Free audit tool · IA QuickTools
Defensible samples in seconds. Reproducible forever.
mySampler selects audit samples the way a reviewer wishes you had: seeded, so the exact selection can be reperformed by anyone, any time — with random, systematic and monetary-unit methods, statistically grounded sample sizes, and a selection memo that documents itself. No signup. No install. Nothing you paste ever leaves your machine.
- 🎲 Seeded & reproducible
- 🔒 100% in-browser
- Random · systematic · MUS
- Exact-binomial sample sizes
- Free forever
Works in every modern browser · happiest on a desktop
mySampler · yours instantly — nothing to set up
mySampler
The seed — why it matters
“How did you select these?” Answered. Permanently.
Excel’s RAND() gives you a different sample every time you open the file — which means your workpaper documents a selection nobody can ever recreate. mySampler runs on a seeded, deterministic generator: the seed string is hashed into the random stream, so the same seed with the same inputs produces the same selection on any machine, on any day. The seed prints on every export.
Reviewers reperform instead of taking your word. QA re-runs the selection years later. And because pasted populations are fingerprinted (SHA-256), your memo can also prove which population the sample came from. That is what “defensible” means.
What’s inside
Every selection method fieldwork actually uses.
Sample from a population
Paste a list — invoice numbers, JE IDs, employees — or give a numeric range, and draw n at random, without replacement by default. Duplicate detection warns you before the sample does.
Every k-th, seeded start
Interval k = N÷n with a random start drawn from the seed. The memo records both — and warns you about the one real weakness, cyclically ordered populations.
Monetary unit sampling
Paste amounts and select monetary units — bigger items proportionally likelier, and anything at or above the sampling interval pulled as a 100% top-stratum certainty, the standard treatment. Credits excluded with a warning, as they should be.
Exact-binomial attribute sizes
Confidence, tolerable rate, expected rate → the required n and allowable deviations, computed from the exact binomial — the statistics behind the AICPA tables — then carried into the sampler with one click.
A selection record that documents itself
Every run produces method, parameters, seed, timestamp and population fingerprint — exported at the top of the CSV, or printed as a memo with prepared-by / reviewed-by sign-off lines.
Plus the classic RNG
Uniform integers and decimals, normal distributions, custom weighted values — the original myRNG lives on as a tab, now seeded and reproducible like everything else.
Questions auditors actually ask
Straight answers.
If it’s seeded, is it still really random?
Yes — in the way that matters for audit. The selection is unpredictable before you run it (you choose or generate an arbitrary seed) and perfectly reproducible after. Statistically the output passes as uniform random; operationally a reviewer can reperform it. Unseeded randomness — Excel’s RAND(), dice — is actually the inferior choice for audit work, because it can never be verified.
Can this site see the population I paste?
No — verifiably. Everything stays in your browser; there is no server behind the tool and no account. Watch the Network tab in your browser’s developer tools while you paste and sample: no request carries your data. One honest caveat: on a company-managed device, your organization’s own monitoring can see anything you do — the tool guarantees it adds zero exposure of its own.
Do the sample sizes match the AICPA tables?
They come from the same place — the exact binomial distribution — so they match the published attribute-sampling tables, occasionally differing by a row where the tables applied rounding conventions. The methodology inside the tool states the computation precisely, which is more than most tables do.
How do I sample from an Excel column?
Copy the column, paste it into the population box — one item per line is exactly what Excel gives you. For monetary unit sampling, copy two columns (ID and amount) and paste; tabs and commas both parse.
Does this replace IDEA, ACL or my GRC sampling module?
For selection and sizing — the everyday 90% — it does the same job with less friction and better documentation. Full analytics platforms still earn their keep on evaluation of results, stratified estimation and big-data work. Use this to select defensibly; use whatever your firm runs to do the rest.
What happens if I clear my browser cache?
Saved inputs and history are erased — nothing is ever stored anywhere else. But here’s the elegant part: your CSV export carries the seed and parameters, so any selection remains reproducible forever from the export alone. The workpaper is self-sufficient.
Go deeper: the math behind your sample
Wondering why the attribute helper suggests the numbers it does? Our guide Audit Sample Sizes Demystified: Where 25, 40, and 60 Actually Come From derives the standard conventions from the underlying binomial math, shows exactly when they mislead, and gives you a defensible parameter grid for non-SOX audits. And to see a seeded, reperformable selection documented inside a real testing file, read our annotated model workpaper — the sampling section is built around the memo this tool exports.
The toolkit
Your sample came from a plan. Build that here too.
RCM Workbench
The planning end of the audit: risk register, control matrix, DE/OE test plans, issues log and team hours — with a live residual-risk model. Build the test plan there; select its samples here.
Open the Workbench → Log the daymyDayLog
Five-second logging of what the fieldwork actually took — entries roll up by engagement into timesheet-ready CSVs. Installs to your phone’s home screen like an app. Same guarantee: your log never leaves your device.
Open myDayLog → All toolsIA QuickTools
Every free tool on this site — built to the same standard: single-file, in-browser, private by architecture, and free forever. More are being rebuilt to this standard now.
Browse the toolkit →mySampler v1 — July 2026. Successor to myRNG: everything the old tool did, now seeded, documented and workpaper-ready.