--- title: "Define user actions on your web app with visual labeling in Product Analytics" description: "Learn how to label autocaptured web actions in Datadog Product Analytics without code changes, then reuse those definitions across behavioral analysis." author: "Sharon Ye, Adam Virani" date: 2026-10-08 tags: ["product analytics", "digital experience monitoring", "funnels"] blog_type_id: the-monitor locale: en --- Adding or renaming a product event has traditionally meant creating an engineering ticket. [Datadog Product Analytics](https://www.datadoghq.com/blog/datadog-product-analytics.md) uses the same SDKs and configuration as [Real User Monitoring (RUM)](https://www.datadoghq.com/blog/rum-without-limits.md), and those SDKs autocapture actions such as clicks and taps. But an automatically generated action name describes the element rather than the user intent behind it. That can make a business-level question such as "How many users started checkout?" depend on filters, custom events, or frontend instrumentation an engineer must define and maintain. With the Visual Labeler in [Datadog Action Management](https://docs.datadoghq.com/product_analytics/action_management.md), you can navigate your web application, select an element, and give the interaction a name that reflects what the user did, all without writing code or new deployments. Labeling now runs through the browser extension in your site's own context, so it works on the authenticated and gated pages (e.g., checkout, logged-in dashboards, account flows) that you could not label before. And Datadog applies each label to all matching interactions it has already recorded within your retention window, so trends and funnels work retroactively the day you define them. In this post, we'll show how you can: - [Label actions that matter by pointing at them](#label-actions-that-matter-by-pointing-at-them) - [Create shared definitions for key user actions](#create-shared-definitions-for-key-user-actions) - [Reuse labeled actions across Product Analytics](#reuse-labeled-actions-across-product-analytics) - [Apply labels to behavior you've already captured](#apply-labels-to-behavior-youve-already-captured) ## Label actions that matter by pointing at them [Visual labeling](https://docs.datadoghq.com/product_analytics/action_management.md#visual-labeling) provides a no-code alternative for labeling web application elements. From the Actions page, select an application and open the Visual Labeler. The Visual Labeler uses the [Datadog test recorder Chrome extension](https://chromewebstore.google.com/detail/datadog-test-recorder/kkbncfpddhdmkfmalecgnphegacgejoa) and lets you navigate to the relevant part of your site, switch to Label Actions mode, select an element, and save a labeled action. Elements that already have labeled actions are marked in the interface, which helps you understand what your team has defined before you add another label. When you click an element, Datadog shows how many times users triggered it over the past 7 days. If several interactions represent the same user action, you can click each of those elements to capture them under one label. By default, Datadog tracks clicks on the element only on the page where you labeled it. You can select Target all pages to track it across your application, and edit the element's CSS selector directly if you need to refine which elements the label matches. ![Visual Labeler defining a “Started checkout” action for a selected checkout element.](https://web-assets.dd-static.net/42588/1791480017-shopist-io.png) For example, suppose your application includes a checkout button whose autocaptured action name reflects its UI text. A product manager can select that element and define a label, such as “Started checkout.” This creates a business-level definition without changing the application's code. For cases that need more control, Action Management also supports manual labeling for web and mobile applications, including definitions based on action type, action name, CSS selectors for clicks, and page location. ## Create shared definitions for key user actions Labeled actions help prevent the meaning of an important product event from being redefined every time someone analyzes it. Without a shared definition, two analysts investigating the same interaction might use different action names, page filters, or variants of an element. Action Management gives teams a central place to define and retrieve labeled actions for an application. A manual labeled action can include a name, an optional description, and optional tags, as well as one or more event definitions. You can also combine several autocaptured actions under one labeled action when multiple interactions represent the same meaningful user behavior. ![Actions page listing curated labeled actions with descriptions and tags.](https://web-assets.dd-static.net/42588/1791480088-visual_labeling_2.png) Descriptions and tags are particularly useful as the number of labeled actions grows. A name such as “Started checkout” tells an analyst what the user intended to do, while its description can record what the definition includes. Tags can provide additional context for how your organization manages the action. Establishing conventions around these fields helps keep the action list useful as more teams contribute definitions. A practical convention is to name an action for user intent rather than the UI control that generated it. “Started checkout,” for example, remains more meaningful to an analyst than a name tied to a particular button implementation. Teams should also document the intended scope in the description and periodically remove definitions they no longer use. Note that deleting a labeled action also deletes it from dashboards where it is used, so check its consumers before removing it. ## Reuse labeled actions across Product Analytics A labeled action becomes a reusable event that you can use in Product Analytics charts. Instead of reconstructing the definition of “Started checkout” for each analysis, teams can select the labeled action and build analyses around the same underlying definition. [Funnel Analysis](https://docs.datadoghq.com/product_analytics/charts/funnel_analysis.md) steps can use actions or views, and a step can combine multiple events when users can reach the same outcome through different paths. For example, you can use “Started checkout” as an action step and analyze how many users subsequently complete the remaining checkout flow. You can then analyze conversion by session, user, or account and add filters to investigate how conversion differs across dimensions such as device or geography. ![Product Analytics funnel using the “Started checkout” labeled action as a conversion step.](https://web-assets.dd-static.net/42588/1791480139-screenshot-2026-10-06-at-12-55-11-pm.png) Product Analytics analyses can also connect behavioral questions with other Datadog workflows. Funnel visualizations can be shared to dashboards and Notebooks, where teams can review conversion alongside other data. When investigating funnel drop-off, Product Analytics also provides links to RUM, [Error Tracking](https://docs.datadoghq.com/error_tracking.md), and [Session Replay](https://docs.datadoghq.com/session_replay.md?platform=browser) for additional context about issues that might affect conversion. Another useful starting point is [Datadog Heatmaps](https://docs.datadoghq.com/session_replay/heatmaps.md), which aggregates interactions on a page into Click maps and Top Elements views, helping you identify which elements users interact with most. From a Click map, you can inspect action frequency and frustration signals and start building a funnel around behavior that warrants deeper analysis. ## Apply labels to behavior you've already captured When you create a labeled action, Datadog applies its name retroactively to matching interactions in your recorded data. Product Analytics retains session, view, action, and labeled-action events for 15 months, so if the relevant actions are already present in your retained Product Analytics data, you can use a new definition to analyze that historical behavior. Consider a feature that shipped 4 months ago without a custom event for its primary interaction. Now, a product team wants to measure how many users engaged with it. Rather than adding a custom event and waiting to collect a fresh dataset, the team can define a labeled action for the already-autocaptured interaction and analyze matching historical events within the available retention period. This workflow changes when teams need to decide what to instrument. Autocapture can provide a broad behavioral dataset before every future product question is known, while Action Management lets teams add meaningful definitions when those questions arise. Product teams can spend less time waiting on instrumentation changes and more time investigating the behavior that prompted the question. ## Build a common language for product behavior The Visual Labeler is generally available for web applications, including authenticated and gated pages. It turns autocaptured interactions into reusable definitions without a code change, and because Datadog applies each label to interactions it has already recorded, you can use those definitions to answer questions about behavior your application has already seen. To start defining actions, see the [Action Management documentation](https://docs.datadoghq.com/product_analytics/action_management.md) and the [Product Analytics data collection documentation](https://docs.datadoghq.com/product_analytics/data_collected.md). You can also explore how [Funnel Analysis](https://docs.datadoghq.com/product_analytics/charts/funnel_analysis.md) and [Heatmaps](https://docs.datadoghq.com/session_replay/heatmaps.md) help you analyze the behavior those actions represent. The visual workflow also requires the [Datadog test recorder Chrome extension](https://chromewebstore.google.com/detail/datadog-test-recorder/kkbncfpddhdmkfmalecgnphegacgejoa). If you're new to Datadog, .