Within Amplitude Experiment, the **Experiment Analysis** view is where you’ll find the details of your experiment. Visible on the *Analysis* card under the *Analyze* tab, it gives you a convenient way to quickly take in the most important, high-level statistical measurements that help you determine whether your experiment was a success.

In this article, we will briefly describe what each of the columns in this table means, and how they relate to your experiment.

The first two columns, **Metric name** and **Variant**, are straightforward. The first contains the names of the metrics included at the beginning of the experiment. The top metric is the primary metric; all other metrics are secondary metrics. The second contains the names of the variants in the experiment. This includes the control and all treatments.

**Significance** is the likelihood that the performance displayed for each test variant is actually different from zero, and is not due to random fluctuations in the data. The higher this value is, the more confident you can be in your results. More formally, this can be described as* 1 - p-value*.

**Relative performance** is the percent change in performance of the control that would be needed to match the given variant’s performance. In other words, it measures the difference between how the variant performed and how the control performed. (In other products, this is often called **relative lift**.) You can cross-check this value by expanding a single metric’s section and then dividing the absolute lift for a variant by the absolute value of the control for that metric. (The absolute lift is the value in parentheses in the **Absolute Performance** column.)

The specific meaning of the **absolute value** column depends on the metric type. For **unique** conversions, the value here will be expressed as a percentage, indicating the percentage of users (over the total number of exposed users) who converted for each variant.

Otherwise, the value indicates the **aggregate** (total events, sum of property value, average of property value) per exposed user. The denominator used here is the total number of exposures. For example, 10 total events / 4 exposures = on average, an exposed user had 2.5 conversion events.

The **confidence interval** column displays the confidence interval (the probability that a parameter will fall between a pair of values) of the **difference** between treatment and control. Mathematically, it can be expressed as `diff = metric_value(treatment) - metric_value(control)`

.

The confidence interval shown reveals characteristics about what the experiment has observed thus far:

- Confidence Interval
**contains**0: There’s not enough evidence to indicate there’s a difference between control and treatment. - Confidence Interval
**greater than**0: The interval (upper and lower confidence bounds) is greater than zero. Amplitude Experiment has accumulated enough observations to reach statistical significance, and you can conclude that the variant has a**positive effect**compared to control—for example, if you are looking at lift, a variant with a confidence interval greater than zero can be expected to perform better than the control. - Confidence Interval
**less than**0: Amplitude Experiment has accumulated enough observations to reach statistical significance, and you can conclude that the variant has a**negative effect**compared to control. If, as in the last example, you are looking at lift, a variant with a confidence interval less than zero can be expected to perform worse than the control.