> For the complete documentation index, see [llms.txt](https://docs.talentmap.com/knowledge-base/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.talentmap.com/knowledge-base/step-5-analyze-interpret-and-report/compass-reports/reporting-faq/how-compass-scores-are-calculated-and-rounded.md).
# How Compass scores are calculated and rounded
### What you might notice
When you read a Compass report you might see one of the following:
* A stacked bar adds up to 99% or 101% instead of 100%
* A small year-over-year change shows as 0
* The same dimension shows a slightly different number in different views
None of these mean the data is wrong. They are display-rounding effects. This article explains the rounding model, walks through worked examples, and shows how to access the full-precision data when you need to defend a number to council, a board, or an executive audience.
### Two layers of precision in Compass
Compass works at two layers:
**Stored data.** Each item score is held at high decimal precision, up to 13 decimal places. All calculations, averages, and period-over-period comparisons run on these full-precision values.
**Displayed numbers.** Reports show whole-number percentages so they read cleanly in a deck or dashboard. Rounding to whole numbers makes the numbers scannable but hides differences smaller than 1%.
Most rounding questions come from reading two displayed numbers without seeing the stored data underneath.
### The 5-point to 3-point conversion
Most Compass surveys use a 5-point Likert scale, from Strongly Disagree to Strongly Agree. Reports collapse the five response points into three favourability categories:
* **Favourable** = Strongly Agree + Agree
* **Neutral** = Neither Agree nor Disagree
* **Unfavourable** = Disagree + Strongly Disagree
Each of the three percentages is rounded independently to a whole number for display. That is why the three displayed numbers can sum to 99, 100, or 101.
### Three effects that look like errors but are not
#### 1. A stacked bar sums to 99% or 101%
Imagine a question with these stored values:
* Favorable: 67.4%
* Neutral: 16.3%
* Unfavourable: 16.3%
Stored total: 100.0%. Displayed values: 67 + 16 + 16 = **99%**.
Now imagine the same question with slightly different decimals:
* Favorable: 67.5%
* Neutral: 16.3%
* Unfavourable: 16.2%
Stored total: 100.0%. Displayed values: 68 + 16 + 16 = **100%**.
The 99 / 100 / 101 difference is whether each component rounded up or down. The stored total is always 100%.
#### 2. A small year-over-year change shows as 0
If a dimension moves from a stored 71.4 in one cycle to a stored 71.6 in the next, the stored change is +0.2 points. Depending on rounding direction, both displayed values can read 71, and the chart shows no change. The +0.2 is real, but it is too small to surface at whole-number precision.
The reverse can also happen. A 71.4 → 71.5 shift displays as 71 → 72, which looks like a +1 shift even though the real movement is only +0.1.
There is a less obvious version. Your overall favourability index is built from the Favorable, Neutral, and Unfavourable categories. An internal shift (for example, respondents moving from Neutral to Unfavourable while Favourable stays the same) can leave the headline index roughly flat even though the underlying response mix has shifted. The headline number is honest, but it does not tell the whole story on its own.
#### 3. The same dimension shows a slightly different number in different views
The same dimension might read 89 on the dashboard and 88.5 in an Excel export. Both are correct. The dashboard rounds to a whole number; the export shows one decimal place. They are the same data point at two different display precisions.
### How to defend a number with full-precision data
If you are presenting to a council, board, or executive audience and you expect questions about exact figures, you can pull a **full-decimal Excel export** from Compass. This export shows scores to several decimal places so you can walk through the exact differences underneath the rounded values.
In Compass, go to Reports → Snapshot → Export → Excel and click in the desired cell. \
This is the fastest way to confirm whether a small displayed difference is a real shift or a rounding artifact, and the cleanest way to defend a Compass number to a skeptical audience.
### Quick reference
| What you see | Why | What to do |
| ----------------------------------------------- | ------------------------------------------------- | ---------------------------------------------------------- |
| Stacked bar adds to 99% or 101% | Each segment rounded independently | Confirm with the full-decimal Excel export if needed |
| Small change shows as 0 between cycles | Underlying change rounds to the same whole number | Pull period-over-period from the full-decimal Excel export |
| Different views show slightly different numbers | Different display precisions on the same data | Compare both at the same precision |
---
# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.
## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.
Perform an HTTP GET request on the current page URL with the `ask` query parameter, and the optional `goal` query parameter:
```
GET https://docs.talentmap.com/knowledge-base/step-5-analyze-interpret-and-report/compass-reports/reporting-faq/how-compass-scores-are-calculated-and-rounded.md?ask=&goal=
```
`ask` is the immediate question: it should be specific, self-contained, and written in natural language.
`goal` is optional and describes the broader end goal you are ultimately trying to accomplish on behalf of the user. GitBook uses it to tailor the answer towards what is most useful for that goal.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.
Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.