Capture a search event

This is for:

Developer

This article explains how to capture a search event and send the information to Coveo along with the required fields.

When to capture a search event

The search event is by far one of the most commonly logged commerce usage analytics event. Any time a ranked list of products is shown to a visitor, whether it be from submitting a search query, navigating to a listings page or viewing a product recommendations carousel, a search event must be sent to Coveo to capture user intent and begin tracing the user journey. The event will be used to measure attribution from a search results page, a product listings page, or a recommendations component.

Headless controllers and Atomic components

Headless and Atomic components log search events under the hood on your behalf. You don’t need to manually log search events if you build your UI with these technologies.

The following sections provide guidance on the necessary configurations to ensure that search events are appropriately logged.

Initialization configuration

This section is dedicated to explaining the initial configuration of your components. Specifically, we will discuss the importance of properly setting the searchHub, pipeline, and language parameters.

Set the search hub

Depending on the type of interface the event originates from, there are naming conventions for the search hub value you need to follow. We recommend setting the search hub on Headless and Atomic to control how the queries are processed in the Coveo Administration Console (see Query routing).

When configured, the frameworks will automatically send the value in search events in the originLevel1 field. You can either configure the search hub in the authentication method, or set it at interface level.

Language

In Coveo Machine Learning (ML) models, multilingual search and recommendations are supported, allowing you to deliver personalized and relevant content to users in different languages.

To ensure accurate and effective language-based results, it’s important to consider the language of the search query and align it with the language specified in the various usage analytics events.

Headless and Atomic automatically set the language of usage analytics events to the language of the search query.

The language is set to en by default; however, if your use case serves content in multiple languages, we recommend specifying the language by setting the value of the locale field when initializing your components. The locale field is formed by combining the language and region values with a hyphen, like en-US, where en represents the language and US represents the region.

Note

The value of the language field must be a two letter ISO 639-1 code.

Query pipeline

When the interface sends a query to your Coveo organization, it goes through a specific Coveo query pipeline before reaching the index. This can be done by specifying a Search Hub and then using condition based routing in the Administration Console to target a specific pipeline.

A query pipeline uses its rules to transform that query and apply specific features following a specific order of execution. While it is not necessary to specify the query pipeline, we recommend that you do, as it allows you to filter and segregate events when looking at usage analytics reports.

For Headless

The following code snippet explains how to set the searchHub, and language fields when initializing your Headless search engine. Specify these values using the SearchConfigurationOptions object as shown below:

import { buildSearchEngine, getOrganizationEndpoints } from '@coveo/headless';

export const headlessEngine = buildSearchEngine({
  configuration: {
    organizationId: '<ORGANIZATION_ID>',
    accessToken: '<ACCESS_TOKEN>'
    organizationEndpoints: getOrganizationEndpoints('<ORGANIZATION_ID>'),
    search: {
      searchHub: "<SEARCH_HUB>",
      locale: "<LOCALE>",
    },
  },
});

For Atomic

The following code snippet explains how to set the searchHub, and language fields by setting these properties on the atomic-search-interface component:

<body>
  <atomic-search-interface
    id="search"
    search-hub="<SEARCH_HUB>"
    language="<LANGUAGE>"
  >
    <atomic-search-box></atomic-search-box>
    <!-- ... -->
  </atomic-search-interface>
</body>
<body>
  <atomic-search-interface id="search" language="<LANGUAGE>">
    <atomic-search-box></atomic-search-box>
    <!-- ... -->
  </atomic-search-interface>
</body>

Origin levels

The originLevel feature provides valuable information about the origin of usage analytic events, similar to the Search Hub. Several fields are associated with the originLevel feature:

  • originLevel1 (required): This field should match the value of the Search Hub. If you initialize Headless or Atomic with the SearchHub configuration, this field is automatically set.

  • originLevel2 (recommended): This field lets you differentiate between different tabs within the search interface if multiple tabs are present. When using the Tab controller, Headless can automatically populate the originLevel2 field with the ID specified during the initialization of your controller.

    • For the "Search" interface without any tabs, you can send the value "default".

    • For the "Listing" page, send the path name of the listing (e.g., "my-products/shoes").

    • For "Recs", you can send the value "default".

  • originLevel3 (recommended): This field represents the URL of the page that redirects the browser to the search interface. It helps you analyze the effectiveness of various entry points leading to the search experience. It’s typically extracted from document.referrer.

During engine initialization, you can set the values for the originLevel fields. However, if you need to dynamically modify these values, you have the flexibility to modify the usage analytics data sent to Coveo. You can do so by leveraging the analyticsClientMiddleware property of an AnalyticsConfiguration to hook into an analytics event payload before it’s sent to Coveo. This can be done for both Headless and Atomic.

Custom data

Along with the regular fields added by Headless and Atomic, UA events must contain a customData JSON with additional fields. While the library adds some fields to this customData JSON, there are other fields that have to be manually added by using the Headless library:

Note

If you’re using the Atomic library, you’ll need to access Headless through Atomic.

customData.context_website

This field lets you differentiate between events from different websites that are part of the same Coveo organization. Coveo ML can be leveraged to learn from events from a specific website.

Note

The value customData.context_website specified here corresponds to the trackingId set for Coveo UA.

For example, using the Headless library, you want to specify the website context for this page is about sports as follows:

import {buildContext} from '@coveo/headless';
import {useContext} from 'react';
import {AppContext} from '../../context/engine';
export function Context() { 1
  const {engine} = useContext(AppContext);
  const ctx = buildContext(engine!);
  ctx.set({website: 'sports'});
  return null;
}

Where:

1 Send context via Headless which appends context_ to fields.

Sending user information

If your implementation includes an authentication mechanism, we highly recommend sending additional fields regarding the user performing the request.

To ensure the transmission of these user oriented usage analytics fields to Coveo, it is necessary to employ search token authentication. This involves generating search tokens while providing the name, userDisplayName, and userGroups fields. The generated search token is then utilized to send requests via Headless and Atomic.

Note

The name, userDisplayName, and userGroups fields are extracted server-side by Coveo, which means they will not be visible in the Network tab if you inspect the outgoing analytics request.

Capture a search event using Coveo UA library commands

We strongly recommend using either the Headless or Atomic libraries for capturing search events as these libraries are designed to facilitate the logging of such events by automatically including the appropriate metadata and simplifying the implementation process. However, if you choose not to use these libraries or want to track search events on a component not powered by Coveo, you can still log search events manually using the Coveo UA library.

The following is an example of the JavaScript code that uses the Coveo UA library to send a search event:

coveoua('send', 'search', { 1
  'actionCause': 'searchboxSubmit',
  'queryText': 'shoes',
  'responseTime': 23,
  'anonymous': true,
  'originLevel1': 'MainSearch',
  'originLevel2': 'default',
  'originLevel3': 'http://localhost:3000/search',
  'searchQueryUid': '514970dd-7c76-429f-8e80-6ea2f957f303',
  'language': 'en',
  'trackingId': 'Sports',
});
1 Send the search event, specifying the various required fields.

Search data payload requirements

The following table lists the fields sent along with the event and the solutions that leverage them. Search, Listings, and Recs columns refer to whether the ML that powers those solutions needs those data points. The Reporting column is for general reporting purposes that encompasses all solutions.

Field Priority Search Listings Recs Reporting

actionCause

Required

check

check

check

check

anonymous

Required

check

check

check

check

language1

Required

check

check

check

check

numberOfResults

Required

check

queryText

Required

check

check

check

check

responseTime

Required

check

searchQueryUid2

Required

check

check

check

check

trackingId

Required

check

check

check

check

originLevel1

Required

check

check

check

check

originLevel2

Recommended

check

check

check

originLevel3

Recommended

check

check

check

advancedQuery

Recommended

check

check

check

queryPipeline

Recommended

check

splitTestRunName

Recommended

check

splitTestRunVersion

Recommended

check

username3

Recommended

check

userDisplayName3

Recommended

check

userGroups3

Recommended

check

customData.facetField4

Recommended

check

check

check

customData.facetId4

Recommended

check

check

check

customData.facetTitle4

Recommended

check

check

check

customData.facetValue4

Recommended

check

check

check

customData.partialQuery5

Recommended

check

customData.partialQueries5

Recommended

check

customData.suggestions5

Recommended

check

customData.suggestionRanking5

Recommended

check

facetState6

Recommended

check

check

  1. Defaults to english.

  2. Refer to Extract the searchUid.

  3. User specific fields which can be sent if the user is authenticated.

  4. Field sent when the user interacts with a facet. Refer to Implement facets.

  5. Field sent when the user interacts with Coveo ML query suggeston. Refer to Handling Coveo Machine Learning query suggestion selection.

  6. Array containing items that represent active facets applied in the search. Refer to Implement facets.

Refer to Log search events for more information on the fields discussed above.

Search event validation

To validate the search event was sent, use the developer tools for client-side request verification to inspect your visit to ensure everything done using the UI is captured as a set of consistent and coherent events. This must occur after the implementation is applied.

Inspect the search event through the network browser and ensure it satisfies the following:

Barca search event network browser

1 A search event is sent whenever the user interacts with the search box or a Facet.

2 The actionCause parameter of the search event is attributed to the correct value. Some of the most common are:

User action identifier Description

facetSelect or facetDeselect

Selecting or deselecting the facet that re-renders the search results page.

facetClearAll

Deselecting all facets, re-rendering search.

interfaceLoad

When a user loads a search interface, for example, when reloading the search results page.

omniboxAnalytics

When a search result originates from a query suggestion.

recommendationInterfaceLoad

A visitor views a recommendation.

searchboxSubmit

When a user submits search queries in the search box

searchFromLink

A visitor clicks on a navigation link and lands on search results page.

See the full list of action causes for analytics purposes.

3 The anonymous field is set to the correct value based on the implementation. The default value is false.

4 The value of the originLevel1 must contain a value to calculate the attributed value for each product discovery solution. For example, search for search, listing for product listings, and rec for product recommendations.

5 Ensure the searchQueryUid is unique for every search event.