# Metrics URL: https://docs.ablyft.com/guides/results/metrics/ > What the numbers on the results page mean. The results page has a **Summary** for the whole experiment and one card per goal. This page explains every number you find there. ## Overview at the top | Tile | Meaning | | --------------------- | ------------------------------------------------------------------------------------------------------------------------- | | **Runtime** | How long results have been counted, with the start date and the day number, for example "Since 2026-03-01 09:00 (day 5)". | | **Participants** | Number of participants, with the average per day. | | **Experiment status** | The status of the experiment and the time of the last change. | | Device chart | The share of participants per device type (desktop, mobile, tablet, other). | ## Summary table The **Summary** has one row per variation. The baseline is marked with "Baseline". The columns are: * **Variation** and its share of the traffic, shown as a donut. * **Participants**, with the share of all participants. * One column per goal with the **improvement** in percent, the conversion rate and total conversions ("CR / Σ"), and the confidence. Green means winner, red means loser, gray means inconclusive or baseline. The column of the primary goal is highlighted. If SRM check is enabled in the project, a badge **SRM** appears next to a variation whose traffic split does not match the planned split. See [Interpret results](https://docs.ablyft.com/guides/results/interpret-results/#check-the-traffic-split-srm). ## Goal cards Every goal has its own card with the goal name, its ID and a sentence such as "This is a *click* goal that counts *one* *conversion* per participant (where *increase* is better)". The table in the card depends on the goal. ### Goals that count conversions | Column | Meaning | | ------------------------------ | ----------------------------------------------------------------------------------------- | | **Conversions** (Participants) | Number of conversions and the participants they belong to. | | **Conversion rate** (Range) | Conversions divided by participants, with the range in which the true rate probably lies. | Example: 120 conversions out of 2,000 participants is a conversion rate of 6%. ### Goals that measure a value This applies to revenue goals and to custom goals with the measurement type **Value**. | Column (revenue) | Column (value) | Meaning | | --------------------------- | ------------------------- | --------------------------------------------------------------------------------- | | **Total revenue** | **Total value** | The sum of all values, with participants and conversions. | | **Revenue per participant** | **Value per participant** | Total value divided by **all** participants, including those who did not convert. | | Revenue per order (AOV) | Value per conversion | Total value divided by the number of conversions. | For value goals, the number of conversions is the number of recorded values. Revenue is shown with the currency sign of the goal. If you set a [value filter](https://docs.ablyft.com/guides/goals-and-measurement/goal-types/#value-filters), the filtered values are used. ### Columns for every goal | Column | Meaning | | ------------------------------------------------ | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | **Improvement** (Range) | How much better or worse the variation is than the baseline, in percent. For conversion goals it compares conversion rates. For value goals it compares the value per participant. The range shows the lower and upper limit. | | **Confidence Interval** or **Credible Interval** | A bar that shows the uncertainty of the improvement. The wider the bar, the less sure the result. A bar that crosses zero means the variation could be better or worse. | | **Certainty** | How sure ABlyft is that the variation really differs from the baseline, in percent. See below. | | **Statistical State** (Remaining duration) | **winner**, **loser** or **inconclusive**. For inconclusive variations, an estimate such as "≅ 12 days (≅ 3,400 participants) left". | Improvement is relative: the formula is (variation − baseline) ÷ baseline. A baseline rate of 5% and a variation rate of 6% give an improvement of +20%, not +1%. ### Certainty The **Certainty** column depends on the [statistical method](https://docs.ablyft.com/guides/results/statistics-methods/): * Frequentist and legacy: shown with `(p-value / z-score)`. A higher certainty means the observed difference is less likely to be random. * Bayesian: shown with `(posterior probability)`. The percentage is the probability that the variation is better than the baseline. The remaining-duration estimate is only a rough guide. It shows "The duration cannot be determined yet" if there is not enough data. ## Charts Open **Charts** in a goal card to see how the numbers developed over time. Choose between **Chart view: Cumulative Total** and **Chart view: Daily Breakdown**, then pick a tab: | Goals that count conversions | Value and revenue goals | | ------------------------------------------------------------------ | -------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Participants, Conversions, Conversion Rate, Improvement, Certainty | Participants, Total Value, Value per Participant, Avg. Order Value (revenue) or Avg. Value, Conversions, Conversion Rate, Improvement, Certainty, Value Distribution | If an experiment has three or more goals, an **Improvement Overview** chart compares the improvement of all goals. ## Filters and settings Open **Filters and Settings** above the summary to narrow down the results: | Filter | Effect | | ---------------------------- | ------------------------------------------------------------ | | **Start date**, **End date** | Only count data in this date range. | | **Device types** | Only count selected devices: Desktop, Mobile, Tablet, Other. | | **Statistical method** | Calculate with another method, temporarily. | While a filter is set, the page shows "You are viewing filtered results". ## Download Each goal card has a download icon (tooltip **Download CSV**). The file contains one row per day, variation and device type with the columns `goal_id`, `day`, `variation_id`, `device_type`, `participants` and `occurrence`. For value goals it also contains `values` and `sum`. ## Next steps * [Statistical methods](https://docs.ablyft.com/guides/results/statistics-methods/): how certainty is calculated * [Interpret results](https://docs.ablyft.com/guides/results/interpret-results/): when a result is reliable