Sales-Tax Audit Sampling: How Errors Become Projected Assessments

In a sales-tax audit, a sample tests a limited group of transactions and projects the measured error across a larger population. A small number of errors can therefore produce a large assessment. The population, sample design, error classification, and projection method deserve review before the result is accepted.

Why auditors use samples

Transaction records may be too numerous or complex for a detailed review of every invoice. Sampling can reduce the work while estimating the error for the full audit period. Some states authorize sampling when a complete audit would be unreasonable or impractical, records are inadequate, or sampling is expected to produce a reasonable result.

How projection works

The auditor defines a population, selects transactions or periods, determines the taxable error in the sample, calculates an error factor, and applies that factor to the population base. Methods vary. The result may be projected by dollars, units, time periods, accounts, or another measure.

For example, if the sampled transactions show a 2% taxable error and the relevant population is $5 million, a simple projection could produce $100,000 of additional taxable measure before tax, interest, and penalties. The actual method may be more complex.

Four decisions that drive the result

1. Population

The population must include the right entity, tax type, accounts, locations, dates, and transaction types. Duplicates, credits, non-audit periods, and unrelated entities can distort the projection.

2. Sampling unit

The unit might be an invoice, line item, customer, day, month, or dollar. The choice affects selection probabilities and how errors are weighted.

3. Error classification

Each sampled item should be evaluated under the correct law and supported by the available documentation. Missing certificates, sourcing, taxability, use-tax accruals, and rate errors are different issues and may need different treatment.

4. Projection

The error factor must be applied to a population base that corresponds to the sample. Credits and overpayments should be handled consistently. Changes in business operations, tax rates, products, locations, or systems may justify stratification or separate periods.

Questions to ask before agreeing

• What records and accounts define the population?
• Which transactions are excluded, and why?
• How will credits and negative transactions be treated?
• Is the sample statistical, block, stratified, or judgmental?
• How will missing documents that are later obtained be handled?
• Can the taxpayer reproduce the random selection?
• How will unusual one-time transactions be treated?
• Does the audit period include material business or system changes?

Review each sampled error

A projected adjustment is only as sound as the sample errors supporting it. Create an issue file for every proposed error with the invoice, contract, certificate, payment evidence, taxability analysis, auditor explanation, and taxpayer response. Correct classification mistakes before projection.

Preserve reproducibility

Keep the original population data, extraction instructions, field definitions, sample selection, replacement items, auditor schedules, and every revised version. Texas audit guidance, for example, states that information necessary to duplicate a computer-assisted sample must be retained.

When a sample may need adjustment

Potential problems include an incomplete population, nonrandom selection, inconsistent treatment of credits, a sample that spans materially different business periods, use of the wrong projection base, or extraordinary items that are not representative. The available remedy depends on the jurisdiction and audit stage.

Official reference points

• Texas Comptroller audit sampling guidance
• California audit publication

Sampling authority and procedures vary by jurisdiction. This article provides general information.