Launching chatbot for Looker support chat widget

  • 4 November 2021
  • 2 replies

Userlevel 2
  • Looker Staff
  • 16 replies


  • The chatbot is only accessible to users who are permitted to use chats; there’s no change in our chat access policy. Users will still be able to engage with a support analyst by clicking “Get in Touch” in the chat window.

  • The early launch starts from November 22nd, at which time it will be by default switched on for instances on the current non-supported versions. If you want to switch the chatbot off, please upgrade to one of our supported releases. In early December, it will be available to all customers; after that, if you want to opt out, please let us know by filing a support request.

  • The chatbot can be toggled for the whole instance. We cannot selectively switch off the chatbot for a particular user.



Today, authorized users can reach Looker support analysts via the in-product chat widget. Our chat support feature has been available to both developers and admin users for Looker related technical requests. Working with the Looker support team has been considered helpful for developing Looker knowledge and resolving technical challenges. On the other hand, we have heard some feedback from our customers on the following topics:

  • Quick questions are not always quick: the solution to your question might have been documented in our knowledge base, however you might have to take time to go through the conversation with our analysts and wait before you get the answer.

  • Explaining all your pre-chat research when you talk to the agents: You have done lots of pre-work before chatting to our agent, however it might be time consuming to make sure the agent is aware of your request and knowledge level before they can jump into the investigation/solution.

To address the above problems, we are going to add a machine learning powered chatbot support layer on top of our current chat support. 

How does it work?

When you initiate the chat widget, you will be able to ask questions to the chatbot, and the chatbot will suggest relevant articles from our Help Center. If the suggested article answers your question, you can directly apply the solution. In case the suggestions do not fully answer your question, you can still get in touch with our support analysts, and they will be able to see what research you have done to avoid any duplicated work.


What does this mean for you and your users?

  1. Save time for quick questions: you get answers instantly, no need to wait for the support analyst to get back to you.

  2. Fully take advantage of our content resources: Our Help Center is the one-stop-shop where you can find the most common questions and use cases, you don’t need to go to different places again for an answer.

  3. Chat support can be a continuation of your research: in case you still need our support analysts’ help, they can see your current research path, and be able to continue from where you left off. In this case you will likely spend less time on the support conversation so you can focus on your core work.

What has not changed about our support?

  1. Our chat support is still available 24/5 via the chat widget; you can reach out by clicking on the “Get in touch” button.

  2. Our documentation, Help Center, and Community Technical Tips section are still publicly accessible and will continue to support you.

What is the Looker support chatbot great for:

Our chatbot can be helpful for these types of questions:

  • How do I do XYZ in Looker? (How do I create cohort analysis in Looker?)

  • What is XYZ about? (What is Aggregate Awareness and how does it work?)

  • What’s the best practice to achieve XYZ? (What’s the best way to migrate a project to a new git repo?)

  • I got an error, what does it mean and how do I troubleshoot it? (How do I fix “Unknown or Inaccessible Field”?)

What is Looker support chatbot not used for:

  • A complete replacement of Looker documentation and knowledge base: currently we are in the process adding more resources into our chatbot’s search range, meanwhile most of our documentation and knowledge resources will still reside in, and We expect to see all our knowledge resources being suggestable by chatbot within H1 2022.

  • Very customized use cases/investigation that requires context understanding: for use case or investigation that requires an understanding of business or technical context, reaching out to our support analyst might be the best option.

  • Reporting a bug or sharing product feedback: we suggest talking to our chat agents for raising and troubleshooting bugs, while using the product ideas platform for feature requests.

  • High priority requests

What can you expect during the early launch:

  • All the articles suggested by our chatbot have been reviewed by our team; they are trusted knowledge resources created by Looker consultants and analysts.

  • Since the chatbot is powered by machine learning, it will improve over time. If the suggestions are not helpful at first, you can help the chatbot improve by giving feedback. When you view any suggested articles, the chatbot will collect your feedback by asking you to confirm if it’s “Helpful” or “Not Helpful” with two options at the bottom of the chat widget. Please share your feedback; that will help us provide more accurate suggestions.

  • We will actively track the requests and solutions during the pilot to help ensure we are surfacing the most useful content. However, if you have any particular resources/article you would like to see on our help center, please feel free to share the feedback with our support analysts via chat or email.

  • All your users who are permitted to use chat support will receive the chatbot service. We cannot selectively switch off the chatbot for a particular user.

What can you expect in Q1 2022:

  • The content resources will keep growing.

  • The chatbot flow and process might change/improve.

  • The chatbot suggestions might become more accurate.

2 replies

Userlevel 2


We are pushing off the early launch date to Nov 22nd due to some delay in our preparation and approval.

Userlevel 3

Thank you @xin for sharing this news!