# Specify Results Calculation

## Overview

The results calculation determines whether overall scoring uses each metric’s performance score alone, or also incorporates stability (consistency over time) when computing that metric’s contribution to the overall score. As with [prioritizing metrics](https://docs.thousandeyes.com/product-documentation/internet-insights/provider-intelligence/build-a-query/prioritize-metrics), this selection directly affects providers’ overall scores (see [Overall Scoring](https://docs.thousandeyes.com/product-documentation/internet-insights/provider-intelligence/scoring-methodologies/overall-scoring) for more information).

## Calculation Options

* **Performance mode**: Uses each metric’s performance score only for overall scoring. That performance score is calculated from average metric values relative to peer providers in the same city (and same destinations) over your selected time frame.
* **Performance & Stability mode**: For overall scoring, each metric combines that metric’s peer-relative performance score with its stability score using a calculation that is weighted toward the lower of the two; a provider must score well on *both* dimensions to receive a high combined score. Stability scores reflect how consistent each provider’s own values are over time relative to their median value, and are independent of how peers behave.

For more information about how each mode is calculated, see [Scoring Methodologies](https://docs.thousandeyes.com/product-documentation/internet-insights/provider-intelligence/scoring-methodologies).

### Performance Mode

**Performance** mode answers the question: *"Which provider has the best overall raw metric values?"*

* Uses only performance scores to derive the overall score (stability scores still appear in results but do not affect overall scoring in this mode).
* Raw measurements feed daily aggregates over the chosen time frame; the performance score itself is a peer-relative index, not a simple average in isolation.
* If latency is your priority metric, a provider with 15 ms latency will likely score higher than a provider with 20 ms latency. Because performance is peer-relative, the same raw latency can map to different scores across locations if peer performance shifts there.
* Doesn't consider whether those values are consistent over time for overall scoring.

Use **Performance** mode when:

* You care primarily about raw speed and responsiveness.
* Short-term variations are acceptable for your use case.
* You want to identify the strongest peer-relative performer on a particular metric.

**Example**: A financial services company needs to select an internet provider for their corporate office where employees use cloud-based trading platforms requiring high-speed access. If their priority is to maximize average browsing and application speed to reduce latency in trades, they would choose **Performance** mode. This mode prioritizes the provider with the best average speed, accepting some variability, since faster average performance can provide a competitive advantage despite occasional fluctuations.

### Performance & Stability Mode

**Performance & Stability** mode answers the question: *"Which provider delivers strong performance and keeps it consistent?*"

* For overall scoring, each metric incorporates both the peer-relative performance score and the absolute stability score.
* Stability scores are derived from each provider’s own values over the time frame, using the median as reference, which reduces the influence of outliers compared with looking only at simple daily means.
* Per metric, performance and stability scores are combined using a harmonic mean of the two scores (see [Performance & Stability Scoring](https://docs.thousandeyes.com/product-documentation/internet-insights/provider-intelligence/scoring-methodologies/performance-and-stability-scoring) for more information about the harmonic mean). That mean penalizes imbalance: excellent performance with poor stability (or the reverse) yields a low combined contribution, so strong overall scoring requires both inputs to be strong.
* With latency as your priority metric, a provider with 15 ms latency but high variability might score lower than a provider with 18 ms latency that is very consistent.
* The combined **Performance & Stability** values are not shown as a single column in the results table; they feed the overall score calculation.

Use **Performance & Stability** mode when:

* Consistency is important for your applications (for example, voice/video calls, real-time collaboration).
* You're evaluating for mission-critical applications.
* You want to minimize the risk of performance variability.

**Example**: If you're evaluating providers for a branch office running video conferencing and VoIP, select **Performance & Stability** mode so overall scoring rewards both good performance and consistent quality without favoring jitter spikes hidden behind a good average.

## Key Considerations

* **Performance scores are always visible alongside stability scores**: Each metric shows two scores, the peer-relative performance score and the absolute stability score. Even if you select **Performance** mode as your results calculation, you still see stability scores in your results, they just don't affect the overall score.
* **The combined scores are not displayed as their own column**; you infer the effect through changes in overall score when you switch modes.
* **Scores might change**: The harmonic mean used in **Performance & Stability** mode ensures a high combined value only when both performance and stability are high; one very high input cannot fully compensate for a very low one. As such, switching between the two modes can significantly change provider overall scores. See below.

## How Scores Can Change

When you switch between **Performance** and **Performance & Stability** modes, you could see significant changes in overall scores.

**Example** (illustrative numbers; same peer set and time frame):

| Provider       | Performance score | Stability score | Overall score (Performance mode) | Overall score (Performance & Stability mode) |
| -------------- | ----------------- | --------------- | -------------------------------- | -------------------------------------------- |
| **Provider 1** | **90**            | **90**          | **88**                           | **92**                                       |
| **Provider 2** | **95**            | **50**          | **90**                           | **78**                                       |
| **Provider 3** | **85**            | **85**          | **84**                           | **86**                                       |

Interpretation:

* In **Performance** mode, Provider 2 can have the highest overall score because overall scoring ignores stability; the high performance score (95) dominates even though stability is weak (50).
* In **Performance & Stability** mode, Provider 1’s overall score can rise above Provider 2’s because the harmonic mean penalizes Provider 2’s instability. If you sort the table by overall score in this mode, Provider 1 would place above Provider 2 even though Provider 2’s standalone performance score is higher.

The example shows how **Performance & Stability** mode can surface providers that look excellent on performance but deliver a less consistent real-world experience.

## How to Select Results Calculation

![Results calculation toggle for Performance only or Performance & Stability](/files/6kwyMC5HiT4zB2o1uVHq)

1. Use the **Results Calculation** toggle in the query builder to switch between the two modes.


---

# Agent Instructions: 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:

```
GET https://docs.thousandeyes.com/product-documentation/internet-insights/provider-intelligence/build-a-query/specify-results-calculation.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
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.
