This Quickstart guide is written to help new Answer Bot users get a simple Answer Bot configuration up and running with their existing help center content.
Answer Bot is included with all Zendesk Suite plans and is available as an add-on for Zendesk Support + Guide users. To get the most out of Answer Bot, your help center should have at least 10 articles that cover commonly-asked questions.
This article contains the following steps:
Step 1 : Activate Answer Bot
If you're a Zendesk Suite customer, you'll need to activate Answer Bot on your account. If you're using Support + Guide, you need to sign up for the 30-day trial to start using Answer Bot If you’ve already done this, skip ahead to Step 2.
To activate Answer Bot
- In Admin Center, click the Channels icon () in the sidebar, then select Bots and automations > Article recommendations.
- Click the Activate Answer Bot button (Zendesk Suite users) or Try Answer Bot for 30 days to activate your 30-day unlimited trial (Support + Guide users).
Step 2: Evaluate your current help center content
Take a look at the content offered in your help center. You can optimize your articles for Answer Bot with some small changes.
In particular, look at the following article elements:
- Titles: Try to title your articles using language and phrasing your customer might use in a search, or in a ticket description. Simple questions ("How do I reset my password?") or phrases ("Resetting a password") are a good approach.
- Introductions: Answer Bot weighs the first 75 words of an article most heavily, so getting relevant keywords in your introduction is key. Try to start your articles with clear, focused paragraphs.
- Topics: Create bite-sized articles that address single, narrowly-focused topics, rather than long articles with multiple separate, if related, sections. For example, instead of writing one article including all profile-related settings, break that information up into separate articles for each setting. You can connect these smaller articles together using a related articles list, or collect them in a resource article containing links to all profile-related articles.
If you are using Content Cues, you can leverage the suggested articles feedback to improve your Answer Bot results.
Keep in mind that these are just suggestions for building a help center that will work with Answer Bot. You can apply any, all, or none of them to your help center content at any time during the trial.
Step 3: Enable Answer Bot
Next, you can enable Answer Bot throughout Zendesk. Where you add it depends entirely on your Zendesk products and integrations, and where you think it will best fit into your setup. You can use it to offer Article Recommendations through email notifications or web forms, in the Web Widget (Classic) or the mobile Support app, as part of a Zendesk messaging conversation flow, and in conjunction with the Slack integration with Zendesk. For information on your options for using Answer Bot across Zendesk, see Understanding everywhere you can use Answer Bot.
- Enabling and configuring Article Recommendations for email notifications
- Enabling and configuring Article Recommendations for web forms
- Using the Answer Bot for Slack integration
- Searching and linking articles using the Knowledge Capture app
- Enabling and using Answer Bot in Web Widget (Classic)
- Answer Bot API
Step 4: Assess Answer Bot performance
With a few simple configurations, you can evaluate how Answer Bot impacts your customer interactions. Creating ticket views, as well as using your Insights or Explore dashboards, can help you decide if Answer Bot is the right tool for you.
Creating views for Answer Bot tickets
You can create views for tickets affected by Answer Bot, to easily see how your end users are interacting with Answer Bot, and to assess whether it is working for you.
You’ll want to use triggers and tags to identify Answer Bot tickets and sort them into Answer Bot-only views. Your basic Answer Bot views can include (but are not limited to):
- Tickets Answer Bot has fired on
- Tickets Answer Bot has helped customers self-solve
- Tickets Answer Bot has fired on but not solved
Answer Bot has a dedicated dashboard in Explore that monitors metrics for the feature's two main usage types:
- Article Recommendations (for email, web forms, and live chat). Key metrics include Suggestion rate, Click-through rate, Resolution rate, and Rejection rate.
- Flow Builder (for messaging). Key metrics include Total users, Engaged with bot, and Transferred to agent.
Answer Bot begins gathering usage data immediately upon implementation. You can begin monitoring performance through your Explore dashboard right away, or wait until more information is available. Commonly, users check data after 24 hours, 48 hours, one week, and three weeks to see how Answer Bot performance changes over time.
"Step 2: Evaluate your current Help Center content"
Where do I view that?
Are you able to link where you saw this specific bullet point? In general regarding Answer Bot since it is directly linked to your Help Center you will want to evaluate your articles etc to make sure when you are live with your Answer Bot that it is giving the most accurate information. There is some additional information on how to do this below:
Guide (Help Center) enables you to provide end users with a complete self-service help center that helps your agents help your customers. With Guide, you can create a help center that includes:
A branded customer-facing support site
A knowledge base, for publishing self-service content
Community, for customer collaboration
A customer portal, where customers submit tickets
I hope that helps! Have a great day.
Jason Schaeffer | Customer Advocate | Support@zendesk.com
Thanks for bringing this to our attention! I'll loop in our documentation team so they can get this information updated for you.
I'd recommend taking a look at this article in the meantime: Explore recipe: Analyzing Answer Bot activity
I hope this helps!
Can you test it in Sandbox ?
We are working with different brands and I have labelled the articles accordingly, but yet I don't to activate the bot and have customers from X brand to be able read the articles related to Y brand etc ?
We are facing a weird problem, I can login to zopim.com dashboard but using the same credentials not on Zendesk.com and also, there is no Zendesk url associated with the registered email id. So not able to access Admin area to enable the answerbot. What should we do in order to get it enabled? Or get our own Zendesk url?
Hello! I have few questions:
- is there a way for coordinators that work on tickets to evaluate the answer bot suggestions provided to a ticket? Coordinators can see the suggestions but cannot evaluate or suggest updates/changes directly from there (like Knowledge capture).
- is possible to evaluate the Rejection rate in the metrics? So for example, knowing why users rejected an article? Is there a sort of survey that we can implement to know why a user rejects an article (not useful at all, missing information, etc...)
- there is a way to introduce a support link directly in the classic web chatbot? We are currently using an easy web chatbot that only suggests articles but I would like to introduce there a link to support so users can directly open a ticket if they cannot find an answer. Also, is possible to track the activity of the chatbot in some way?
We implemented simplified email threading for email applications, and we noticed that our answer bots are no longer embedding the help center article in the email. We've seen several end-users say that they don't see the article within the email. Is the simplified email threading not supported in the answer bots?
Hi Allen Lai | Head of CX, Otter.ai
I went ahead & created a ticket on your behalf so we can look into the issue with the Answer Bot. Please keep an eye out for our update
Hi! I had a couple questions about Answer Bot.
Let me address the 3 points you talked about:
Again, feel free to create a ticket if you want something more specific to your instance.
I don't find any mention of Answerbot behaviour with articles with user segments,
Please sign in to leave a comment.