[[["易于理解","easyToUnderstand","thumb-up"],["解决了我的问题","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["很难理解","hardToUnderstand","thumb-down"],["信息或示例代码不正确","incorrectInformationOrSampleCode","thumb-down"],["没有我需要的信息/示例","missingTheInformationSamplesINeed","thumb-down"],["翻译问题","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["最后更新时间 (UTC):2025-08-18。"],[[["\u003cp\u003eExperiments compare multiple flow versions (variants) against a control version to assess performance using live traffic.\u003c/p\u003e\n"],["\u003cp\u003eKey metrics monitored during experiments include session outcomes like containment, live agent handoff rate, callback rate, abandonment rate, session end rate, no-match count, and turn counts.\u003c/p\u003e\n"],["\u003cp\u003ePreparing for an experiment involves choosing a flow, creating multiple versions of it, and deciding how much traffic to allocate.\u003c/p\u003e\n"],["\u003cp\u003eExperiments can be started, stopped, edited, and deleted, with status indicators including Draft, Pending, Running, and Completed.\u003c/p\u003e\n"],["\u003cp\u003eExperiment results are displayed in green or red to represent favorable or unfavorable outcomes respectively, with a note that higher or lower values are not always better depending on the metric in question.\u003c/p\u003e\n"]]],[],null,["# Experiments are used to compare the performance\nof multiple flow versions (*variant* versions)\nto a *control* version (normally a production version) while handling live traffic.\nYou can allocate a portion of live traffic to each flow version\nand monitor the following metrics:\n\n- **Contained** : Count of sessions that reached [END_SESSION](https://cloud.google.com/dialogflow/cx/docs/concept/handler#symbolic) without triggering other metrics below. Only available to agents using a telephony integration.\n- **Live agent handoff rate** : Count of sessions [handed off to a live agent](/dialogflow/cx/docs/concept/fulfillment#handoff).\n- **Callback rate**: Count of sessions that were restarted by an end-user. Only available to agents using a telephony integration.\n- **Abandoned rate**: Count of sessions that were abandoned by an end-user. Only available to agents using a telephony integration.\n- **Session end rate** : Count of sessions that reached [END_SESSION](https://cloud.google.com/dialogflow/cx/docs/concept/handler#symbolic).\n- **Total no-match count** : Total count of occurrences of a [no-match event](/dialogflow/cx/docs/concept/handler#event-built-in).\n- **Total turn count**: Total number of conversational turns (one end-user input and one agent response is considered a turn).\n- **Average turn count**: Average number of turns.\n\nPreparation\n-----------\n\nTo prepare for an experiment:\n\n1. Decide which flow will be used for the experiment. You cannot run multiple experiments on a single flow, so ensure that you have partitioned your agent into multiple flows.\n2. Create multiple [versions](/dialogflow/cx/docs/concept/version) for your flow. The differences between each version could be small or large, depending on what you want to compare.\n3. Decide on the amount of traffic that will be allocated to your experiment. If you are testing minor changes, you might start with a higher amount of traffic. For large changes that may be disruptive, consider allocating a small amount of traffic to your experiment.\n\nCreate an experiment\n--------------------\n\nTo create an experiment:\n\n1. Open the [Dialogflow CX console](https://dialogflow.cloud.google.com/cx/projects).\n2. Select your project to open the agent selector.\n3. Select your agent to open the agent builder.\n4. Select the **Manage** tab.\n5. Click **Experiments** to open the Experiments panel.\n6. Select the **Status** tab.\n7. Click **Create**.\n8. Enter a description.\n9. Select the [environment](/dialogflow/cx/docs/concept/version) that you want to run the experiment from.\n10. Select the flow for the experiment.\n11. Optionally enter the number of days in which the experiment will automatically stop.\n12. Enter the control flow version and the percentage of traffic that will go to the control version.\n13. Enter one to four variant flow versions, and the percentage of traffic that will go to the variant version.\n14. Optionally, click **Enable auto rollout and steps** for a gradual rollout of traffic to the variant flow. An automated experiment is based on *steps* , which are time durations in which a percentage of traffic is increased to the variant flow. Auto rollout only supports one variant flow.\n 1. Under **Rollout rules** , you can set one or more conditional rules to determine how the experiment should proceed through the steps.\n 1. If you select **Match at least one rule**, the experiment proceeds to the next step if at least one rule and the time duration for the current step are met.\n 2. If you select **Match all rules**, the experiment proceeds to the next step if all rules and the time duration for the current step are met.\n 3. If you select **Steps only**, the experiment proceeds according to the time durations for each step.\n 2. Under **Increase steps**, define a percentage of traffic to allocate to the variant flow and a time duration for each step. The default time duration for each step is 6 hours.\n 3. Select **Stop conditions** to set one or more conditions under which to stop sending traffic to the variant flow. Note that you cannot restart a stopped experiment.\n15. Click **Save**.\n\nStart and stop an experiment\n----------------------------\n\nYou can start a saved experiment\nor manually stop a running experiment at any time.\nStopping an experiment will cancel the traffic allocation\nand will revert traffic to its original state.\n| **Note:** If you stop an experiment while it is [pending](#monitor), results will not be available. If you stop an experiment while it is [running](#monitor), results might be inconclusive or missing.\n\nTo start or stop an experiment:\n\n1. Open the Experiments panel.\n2. Select the **Status** tab.\n3. Click **Start** or **Stop** for an experiment in the list.\n\nManage experiments\n------------------\n\n| **Note:** You can change variant traffic allocation while an experiment is running.\n\nYou can edit or delete experiments at any time:\n\n1. Open the Experiments panel.\n2. Select the **Status** tab.\n3. Click the option *more_vert* menu for an experiment in the list.\n4. Click **Edit** or **Delete**.\n\nMonitor status of experiments\n-----------------------------\n\nAll experiments, regardless of their status,\ncan be found on the experiments panel.\nExperiments can have four different statuses:\n\n- **Draft**: Experiment has been created, but it has never run.\n- **Pending**: Experiment has started recently, but results are not available yet.\n- **Running**: Experiment is running and interim results are available.\n- **Completed**: Experiment has been completed due to automatically or manually being stopped.\n\nViewing experiment results\n--------------------------\n\nTo see experiment results:\n\n1. Open the [Dialogflow CX console](https://dialogflow.cloud.google.com/cx/projects).\n2. Select your project to open the agent selector.\n3. Select your agent to open the agent builder.\n4. Select the **Manage** tab.\n5. Click **Experiments** to open the Experiments panel.\n6. Select the **Results** tab.\n7. Select an environment and experiment to see the results.\n\nGreen colored results suggest a favorable outcome,\nwhile red suggests a less favorable result.\nNotice that in some cases,\nhigher/lower numbers are not necessarily better\n(high abandonment rate / low abandonment rate).\n| **Note:** You will see \"no experiment result\" if not enough conversations have been through each variant of the experiment.\n\nLimitations\n-----------\n\nThe following limitations apply:\n\n- The [Enable interaction logging](/dialogflow/cx/docs/concept/agent-settings#settings-general) agent setting must be enabled."]]