docs: Add recommended decisions documentation for experiment configuration (#594)

* docs: Add recommended decisions documentation for experiment configuration

* docs: Update Metric Properties documentation with configuration details

* Restore propertysql page

* docs: Update internal links for Metric Properties and Explores documentation

Author: emetelka

docs/administration/property-cardinality.md

@@ -0,0 +1,9 @@
+# Metric Property Cardinality Limit
+
+You can minimize warehouse costs by limiting the number of computed values for all metric properties. Applying a limit here will cap the number of values computed for Explores and Property Analysis on experiments.
+
+The cardinality limit will limit computation to the top most frequently occurring values per property.
+
+You can configure the cardinality limit by clicking on the Metric Property Cardinality Limit tile in Admin.
+
+![Settings to add metric properties to a Protocol](/img/administration/property-cardinality-tile.png)

docs/administration/recommended-decisions.md

@@ -0,0 +1,42 @@
+# Recommended Decision Defaults
+
+The decision criteria aligns the team around a ship decision based on the measurement of the primary and [guardrail](/data-management/organizing-metrics/guardrails) metrics set in experiments. This recommendation will appear on the experiment when the experiment end date is reached or if the primary or guardrail metrics have achieved an outcome that is statistically significant.
+
+The default decisions for Standard and Non-Inferiority tests can be set it the Admin menu by clicking on the `Recommended Decisions` tile in Admin.
+
+![The Recommended Decision tile in Admin](/img/administration/recommended-decision-tile.png)
+
+:::note
+The recommended decision defaults can be overwritten based on the Experiment Protocol selected or edited with individual experiments.
+:::
+
+## Recommended Decision options
+
+Recommended decisions include:
+* Roll out winning variant - Ship the variant that performed the best compared to the control
+* Extend run time - Continue running the experiment to reach more precision
+* Discuss with stakeholders - Discuss the results with stakeholders to determine the best course of action
+* Do not roll out - Roll back to the control and turn off the treatment variant(s)
+
+## Standard Experiment Default Recommended Decisions
+
+In this first step, you can define the recommended decisions for all possible experiment outcomes when using a Sequential confidence interval method.
+
+If the confidence interval method for an experiment is set to Fixed Sample or Sequential Hybrid, any default recommendation set to ‘Extend Run Time’ will automatically change to ‘Discuss with Stakeholders.’ This is because these methods require a fixed run time to be determined before the experiment begins.
+
+![The Standard Experiment Recommended Decision page](/img/administration/standard-default-decision.png)
+
+## Non-Inferiority Experiment Default Recommended Decisions
+
+A non-inferiority test is used to determine whether a new treatment is not significantly worse than an existing one, within a predefined acceptable cutoff. In this context, the primary and guardrail metrics should remain within the acceptable risk threshold defined by this cutoff.
+
+For Non-Inferiority tests, define the recommended decisions for all possible experiment outcomes when using a Sequential confidence interval method defined in the Statistical Analysis Plan.
+
+If the confidence interval method for an experiment is set to Fixed Sample or Sequential Hybrid, any default recommendation set to ‘Extend Run Time’ will automatically change to ‘Discuss with Stakeholders.’ This is because these methods require a fixed run time to be determined before the experiment begins.
+
+![The Non-Inferiority Experiment Recommended Decision page](/img/administration/non-inferiority-default-decision.png)
+
+
+
+
+

docs/data-management/definitions/properties.md

@@ -70,9 +70,13 @@ Metric Properties can be used to refine Metrics or split experiment metrics when
 
 ![Configuring a Fact SQL](/img/properties/metric-property-fact-sql.png)
 
-Add a Metric property to any of your existing Facts or by creating a new one.
+Add a Metric Property to any of your existing FactSQLs or by creating a new one. A Metric Property only needs to be configured once and is able to be re-used many times for any Metrics created.
 
-This only needs to be configured once and re-use many times for any Metrics created.
+By default, Metric Properties will not be included in experiment computation to reduce experiment run time. This means that they are not available for analysis in [Explores](/experiment-analysis/deep-dive/explores) and [Property Analysis](/experiment-analysis/deep-dive/property_analysis) for a given experiment.
+
+This can be changed by toggling the Metric Property to enabled, which will make it available in all experiments. However, it is recommended instead that you set Metric Properties to be available for experiment analysis by choosing to include them in Protocols settings and in individual experiment settings.
+
+![Metric property with toggle to include in experiment computation](/img/properties/metric-property-toggle.png)
 
 ### Configure a Metric with a property filter
 

docs/data-management/definitions/property-sql.md

@@ -30,4 +30,4 @@ To create an Entity Property SQL follow the steps below:
 
 ![Write SQL Properties](/img/building-experiments/dimension-sql-query.png)
 
-Once you're finished adding mapping columns into Eppo, click **Save & Close**.
+Once you're finished adding mapping columns into Eppo, click **Save & Close**.

docs/experiment-analysis/configuration/experiment-advanced-settings

@@ -0,0 +1,14 @@
+# Experiment Advanced Settings
+
+Advanced Settings lets you choose which Metric Properties to include in your experiment analysis. To access these settings, click the edit icon in the top right corner of your experiment.
+
+Any Metric Properties you choose to include will make them available for analysis in [Explores](/deep-dive/explores) and [Property Analysis](/deep-dive/property_analysis).
+
+:::note
+Metric Properties will be available after the next experiment refresh
+:::
+
+To select which Metric Properties to make available, toggle `Use the company defaults` to off and select which Properties you want to include from the drop-down.
+ 
+![Settings to add metric properties to a Protocol](/img/experiments/protocols/protocol-advanced-settings.png)
+

docs/experiment-analysis/configuration/protocols.md

@@ -36,7 +36,7 @@ Specifying a Primary Metric is required. This is the metric that your experiment
 
 You are also able to specify optional [guardrail metrics](/data-management/organizing-metrics/guardrails). Guardrail metrics provide a level of governance and are primarily useful to track metrics that an experiment is not supposed to positively impact, but you want to ensure that there is no negative effect either. Guardrail metrics can use any cutoff bounds previously set or can have custom cutoffs set for the Protocol.
 
-After the metrics are set, you can then specify the recommended decision based on metric outcomes. This recommendation will appear on the experiment when the experiment end date is reached or if the primary or guardrail metrics have achieved an outcome that is statisically significant.
+After the metrics are set, you can then specify the recommended decision based on metric outcomes. This recommendation will appear on the experiment when the experiment end date is reached or if the primary or guardrail metrics have achieved an outcome that is statistically significant.
 
 ![Protocol decision criteria screen](/img/experiments/protocols/protocol-decision-criteria.png)
 
@@ -80,3 +80,11 @@ A recommendation will display when:
 * The Guardrail metrics are statistically significant positive or negative, or the guardrail cutoff threshold is breached
 
 The decision made can be recorded by clicking the `Make Decision` button. Note that the recommended decision does not have to followed, but we suggest adding notes as to why the team decided to deviate from the Protocol.
+
+## Advanced settings
+
+If your Primary or Guardrail metrics have metric properties, you can choose which Metric Properties to include in the experiment. This means that they will be available for analysis in [Explores](/experiment-analysis/deep-dive/explores) and [Property Analysis](/experiment-analysis/deep-dive/property_analysis) in all Experiments that use this Protocol.
+
+To select which Metric Properties to make available, toggle `Use the company defaults` to off and select which Properties you want to include from the drop-down.
+ 
+![Settings to add metric properties to a Protocol](/img/experiments/protocols/protocol-advanced-settings.png)