Benford’s Law says that in many real-world sets of numbers, the first digit is much more likely to be small. About 30% of numbers start with 1, while fewer than 5% start with 9. Auditors and fraud investigators compare financial records with this pattern. When invented numbers are mixed into real data, the pattern often breaks, which gives investigators a clue about where to look more closely.
Table of Contents
A Simple Puzzle
Imagine you collect the first digit of thousands of real numbers: the populations of towns, the lengths of rivers, the amounts on a company’s invoices. You might expect each digit from 1 to 9 to appear about equally, around 11% of the time each.
In practice, that is not what happens. The digit 1 appears far more often than any other, and each larger digit appears a little less often, with 9 being the rarest. This surprising pattern is called Benford’s Law, and it has become a useful tool in checking financial records.
What Benford’s Law Says
The law predicts how often each digit should appear as the first digit in a large set of naturally occurring numbers:
| First digit | Expected share of numbers |
|---|---|
| 1 | 30.1% |
| 2 | 17.6% |
| 3 | 12.5% |
| 4 | 9.7% |
| 5 | 7.9% |
| 6 | 6.7% |
| 7 | 5.8% |
| 8 | 5.1% |
| 9 | 4.6% |
In other words, a number is more than six times as likely to start with 1 as with 9. For readers who like math, the share for each digit d comes from the formula log10(1 + 1/d). For the digit 1, that gives log10(2), which is about 0.301, or 30.1%.
Who Discovered It
The pattern was first noticed in 1881 by the astronomer Simon Newcomb. In those days, scientists used printed books of logarithm tables for calculations. Newcomb saw that the early pages, covering numbers starting with 1, were much more worn and dirty than the later pages. People were simply looking up numbers starting with 1 more often.
Newcomb wrote a short note about this, but it was mostly forgotten. In 1938, a physicist named Frank Benford noticed the same thing independently. He went much further, collecting more than 20,000 numbers from very different sources, including river areas, city populations, scientific constants and numbers printed in newspapers. The same pattern appeared again and again, and the law took his name.
For decades it stayed a mathematical curiosity. In the 1990s, the accounting researcher Mark Nigrini showed how it could be used to spot suspicious figures in business and tax records, and it became a regular part of forensic accounting.
Why Real Numbers Behave This Way
The pattern seems strange until you think about how numbers grow.
Imagine a savings balance that grows by 10% every year, starting at 100. To go from 100 to 200, the balance must double, and at 10% a year that takes about seven years. During all that time, the balance starts with 1.
Now look at the other end. To go from 900 to 1,000, the balance only needs to grow by about 11%, which takes roughly one year. So the balance spends only about a year starting with 9.
Once the balance reaches 1,000, it starts with 1 again, and it will stay that way until it reaches 2,000, which takes another seven years or so.
Many things in the real world grow or vary in a similar way, by percentages rather than by fixed amounts: prices, populations, company sizes and transaction amounts. When a set of numbers covers a wide range, from tens to thousands or millions, this kind of growth naturally produces more numbers starting with small digits.
The pattern has another useful feature. It stays the same even if you change the unit of measurement. Convert a company’s sales from dollars to euros or rupees, and the first digits will still follow roughly the same shares.
How Auditors Use It
People who invent numbers rarely think about first digits. When someone makes up figures, they tend to spread digits fairly evenly, and they often avoid repeating patterns because those feel less random. The result is too few numbers starting with 1 and too many starting with larger digits.
Auditors and fraud examiners take a large set of figures, such as a year of invoices, expense claims or payments, and count how often each first digit appears. Then they compare the result with the Benford pattern.
If the data follows the pattern closely, nothing stands out. If certain digits appear much more or less often than expected, the auditor looks at those transactions in more detail.
Experienced analysts also use related tests:
- a first-two-digits test, which checks combinations from 10 to 99 and can reveal spikes at very specific amounts
- a last-digits test, which can show whether too many figures have been rounded to 0 or 5
- a duplicate test, which looks for the same amount appearing more often than it should
A Simple Example
Suppose a company requires a manager’s approval for any expense claim over 5,000. A dishonest employee who wants to avoid checks might submit several claims just under the limit, such as 4,850, 4,900 and 4,975.
On their own, these claims look normal. But when an auditor runs a first-digit test on all the company’s expense claims, a clear problem appears. The digit 4 shows up far more often than the expected 9.7%. A first-two-digits test would point even more directly at the 48 and 49 range.
The test does not prove anyone did something wrong. It simply tells the auditor which claims deserve a closer look.
When the Law Does Not Apply
Benford’s Law is powerful, but only for the right kind of data. It usually does not work well for:
- numbers with a narrow range, such as adult heights or exam scores out of 100
- assigned numbers, such as phone numbers, invoice numbers or postal codes
- prices set on purpose, such as items priced at 9.99
- data with fixed limits or minimums, such as a cash machine that only gives out amounts in steps of 20
- small data sets, where random variation can hide or create patterns
Using the test on the wrong kind of data can create false alarms, so analysts first check whether the numbers are suitable.
What Benford’s Law Cannot Do
It is important to understand the limits of this tool.
A mismatch with the pattern is not proof of fraud. There can be normal business reasons for it. A shop that sells only a few products at fixed prices, for example, will naturally have unusual first digits.
In the same way, data that fits the pattern is not proof that everything is honest. A careful fraudster could create figures that follow the pattern, and some types of fraud, such as recording a real payment to the wrong person, do not change the numbers at all.
This is why auditors treat Benford’s Law as a screening tool. It helps them decide where to spend their time among thousands of transactions. The real investigation still depends on documents, interviews and other evidence.
Key Points to Remember
- In many natural data sets, about 30% of numbers start with 1 and fewer than 5% start with 9.
- The pattern comes from the way numbers grow by percentages across a wide range.
- Invented numbers often break the pattern, because people spread digits too evenly.
- Auditors use it to find unusual groups of transactions, not to prove fraud.
- It only works on suitable data: large sets of natural amounts that cover a wide range.
A pattern first spotted on the worn pages of a scientist’s book of tables now helps auditors check millions of transactions. It is a good reminder that numbers carry more information than their face value, and that the digits themselves can tell a story.