Recently the Community hosted a virtual event where we took a look into the way we do Knowledge Management here at Zendesk. Our experts from our Self Service Team showcased their best practices around KCS, explained the importance of the solve/evolve loop and how it can identify any gaps missed by Knowledge-Centered Service. We also sat down with customers and addressed questions live from the community on our topic of focus!
Video Recording
Questions from the Q&A
Q: I'd love to get some insights or tips on how to segment our Help Centre between public documentation and restricted documentation. Some of our documentation should ideally be restricted to a specific audience and not visible to all users.
A: We use user segments to partition our help center between different groups of users and specify if an article is only internal-facing or public. Each internal section is marked "iKB" to designate that the documentation is not public-facing.
Supporting Documentation:
Creating user segments for Guide user permissions
Q: What are your best practices to manage article updates, specifically with translations?
A: We use a combination of periodic reviews of content (Self Service Analyses) where we analyze tickets and determine how well our content matches up to the questions that are raised by customers for a given topic (typically a product topic such as Talk configuration or Explore reporting).
We rely on a combination of machine translation and manual translation for our content. The articles with higher traffic are manually translated.
We also rely on demand-driven updates - if an advocate flags an article for an update or an incorrect translation, we work with them to correct the article to make the content clearer.
Q: Can you share any user permission management and article publishing and approval flows?
A: Our documentation team and KCS publishers are the only agents with publishing permission (or Guide Admin access) in the help center. However, all agents have access to flag articles or suggest a new article using the Knowledge app in the context panel.
Once an agent flags an article or submits an article draft, the agents in the KCS group review the ticket, make any changes necessary, and publish the article in the appropriate section.
Supporting Documentation:
Searching, linking, and quoting content in tickets
Q: What is the most efficient workflow for getting out of date articles reviewed/assigned?
A: We use the Team Publishing feature to allow agents to create new articles or update existing ones without publishing them. This way, our super Knowledge Champions can go ahead, review the article as per our content standard and then publish the article. Content standards are lengthy and not everyone likes them. We have a limited number of KCs. This workflow allows us to encourage agents to share their knowledge without feeling threatened by writing guidelines.
Supporting Documentation:
Zendesk on Zendesk: How we do KCS
Q: How can I use Content Cues and make Answer Bot improvements?
A: Content Cues and AB are two different features that work separately but depending on your workflow could be used as a whole. From Content Cues, you can get ideas for new topics of articles. You can then use labels to target those articles in your Answer Bot. We call it "feed the bot". Just make sure articles are created following the AB writing guidelines.
Supporting Documentation:
Optimizing your articles for Answer bot
Q: How do you use Search metrics for content ideas?
A: We use both the Search dataset in Explore and the Search Console from Google. Our help center is a public help centre, which means that our customers can easily find their information through Google, and that is why we use the Console. The process in both datapoints is very simple. We look at large spreadsheets and make sure that the search query used by our customers surfaces the right content. If nothing exists, then we create an article. We don't target all the search queries, just the ones that are relevant.
Q: Does Guide track who makes changes?
A: Yes, you can find this information in the version history.
Q: I would love to learn how to efficiently share knowledge across multiple brands (and how to organize content blocks) as well as advise for working with multiple Zendesk instances.
A: Each help centre is separate, but multi-brand can be used to support multiple brands using content from a single knowledge base. If you create separate help centers, the content is also separated. Now, in terms of sharing knowledge, your agents can use the Knowledge feature for AW and quickly find the right article in the right brand regardless of the brand in which it is published.
Q: I would love to learn how to efficiently share knowledge across multiple brands (and how to organize content blocks) as well as advise for working with multiple Zendesk instances.
A: Each help centre is separate, but multi-brand can be used to support multiple brands using content from a single knowledge base. If you create separate help centers, the content is also separated. Now, in terms of sharing knowledge, your agents can use the Knowledge feature for AW and quickly find the right article in the right brand regardless of the brand in which it is published.
Supporting Documentation:
Searching, linking, and quoting content in tickets
Q: What specifically is in the roadmap to enhance or expand on these features?
A: You can find our current roadmap here.
Supporting Documentation:
Q: We are new to the Guide Explore dataset so want to learn as much as possible about reporting on the Knowledge base.
A: There are five Guide datasets in Explore right now: Knowledge Capture, Team Publishing (Guide Enterprise only), Knowledge base, Search, Community. This article is the source for all the metrics and attributes in the different Guide datasets.
Supporting Documentation:
Metrics and attributes for Zendesk Guide
Q: We have around a dozen departments using the KB within Guide. We want to share data with department leaders on how their teams are interacting with the Guide KB. Is it possible to pull data in Explore by agent role or group? Right now our "role" filters are only agent or admin. Do you have any other recommendations on how to accomplish our goal?
A: If you have a relatively static group of agents, you group your agents (or any other attribute) with advanced groups. You'll have to do this for each dataset in Explore but the number of datasets depends on how many features you use and your leadership values. We rely a lot on the KC app dataset, Knowledge Base and TP.
Supporting Documentation:
Organizing values by groups and sets
Q: Many agents haven't specialized in content writing. How do you at Zendesk make sure the content's still good quality whilst meaningfully connecting them with the process of creating/updating content?
A: Most of our agents haven't specialized in content writing either. We encourage them to share their knowledge in a very raw form and to not bother so much about the package. Because we rely on Team Publishing, the content that they share isn't public immediately. We have Knowledge Champions, agents who are specialized in writing, who then review the content and share it as soon as possible..
Supporting Documentation:
Zendesk on Zendesk: How we do KCS
Q: I would like more insight into Answer Bot.
A: Here is our comprehensive article on Answer Bot.
Supporting Documentation:
Q: How do you manage images and other embedded content in articles?
A: Great question. Never copy and paste images into articles, use the upload image feature in the article editor. Otherwise, you may see some broken images. Good quality images but not too heavy. Use the alt_text tags in the article source and give them an accurate file name (searchability and accessibility reasons). Things become a bit more difficult if you localize your content. We don't localize images, and because of that, we need to make sure that each image is supported by translatable text. Finally, soon images will be available in content blocks, and that will help image management too.
Supporting Documentation:
Inline images in Guide articles display as broken across all browsers
Q: What key metrics do you use for Self-Service and Knowledge Management?
A: From Explore: Views, Links, Flags, and Created articles, Searches and Searches with no results, Articles edited, Articles published, Articles reviewed (Articles archived for monitoring reasons). We don't rely on article verifications and assignments, but those could be good data points too. Because we combine our workflows between Support and Guide, and because we use macros for our workflows, we can also use the Tickets datasets to understand how our Knowledge Champions are working. We also have some custom metrics that allow us to look at other things. For example, the number of users who interact with the Knowledge feature is a good one to understand if we are succeeding at improving participation,
From Guide Articles: There are too many possibilities of reporting through Guide Articles, and much depends on your Guide Article implementation, but generally, we look at sessions, time on page, exits, and new users.
From video host: Plays, for how long they watch the video, etc.
From Google Console: Clicks, Impressions and AVG position".
Supporting Documentation:
Using the metrics that matter to improve your knowledge base
Q: How do you analyze those metrics to ultimately understand if you are deflecting tickets?
A: Firstly, we look at tickets. We usually separate buckets of tickets per topic (a simple About field will help you with that). After targetting a specific topic with new content or updating existing content, we look at trends on the tickets of that topic. If they go down, we are probably being successful. Then, while it’s not a complete story, we frequently monitor utilization like pageviews of articles to understand what our customers are looking for. We also monitor search terms, the links between our Agents’ tickets and articles that are created through use of the Knowledge Capture app.
At the highest level we at least recommend measuring your self-service ratio (also referred to as a Self-Service Score). It’s a simple calculation of the number of pageviews within a time period divided by the number of tickets you received in the same period.
Supporting Documentation:
Using the metrics that matter to improve your knowledge base
Q: How do you communicate impact of smaller changes (as in impact on an individual-article basis) to show you are moving the needle on a smaller scale (not just overall)?
A: That's a very good question because KCS is also about recognition. We try to make sure the agent's manager know about the article that the agent created and then share it in Slack. After some time has passed, we look at how many views the article had, how many times it was reused by other agents and how many times was used by the AB to solve tickets. If the agent has more than one article authored, we can even have a ranking of best performing articles. With Explore, we can create scorecards that show how each agent is doing content-wise.
Q: What are the best practices for using Knowledge Capture app in Support?
A: We actually recommend moving from the Knowledge Capture app to Knowledge in the context panel. This article goes over the similarities and differences.
Supporting Documentation:
About Knowledge in the context panel and the Knowledge Capture app
Q: What's the best way to manage information to make it easier to onboard new support reps?
A: Plain and bite sized knowledge articles with assessments during onboarding. A good taxonomy, good article flow and connectivity are critical. Agents don't need to remember everything, they just need to find it and follow it. We have an iKB with lots of short articles on how to write for KCS, we send slack videos and reminders, we post internally celebrating excellent practices of KCS practicedinternally.
Supporting Documentation:
About Knowledge in the context panel and the Knowledge Capture app
Q: What mistakes have you made while implementing the knowledge management, and how did you correct them?
A: I personally have failed to give reporting the time it needed and I've learned to give this the weight it deserves.
Q: Do you have any tips for managing complex troubleshooting with the Knowledge Articles?
A: Text, videos, and interactive tools like tabs and accordions. All are available on the Zendesk Garden. Cater for many learning styles.
Supporting Documentation:
Q: One of the guiding principles of KCS is that agents should be drafting articles as they work tickets - updating as they go every time they gather more information. How do you do that at Zendesk?
A: Some of our agents are only trained to flag articles or suggest new articles for someone else to write. While other more advanced agents (who often have more flexibility in their day) are able to draft articles and also directly fix articles themselves.
Resources from the event
Presentation Material:
2 Comments
Hi!
In the intro of the video you say that you would post the link to:
Can't find them, or will they come with the email that is mentioned?
Also, these links are broken at the bottom of this article
Hello Kalle Windefalk,
Apologies for those broken links. They should all be in working order, and you can find our past community events in our Past Community Event Playlist.
Please sign in to leave a comment.