# What is a good Apdex score?

> An Apdex score above 0.85 is generally considered good. Here's how Apdex is calculated, what the thresholds mean, and how it relates to load test results.

Source: https://loadster.com/faqs/good-apdex-score/

An Apdex score above 0.85 is typically considered good, 0.94 and above is considered excellent, and anything below 0.7 is a problem.
Apdex (short for "Application Performance Index") condenses all of your measured response times into a single satisfaction score between 0 and 1, where 1 means every user had a satisfying experience according to your satisfaction threshold.

## How Apdex Is Calculated

Apdex starts with a threshold that you choose, called T, representing the response time a user on your site or application finds satisfying. The T value won't be the same for every site or every organization; some user communities tolerate slower response times than others, so the T value you choose should be based on your team's requirements.

When calculating Apdex, requests at or under T count as *satisfied*, requests between T and 4T count as *tolerating*, and anything slower than 4T counts as *frustrated*. The formula is:

```text
Apdex = (Satisfied + Tolerating/2) / Total samples
```


Let's say you set T to 1 second (your satisfaction threshold) and collect 1,000 samples: 800 finish within 1 second, 150 land between 1 and 4 seconds, and 50 take longer. That gives (800 + 150/2) / 1000 = 0.875, a good Apdex score.

Keep in mind that the score is only as meaningful as the T value you pick. The exact same response times might produce a great Apdex score or a terrible one depending on where you set the threshold.
Set a satisfaction threshold that's appropriate for your site.

## Apdex Score Ranges

The typically used interpretation bands for Apdex scores are:

- **0.94 to 1.00** — Excellent
- **0.85 to 0.93** — Good
- **0.70 to 0.84** — Fair
- **0.50 to 0.69** — Poor
- **Below 0.50** — Unacceptable

Treat these as guidelines rather than laws of the universe. What's acceptable depends on context: a checkout page and a heavyweight report export deserve different thresholds because users have different performance expectations around them, and a single site-wide T blurs that distinction.

## Apdex vs Response Time Percentiles

Apdex trades detail for simplicity. One number is easy to put on a dashboard and track over time, so it's a popular way to align teams around performance. However, it sometimes hides how bad the bad experiences actually are. A frustrated request at 4.1 seconds and one at 40 seconds could hurt your Apdex score identically, even though they're drastically different problems for your users.

[Percentile response times](https://loadster.com/faqs/response-time-percentiles/) answer the same underlying question with more nuance: instead of "what fraction of users were satisfied?", they tell you "how slow was it for the slowest 10%, 5%, or 1% of users?".
That's why Loadster doesn't compute an Apdex score; its reports give you detailed percentile response times (p50 through p99) alongside error rates, which together show both the median experience and the longtail.
See [Analyzing Test Results](https://loadster.com/manual/analyzing-test-results/) for more about how to read and interpret these percentiles.

Apdex and percentiles can be complementary rather than competing. If your team already tracks Apdex in an APM tool, a load test's percentile and error graphs will help you understand why the Apdex score moves under heavy load, and help you identify outliers to optimize for.