
  • controlsthat reported “compliant” on empty months now fail. Closed months and deposited payroll can’t be overwritten by later imports, and manual overrides survive a re-import.
  • featuresrequested by client staff: split allocations, Excel paste into grids, multi-month uploads, part-time working patterns. Each checked against production data.
  • handoveronce the client was running the tool on their own, with client updates written in French.

How a check works

Your file is the reference, not the app.

  1. Your source filesThe workbook or export the business already trusts.
  2. What the tool saysThe same numbers, as the app calculates them.
  3. Every difference, listedEach traced to a defect, a data gap or an unwritten rule.
  4. Fixed, or explainedNothing is closed as “probably fine”.

What I do

Find the difference, then close it.

Nobody trusts the numbers

Audit against your files

Every difference listed, with its cause.

The app mostly works

Fix and extend it

Python, SQL/PostgreSQL and React, plus what its users ask for.

The same file imported twice

Controls and guards

Checks that fail on empty data; closed months stay closed.

Checks that always pass

Verify before release

Every change checked against your live data before it goes out.

How it runs

Scoped first, priced after a call.

First step, free

A 20-minute call, or one file sent over.

2 to 6 weeks, fixed price

An audit first, then fixes and controls, quoted after a call.

With your team, documented

Changes reviewed and tested where your team can see them, then a written handover.

Also available for longer remote contracts, full- or part-time. For recruiters →

Send me one file.

The report nobody trusts, or the export behind it. I’ll tell you what I’d check first. Free, and no call needed.

Not comfortable sending data to someone you haven’t met? The column headers and two lines describing the problem are usually enough.