Support/Live Chat/Feedback loops

From MozillaWiki
Jump to: navigation, search

A main goal of the new Live Chat is to create feedback loops to leverage the service to continuously improve Firefox and our site. This page explains how the features in the chat PRD will be used to enable these feedback loops.

A problem is documented, but the solution doesn’t work or isn’t understandable.

  1. A poll targeted to helpers asks whether a linked article helped or not, possibly as a bubble next to the link.
  2. If a helper answers “no” to the above poll, the chat system makes it easy for helpers to leave feedback in the existing Article Forum, including a reference to the chat which triggered the comment.
  3. Content contributors read the article forum, ask Live Chat helpers why the articles didn’t work, and suggest improvements. Content reviewers approve an improved revision.


A new issue/bug is reported in live chat.

  1. If a helper isn’t familiar with the type of issue being reported, other helper(s) are consulted.
  2. A tag is applied to the chat, marking it as a potential new issue.
  3. If the user provides enough information to file a good bug, it should be filed right away. A senior helper can do this, or help can be requested in IRC or the community forum. If a bug is filed, the chat is tagged with this information for future reference.
  4. Administrator(s) regularly review chats with this tag, ensuring that bugs get filed when needed.

Search result broken

  1. A user clicks to Live Chat from a search result page or from a page found via a search result, giving the helper access to both the search terms and the referring page.
  2. If the helper suspects the search result was broken, he/she clicks a link to visit the actual search result tried.
  3. If the search page is indeed broken, the helper is able to mark this search result as ‘broken’, indicating the page that should have been included in the results.
  4. The search engine is tweaked to include the proper pages, after automatic or manual data analysis.

Determining the frequency of reported issues

  1. A tagging interface appears with each chat, allowing helpers to apply tags representing the issue.
    • Tags are added automatically for OS, version, and KB links reported by either the helper or user as successful.
    • Helpers can manually choose tags from a list or with an autocompleting interface.
  2. Automatic or manual data analysis is used to generate a list of the most common issues in Live Chat, based on the applied tags.
  3. Newly common issues are investigated by interested individuals after reports are made. Administrators and moderators can search chat logs to assist with this investigation. Information is added to bugs as appropriate.
  4. Product teams use information submitted to bugs to better understand the problem. They also use the frequency data, along with data from Input and the forum, to prioritize issues.