You can use labels to help with targeting for your Answer Bot triggers. Adding labels to your Answer Bot triggers is optional. Labels enable you to specify a limited subset of articles that you want to search within.
Without labels, Answer Bot searches all article titles and content to identify suggested articles. By adding labels in triggers, you can restrict the search to articles containing those labels.
This article covers three scenarios for using labels in triggers to better target article suggestions in your web forms. For information on using Answer Bot labels in the Web Widget, see Configuring Answer Bot for the Web Widget: Filtering Answer Bot suggestions by label.
Scenario 1: Targeting customer segments
The most common scenario for when to use labels in triggers is when you have different customer segments and you want to show each segment only the relevant articles. For example, suppose you are a mobile game developer and you support both Android and iOS platforms. When you get a request from a customer who's using Android, you want to show only Android articles.
To accomplish this, create an "Android" Answer Bot trigger with the condition based on your custom field "Platform = Android." Then, configure Answer Bot using labels to include only articles that contain the "android" label.
Likewise, set up an additional trigger for the iOS platform and label.
Scenario 2: Reducing the "noise" in your Help Center
Your Help Center might contain a lot of articles, most of which you never want to be used as Answer Bot recommended articles.
In this case, review your articles and add a "use_for_answer_bot" label to ~200-300 of the best articles. This will allow Answer Bot to focus and only suggest articles that make sense.
Scenario 3: Conducting a limited trial for a specific type of inquiry
While not recommended (it's a slippery slope), some customers have proven the value of Answer Bot by focussing it on a specific type of inquiry, such as "password reset" requests.
By creating an Answer Bot trigger that looks for specific words in the subject / description, and then using labels to restrict the articles suggested, you can limit the pilot and get some quantitative data to help support a broader rollout.
27 Comments
It would also be nice to be able to use a tag to exclude certain articles. I think in many circumstances it'll be easier to exclude the articles that you don't want to be included, rather than the ones that you do.
Craig
For scenario 2, can you confirm whether anything needs to be added to the Answerbot itself, as far as trigger rules (and if so what) etc or whether the articles themselves just need the label inserted?
Thanks
+1 to Craig's comment. Excluding by label would be great.
Hey all, Craig: thanks for the feedback. We're continually reviewing the best ways to make it easier to deal with the use cases.
Matthew, once you've added a generic "use_for_answer_bot" style label to all the articles you want to restrict Answer Bot to, you'll just need to add that label to the trigger by using the "Configure and Test" mode which is available when you edit the trigger. Let me know if you have any follow up questions?
Hi Mike,
Ideally I would like to make an article top 50 where keywords are unnecessary taking the queue entirely from Answerbot. Is this an option using the use_for_answer_bot label in articles and adding some kind of tag within the trigger?
I tested with your suggestion but still see answerbot suggested articles outside of those I would wish to be provided.
Can you give any recommendations here?
Thanks
Matthew
"Instead of using labels to improve the performance of Answer Bot, you can use labels to help with targeting for your trigger."
WHAT?
Wow -- I can't believe this hasn't been mentioned, but you seriously need some kind of "boolean logic" for your labels.
For example, I want to limit Answer Bot to focus on just the questions for the "Windows" version of "Program_A". But if I enter both the labels "Windows" and "Program_A", then I might get "Macintosh" labeled articles.
Isn't the point of creating these triggers to narrow down the scope of Answer Bot's suggestions?
We receive many tickets through emails but not all of the end-user's questions or problems can be suggested through the knowledge base. How can I use key words from emails to force Answer Bot to not respond?
For example, a user may email us to unsubscribe from our program, and we do not have that functionality through a button. We do not want Answer Bot attempting to suggest articles which are irrelevant to that, so I want to use keywords such as "unsubscribe", "cancel my membership", etc. to force Answer Bot not to respond and to let one of our Agents work that ticket.
Is something like that a functionality of Labels?
What Craig Willis said! It would be awesome to be able to exclude specific articles.
Hi,
With the new bult article actions for Zendesk Guide is not necessary anymore to use the "Ruby Script" for scenario 2.
https://support.zendesk.com/hc/en-us/articles/360001901768-Introducing-bulk-article-actions-for-Zendesk-Guide
I'm trying to figure out how these labels work with triggers. If I have 2 types of customers (say power and water) and I only want to display certain articles for water customers, how do I use the labels to limit which articles they see?
Hello Mark,
For this type of workflow, you would need 2 triggers and a way to scope the triggers to fire on one group of users vs the other. (I would recommend something like group assignment, user tags, or something easy to control and adjust.)
With that done, in the trigger for "water" customers you can add specific article labels that would pertain to those customer qs. The labels would restrict answerbot from looking at "unrelated" articles on your help center.
If you run into any issues with this setup, or find that articles are still showing up, you can send us an email at support@zendesk.com and we can spin up a ticket and help with the troubleshooting!
Is there then a way to assign all users from a certain domain to a group? I don't see how to create groups for any users other than agent users. This is for end-use customers and not agents.
I really agree with Craig Willis on being able to exclude certain tags from the AnswerBot suggestion. My main pain point is the following:
@AODocs we use the knowledge base to publish our Release Notes in a dedicated section. And as you can imagine release notes are covering a lot of topics, hence the AnswerBot matches nearly everytime a customer opens a ticket to us.
I wouldn't mind if the suggestion was relaying a change in behavior in the last release but when it suggests a Release Note article from our version 37.0 while we are currently in 49.0, you can imagine how far off the AnswerBot misses.
Has anyone already opened a Feature Request on this topic? I couldn't find one in the Guide Community.
Hi Thibault -
I don't see any feedback threads specific to this request currently; you're welcome to initiate one. Here are the guidelines for publishing a good product feedback post. Thanks for sharing your input!
For scenario 1 I am trying to configure this for answer bot with the web form, as I have two clients types.
However only the "configure and test" options seem to be available, with no way to make those labels conditional. So how do I get the answer bot on the web form to only deliver client A labeled article suggestions to clients of type A and client B labeled articles to clients of type B?
This was fairly easy to do for the emailed portion but is this non-functional for the web form??
Hey Michael!
Answer Bot for Web Form does not allow for this same type of conditionality that you can achieve thanks to Business Rules for your Trigger configurations.
With that said, Answer Bot will always respect User Segment restrictions, so if you have configured specific articles to be available only for users who belong to specific segments then you can use a common Article Label across those Articles without that exposing any restricted content to users that should not have access.
Please keep in mind that returning restricted results in this manner will require the users are signed in. Users who are not signed in would not have access to the restricted content.
As an alternative you may consider using multiple ticket forms to achieve this segmentation. If you only want to expose specific forms to users (this workaround will require they are signed in, and due to the type of code used will only apply to desktop browsers) you may consider the customization outlined in Hide ticket forms based on user's organization (Professional Add-on and Enterprise).
Hello,
Scenario 2 sounds like it would solve an issue I have with AnswerBot. I have one product I would like to have AnswerBot respond about, but it picks up KB articles from other products in the same help center. It seems that if I just put "use_for_answer_bot" in the one products KB articles, then AnswerBot will only use those articles, without needing to add or change anything else, including the AnswerBot trigger. Is that accurate?
Thanks!
Hey Eric, that's correct yes. Add that label to all the products articles (using bulk editing in Guide) and then configure your webform in the Answer Bot settings to filter only for that label.
I'm confused. I've written an article that does NOT contain the word 'freebird' in it - but made a label associated only to that article to test the usefulness of labels in articles w/ answer bot.
I then write an email to my helpdesk saying: "I really like freebird" and answer bot will not suggest that article. Am I missing something?
Neil, labels restrict the set of articles that Answer Bot will search within. Based on that reduced set, if there is no article that's relevant or related to the question then it won't suggest anything. We do into detail on using labels with Answer Bot here: https://support.zendesk.com/hc/en-us/articles/115012444447-Best-Practice-Using-labels-with-Answer-Bot
thanks, can't really wrap my head around it. Seems it's either so simple it makes it challenging - or it really is complex.
To me, the function of a label is a quick and easy way of saying "please make this article relevant if this word is used" ie: the article might mention the word "house" but not "home" so I add a label for "home" is that not what the intended function is?
Is what you're saying that a trigger would need to be made for each synonym instead of a label? ie "dwelling" , "residence" "property"
Can't help but wonder what purpose at all labels serve otherwise.
Hey Neil, sorry to hear it's causing confusion and for inception-linking you back to the exact post you were making a comment on... not that helpful!
Labels serve 2 purposes:
It's important to understand that Answer Bot uses a machine learning, natural language processing model to understand the topics contained in the question and then based on that, match that with the closest topic match within the articles. Because of this, misspellings, synonyms, etc, aren't a problem. Answer Bot will know that house, property, residence, dwelling are all topically related. When a question is asked, it uses the fully verbose natural language (not keywords) and determines the most relevant match.
Looking towards the future, now that we have HC segments we will likely be revising how we all for configuration of Answer Bot to use segments instead, rather than conflating labels.
Hope this clarification helps
Thanks Mike. That explanation helped heaps.
How do I add new labels to Configure and Test article labels? I've been trying to add a new label and hit return, but it doesn't get added.
We have lots of products. So instead of creating separate Answer Bot triggers for each product we want to use liquid to set the whitelist label based on keywords found in the query.
Question on this portion of Mike's post:
In this case, review your articles and add a "use_for_answer_bot" label to ~200-300 of the best articles. This will allow Answer Bot to focus and only suggest articles that make sense.
We have a lot of content for one of our brands, 35% of which content is internal. So if I was to use the "use_for_answer_bot" label and choose the top articles to have AB pay attention to, how would this impact the knowledge capture app (and suggesting content to our agents?)
Would I want to label both internal and external? And if I don't have the label to use with Answer Bot, will it never be presented wherever answer bot fires?
Please sign in to leave a comment.