PERCENTRANK in Google Sheets: Documented Exact-Value and Precision Rules

• Introduced: first introduction date is not established by the reviewed official history; no desktop release number is inferred.
• Availability: documented in the rolling Google Sheets service; this guide uses the web worksheet workflow.
• Checked: October 14, 2026; official PERCENTRANK reference. Documentation and independent calculations, with no native Sheets execution or wider client certification.

=PERCENTRANK(data,value,[significant_digits])

The data and target value are required. The optional precision argument defaults to 3. Google describes it as significant figures and says a supplied number less than or equal to zero is ignored in favor of that default.

These are documentation-based rules, not native test results. The page does not give a full precision-rounding algorithm or invalid-type error matrix. The numerical examples deliberately use ranks representable in at most three decimal places, so they do not depend on an unverified rounding-versus-truncation decision.

Keep this identity separate from PERCENTRANK.INC, whose own reference describes estimation for an unobserved value. Do not silently substitute one for the other during workbook migration.

Enter 10,20,40,80,160 in A2:A6. The independent distinct-value rank model uses (position-1)/(n-1) after sorting:

These outputs are independently calculated under the stated ordinal model. The compact reference names percentage rank but does not print its equation; native endpoint and tie controls have not been executed here.

Format the result as Percentage to display 0.75 as 75%. Multiplying by 100 and also applying Percentage format would display 7500%, so choose one presentation method rather than both.

=PERCENTRANK(A2:A6,30)

Google’s documented expectation is #N/A, because 30 is not in any cell of the data range. It lies between 20 and 40, but the dedicated page still states a membership requirement.

A value outside the observed range, such as 200, is absent too and falls under the same documented rule. These are explicitly source-based expectations; if a current native workbook differs, preserve that observation and reassess the documentation rather than concealing the discrepancy.

PERCENTRANK.INC’s interior estimation model gives 0.375 for 30 in this fixture. That comparison explains why blindly renaming a function can change a report’s treatment of unobserved targets.

For the observed target 80:

=PERCENTRANK(A2:A6,80)
=PERCENTRANK(A2:A6,80,0)
=PERCENTRANK(A2:A6,80,-2)

All three have independently expected rank 0.75 under Google’s documented fallback and the distinct-value model. This does not establish that an arbitrary text precision, fractional precision or very large precision is valid.

Because 0.75 is exactly representable at the default decimal detail, the comparison tests the fallback without claiming an unverified precision-shortening algorithm.

The ratio 80/160 is 0.5, while 80’s modeled percentage rank is 0.75. The first calculation compares magnitudes; the second compares ordered positions. They answer different questions.

Append 320 in A7. The target 80 is still fourth, but there are now six observations, so its modeled rank becomes (4-1)/(6-1)=0.6. A rank can change even when the target value does not, because the comparison group changed.

Positive unit conversion preserves the ordering: the values 25,45,85,165,325, obtained with 2*x+5, give the transformed target 165 the same five-value rank 0.75.

All numerical controls use distinct, complete numeric observations. Duplicate-value ranking, singleton data, empty cells, numeric-looking text, source errors and native precision boundaries remain unverified. An error is not automatically a rank of zero; zero already has a meaningful position in the model.

PERCENTILE asks the reverse kind of question by returning a value for a chosen percentile. Its documented interpolation does not override PERCENTRANK’s separately documented missing-value rule.

• Google PERCENTRANK reference.
• Editorial testing standards.

Examples use fictional data, English function names and comma argument separators. Expected results are independent calculations or explicitly documented rules, not native Sheets captures. Check spreadsheet locale when reproducing punctuation-sensitive formulas.