Audit samples a reviewer approves before drawing
How Achekam proposes an audit sample from a risk-scored population, who approves the plan before the draw, and how any draw is reproduced from its seed.
By Achekam · · 4 min read · For Accountants & audit firms in South Africa, Nigeria, Kenya
The reviewer's question is always the same: "How were these items chosen?" On too many files the honest answer is that a senior pasted the population into a spreadsheet, added a random-number column, sorted it and took the top of the list. Then the spreadsheet recalculated, the random numbers changed, and the selection on the working paper can no longer be shown to have come from anywhere.
ISA 530 expects a sample to be selected so that each sampling unit has a chance of selection, and the file to show how it was done. This article explains how audit sampling works in Achekam: the population, the methods, the approval that comes before the draw, and the stored seed that lets anyone reproduce it later.
Where sampling lives
Sampling sits in Audit Intelligence, under Analyze. That is worth saying first, because the menu can mislead. The Sampling, Findings, Audit Procedures, Adjustments and Permanent Files entries in the audit menu do not open screens of their own; they open the working-paper list, its reviewer view or the Central Audit Data Hub. If a new team member clicks "Sampling" and finds a list of working papers, send them to Audit Intelligence instead.
Load and score the population
The team loads the population it wants to test: a debtors listing, a fixed-asset register, the purchases for the year. Achekam risk-scores the loaded population and proposes a sample from it.
Before relying on any of that, do what ISA 530 and ISA 500 expect of every population: establish that it is complete and accurate for the purpose. Both standards appear in full in the IAASB's 2026 Handbook, alongside every other International Standard on Auditing. Agree the listing's total to the trial balance or the lead schedule, and note on the working paper how you did it. A perfectly reproducible sample drawn from an incomplete population is still the wrong sample.
The methods on offer
Achekam offers five ways of choosing items:
- Random selection, where each item has an equal chance.
- Systematic selection, taking items at a fixed interval from a random start.
- Monetary unit sampling, where the chance of selection rises with the item's value, which suits tests for overstatement.
- Specific items, where the team chooses items deliberately, for example every balance over a threshold or every related-party item.
- All items, testing the whole population.
Strictly, the last two are not sampling under ISA 530: selecting all items or specific items is a way of selecting items for testing, and the results are not projected to the rest of the population. Many good audit approaches combine them: test the large and unusual items specifically, then sample the remainder.
No size before materiality
No sample size is proposed without approved materiality and a confidence level. If the engagement partner has not yet approved materiality, the team will see no proposed size, which is the right order: sample sizes depend on tolerable misstatement, and tolerable misstatement depends on performance materiality. The approval of materiality is described in setting and approving materiality on an audit.
A second person approves before the draw
The sampling plan is approved by a reviewer who did not prepare it, and only then is the sample drawn. The preparer cannot approve their own plan.
This changes the conversation on review. Instead of the manager challenging a selection after the testing is done, when changing it means redoing work, the manager challenges the plan: is the population right, is the method right, is the confidence level appropriate for the assessed risk? Once approved, the draw happens and the testing starts on a selection both people have agreed.
Reproducing any draw from its seed
Every draw can be reproduced from its stored seed. A seed is the starting value for the random selection: run the same method on the same population with the same seed and you get the same items, every time.
That solves the spreadsheet problem in the opening paragraph. Anyone reviewing the file, from the manager to the engagement quality reviewer, can have the draw reproduced from the seed and see that the items on the working paper are the items it produced. It also protects the team: nobody can be suspected of quietly swapping a hard-to-test item for an easier one.
Evaluation stays with the team
Achekam's AI engine refuses any task whose type is giving an opinion, reaching a conclusion, approving or signing. The evaluation of sample results, deciding whether a deviation is a control failure, projecting misstatements and concluding whether the population is fairly stated, is the engagement team's work, recorded on the working paper.
A sampling checklist
- Reconcile the population to the trial balance or lead schedule before loading it.
- Confirm materiality is approved; without it, no size is proposed.
- Choose the method and the confidence level, and write down why.
- Separate items you will test specifically from the remainder you will sample.
- Send the plan to a reviewer who did not prepare it, and draw only after approval.
- Note the seed on the working paper alongside the selection.
- Evaluate the results and conclude on the working paper.
Unusual journals are usually selected by criteria rather than by statistical sampling; that is covered in journal entry testing on a whole population.
Try it on one area
On your next audit, run one population through Audit Intelligence from loading to approved draw, and ask the reviewer to reproduce the selection from the seed before testing starts. If that works on one area, there is little reason to keep drawing samples in spreadsheets on the others.
General guidance only, not advice on your own circumstances. Tax and payroll rules change: check the current position with the authority or your adviser.
