Low-Cost Audit Heuristic

principle

A cheap quantitative screen can expose suspicious patterns before expert review begins. Its job is to concentrate costly attention, not deliver a verdict.

A business-fraud investigation can begin with work simple enough for a seventh-standard student: count the numbers and plot them on a graph. The surprise is not that experts become unnecessary, but that expertise need not be the first resource spent.

E1

Spend attention after the pattern appears

The heuristic separates detection from diagnosis. First, reduce a large body of data to a cheap count or visual pattern. Then compare what appears with what would normally be expected. Only the largest departures graduate to closer investigation. This is epistemic-triage expressed quantitatively: inexpensive computation decides where scarce human judgment may have the highest value.

That ordering matters. Beginning with exhaustive review spends expert time evenly, including on ordinary cases. Beginning with a crude screen narrows the field before the expensive work starts, avoiding the demand for total understanding that produces analysis-paralysis.

E1

Where it shows up

The schoolchild’s graph

Counting values and plotting them does not solve the fraud case. It creates a visible shortlist of patterns worth questioning—a low-skill first pass feeding a high-skill second pass.

E1

A flag is not a finding

A deviation can identify where to look, but it cannot establish why the pattern occurred. Treating the count as a verdict turns a useful screen into the McNamara Fallacy: what is measurable displaces context, explanation, and expert judgment.

Build the cheapest shortlist first

Before commissioning a full review, choose one count or graph that can rank cases by unusualness. Send the highest-deviation cases to an expert together with the underlying records, and require diagnosis rather than automatic condemnation.

Episodes that teach this