Enter how many stores, visits or claims you have and how sure you need to be. The calculator returns how many to audit, with and without the finite-population correction, and the interval for picking them systematically.
This is standard survey statistics. Your audit policy, or the client’s contract, decides the confidence level, margin of error and expected rate; the calculator only does the arithmetic. Once the sample is drawn, a photo-verified site audit tool such as AutoAudit can assign and verify each sampled visit.
The calculator uses Cochran’s formula for estimating a proportion, then applies the finite-population correction:
n₀ = z² × p × (1 − p) ÷ e²
n = n₀ ÷ (1 + (n₀ − 1) ÷ N)
Here z is 1.645 for 90%, 1.96 for 95% and 2.576 for 99% confidence; p and e are entered as fractions (5% = 0.05). Both results are rounded up to the next whole unit. The sampling interval is N ÷ n, rounded down: pick a random starting point between 1 and the interval, then take every k-th unit from an ordered list.
The correction matters for small populations. For a few hundred stores it cuts the sample sharply; for tens of thousands of visits, n and n₀ are almost the same.
An illustrative retail programme has 1,200 stores and wants 95% confidence, a ±5 point margin and no prior estimate of the non-compliance rate, so p = 0.5.
If last quarter’s audits suggest about 20% of stores fail, set p = 0.2 and the sample falls to 205 stores. A better estimate of p is the cheapest way to reduce audit effort without loosening the margin.
The sampled visits still have to happen properly and be reviewed. AutoAudit assigns each sampled store to an auditor, captures photo evidence in-app with location and time, and holds every report behind a sampled QC review before it reaches the client. For what each visit should check, see the retail store audit checklist; for scaling from a pilot to a network, read the multi-site audit rollout playbook.
Formula: W. G. Cochran, Sampling Techniques, 3rd edition, Wiley, 1977. Background on confidence intervals for a proportion: the NIST/SEMATECH e-Handbook of Statistical Methods. Illustrative; confirm the confidence level, margin and any regulatory sampling requirement with your audit policy, client contract or adviser.
It depends on how many stores you have, the confidence level and the margin of error your audit policy sets, and the failure rate you expect. At 95% confidence and a ±5 point margin with no prior estimate, about 385 is enough for a very large network; the finite-population correction brings it to about 278 for 1,000 stores.
A common one is Cochran’s: n₀ = z² × p × (1 − p) ÷ e², where z is the z-score for the confidence level, p the expected error rate and e the margin of error. For a known population N, the corrected size is n = n₀ ÷ (1 + (n₀ − 1) ÷ N).
Because p × (1 − p) is largest at 50%, it gives the biggest sample for a given confidence and margin. Use it when you have no reliable estimate from earlier audits; a real estimate usually reduces the sample.
It is the step used in systematic selection: population ÷ sample size, rounded down. Pick a random start between 1 and the interval, then audit every k-th unit on an ordered list of stores, visits or claims.
See it on your own data. AutoAudit — Proof of every field visit. Book a 30-minute working session with an engineer.
