Cost assurance: looking at a sample of invoices doesn’t work

Background: The traditional approach

In any major programme delivered under cost reimbursable or target cost arrangements, assurance over the appropriateness of defined costs applied for by contractors is important. Traditionally the approach to cost assurance (or substantiation) consists of a sample of typically 10% to 20% of costs being agreed to invoices.

This approach doesn’t work as it provides only very limited assurance. Here are the reasons why:

An invoiced cost may be charged more than once

An invoice tested under a 10% sample approach may be charged more than once. The duplicate invoice(s) may appear in the remaining 90% of costs not tested.

Unusual groups of transactions may not be identified

On its own, an invoice included in a sample may be properly supported and pass the test of substantiation. When looked at alongside other transactions however it may look unusual.  For example, ten invoices for the same amount over a short time period would potentially raise questions.

The full implication of small amounts overcharged on individual transactions may be missed

On a sample basis there may be very small amounts overcharged which when viewed as a single transaction have little impact. For example, an “administration charge” of few pennies on a payroll transaction may be ignored.  When looked at as part of costs as a whole the impact of adding all the small transactions together may however be material.

The results of sample testing cannot be extrapolated

Under the traditional approach, samples are rarely selected on a statistical basis and therefore, the result of testing cannot be extrapolated to the rest of costs. In order to use sampling effectively, it would be necessary to stratify the population of costs and to create sample sizes based on expected errors and confidence levels.

Testing individual accounting journals provides no assurance

Accounting journals are a means of moving costs around and introducing and taking out costs using the principles of double entry bookkeeping. Often, traditional approaches ignore accounting journals.  Even if they are included in the sample, however looking at the supporting documentation for individual journals will provide little assurance.

Many contractors use reversing journals to account for accruals. These types of journal have no impact on cost and testing them adds no value.

The impact on costs of accounting journals can only be properly understood by looking at the net effect of all of them.

Testing a sample of invoices for overstatement ignores the understatement of credits

The traditional approach to selecting a sample of costs is an overstatement test.  It tests to see if costs are overstated by the inclusion of inappropriate amounts.  Costs can also however be overstated by understating credits, for example by suppressing credit notes.

Some transactions or invoices are riskier than others

A traditional sampling approach ignores the fact that some invoices or transactions may be riskier than others. Any accounting provisions or estimates introduced into costs are inherently riskier than systematically processed transactions for, for example payroll costs.  Greater assurance could be achieved by testing 100% of the accounting provisions and 1% of the payroll costs.

Sample testing of individual transactions ignores the substance of transactions

Looking at individual transactions may only tell part of the story. An invoice may be properly supported, however, there may be rebates or discounts applied for early payment or quantities which are accounted for separately.


Don’t place too much reliance on the traditional approach to cost assurance.  There is an alternative approach which is a cost effective, best practice and risk based.