RANK.AVG in Collabora Office: Average Tied Ranks and Complete Ranking Checks
Calculate Collabora RANK.AVG with tied scores, both order directions, rank-sum controls and a changed comparison cohort.
Verification status: See the evidence and test scope described in this guide. How formula examples are checked
Quick answer: RANK.AVG assigns tied values the average of the positions they occupy. Its default type 0 ranks in descending order; type 1 ranks in ascending order. Fractional ranks are expected when an even-sized tie group spans consecutive positions.
Availability and version history
- Introduced: original Collabora release and launch date are not established by the reviewed sources.
- Availability: currently documented in Collabora Office 26.04 desktop help.
- Platform scope: online and mobile deployments are not inferred from this desktop-help release.
- Checked: October 14, 2026. Function-specific documentation and independent calculations; no native Collabora execution.
Syntax and duplicate rule
=RANK.AVG(value;data;[type])
The data argument is an array or range. The selected value must occur in it; the source says an absent value produces an error without naming its exact code. These examples supply numeric data and type 0 or 1 only.
For ties, RANK.EQ uses the lower numerical rank, while RANK.AVG averages the occupied positions. This is a difference in ranking policy, not a rounding setting.
Example 1: a five-row ranking with two ties
Enter 90,70,90,50,70 in A2:A6. Put this formula in B2 and fill through B6:
=RANK.AVG(A2;$A$2:$A$6)
Expected results, preserving source-row order:
| Value | Descending average rank |
|---|---|
| 90 | 1.5 |
| 70 | 3.5 |
| 90 | 1.5 |
| 50 | 5 |
| 70 | 3.5 |
The two 90s occupy positions 1 and 2, whose average is 1.5. The two 70s occupy positions 3 and 4, whose average is 3.5. The 50 occupies position 5.
A rank of 1.5 is meaningful. Formatting it as an integer can hide the tie treatment, so show enough decimal places to preserve the result.
Example 2: ascending average ranks
Use an explicit type 1:
=RANK.AVG(A2;$A$2:$A$6;1)
Filled through the same rows, expected results are 4.5,2.5,4.5,1,2.5. The 50 takes position 1, the 70s average positions 2 and 3, and the 90s average positions 4 and 5.
For each observation, its ascending and descending average ranks sum to 6. This provides a check for this five-observation numeric dataset without changing the source order.
Advanced check: the sum of all average ranks
The descending results sum to 15, equal to 1+2+3+4+5. The ascending results also sum to 15. Averaging the occupied ranks within each tie group preserves their total contribution.
For four identical numeric values 8,8,8,8, every descending or ascending average rank is 2.5, the average of positions 1 through 4. Their total remains 10.
These checks include each original observation once. Summing only one row per distinct value would answer a different question and would not preserve the same rank total.
Advanced example: add a new highest observation
Add 100 in A7 and expand the comparison range to A2:A7. The original five descending ranks become 2.5,4.5,2.5,6,4.5. The new 100 has rank 1, and all six ranks sum to 21.
The original ascending ranks remain 4.5,2.5,4.5,1,2.5; the added 100 gets ascending rank 6. The change follows the expanded cohort, not a change in any old value.
Keep comparison ranges absolute when filling formulas, and disclose whether a report ranks within a subgroup or across the full dataset.
Boundaries and format notes
For a requested value 80 absent from the original list, the documented outcome is an error; its exact native code is unverified. Range-text, Boolean, empty-string and source-error behavior are not imported from the separate RANK page.
The source identifies COM.MICROSOFT.RANK.AVG outside the cited OpenFormula 1.3 standard. That note establishes neither its first Collabora release nor universal workbook interoperability. RANK remains a separately documented function with a less explicit tie description.
References and verification
Complete function-specific documentation reviewed. Worked outputs are independent mathematical or explicitly documented rule controls, not application screenshots or native test results. Unresolved ordering, type and error boundaries retain their stated scope.