Architecture

This is for:

In this article

Indexing Zendesk content
Query recommendations

The Coveo for Zendesk integration is installed within your Zendesk organization and connects to your Coveo index, Coveo Search API, Coveo Machine Learning (Coveo ML), and your usage analytics data. The integrations for Zendesk Support and Zendesk Guide work similarly, with the primary difference being that Guide is a self-service customizable portal, and therefore elements of the guide are replaced with Coveo components (see Integration for Zendesk Guide).

The following diagram illustrates the integration architecture.

Note

Coveo generates a search token based on the information retrieved from the Zendesk JWT token. For more information, see Search Token.

Indexing Zendesk content

An index is a data structure used to securely store items from all sources across a given organization, such that they can be retrieved quickly when they match a specific query.

When you integrate Coveo to your Zendesk instance, your Zendesk content is retrieved periodically by using the Zendesk connector, and stored in your Coveo organization index. The indexed items can include tickets, users permissions, articles, and more.

Note

Although the Zendesk source is refreshed every hour by default, you can modify the update schedule to suit your indexing requirements.

Query recommendations

A query performed within the Coveo for Zendesk search interface is sent to the Coveo Search API. It’s then forwarded through the query pipeline and then sent to the index to retrieve relevant results from your Zendesk content. The results are sorted in order of relevance by Coveo ML, after which they’re returned to the Coveo for Zendesk search interface.

The query recommendations become more relevant over time. With repeated use, your usage analytics data is compiled and assessed by Coveo ML to identify trends and patterns. Coveo ML then builds a model around your data and recommends the best results for a given query and user.