Have you ever found yourself searching through countless help center pages for a tiny snippet of information to navigate an app?
If so, you'll be pleased to know that those days might soon be over.
Generative AI (GenAI) is no longer just for chatbots. It's changing the way we interact with knowledge bases too, by streamlining the search experience and making it easier to find information across various sections and articles.
Zendesk Guide is a leading player in knowledge base management software, and Zendesk recently announced the release of Generative Search for articles within both the help center and Agent Workspace.
In this article, we'll take a closer look at what this release offers, what you should consider if you decide to use Zendesk Generative Search, and how to get the most out of it.
What exactly is Generative Search?
Generative Search is an AI-powered tool that searches knowledge base content and provides direct answers to your search queries. Instead of sifting through a plethora of search results and clicking through each article, you can now receive quick, precise answers almost instantly.
Before this, searching for information meant guessing the right keywords and browsing through lists of search results to find the answers you need.
With generative search, you can find answers faster and more efficiently based on the same help center content.
No more keyword guessing or endless scrolling required.
What does Zendesk’s Generative Search do?
Zendesk’s Generative Search leverages OpenAI’s technology to understand search queries and provide searchers with the most relevant information from your knowledge base — whether the searcher is an end user looking for help or a customer support agent trying to assist a customer.
In short, it means quicker access to the right information, delivering a quality of life improvement to both customer and agent experience.
Zendesk Generative Search for your help center
Zendesk Generative Search unlocks a new way to leverage your Zendesk help center.
It allows logged-in users to receive an AI-generated “quick answer” to questions entered in the search field.
It delivers this quick answer above the list of traditional help center search results, making finding information faster and more efficient and enhancing the overall user experience.
Derived from the top search result in your help center, these quick answers show users the information they need directly within the search results – without requiring them to open each link and scan through the articles. And if more information is needed, users can still click through to the full articles.
By leveraging Zendesk Generative Search, you basically gain a bunch of the advantages of generative AI – 24/7 assistance, reduced support volume, and cost savings – without the need for a complex chatbot setup.
Zendesk Generative Search for the Agent Workspace
Internal users, like your customer service agents, also benefit from this feature release.
When support reps need to search through extensive documentation to assist a customer, it takes your team longer to reply and resolve issues. That means they’re capable of helping fewer customers, and that each customer has to wait longer periods of time.
Zendesk Generative Search helps solve this by surfacing the most relevant information at the top of the search results within Zendesk’s Knowledge side panel.
Agents can click through to the source of the answer to read the original content. They can also quickly copy and paste the AI generated answers into their customer replies, further speeding up the support process.
While it’s too early to have hard data on the impact, it makes sense that having this feature embedded directly within the Agent Workspace should unlock a boost in ticket response and resolution times as agents get what they need much faster.
Things to consider with Zendesk Generative Search
While Zendesk Generative Search offers exciting possibilities, there are a few considerations you should be aware of before diving in:
1. Beta version hiccups
Generative AI models are still in their early stages, and Zendesk Generative Search is currently available only through their Early Access Program. When something is in a pilot or beta mode, It doesn’t always provide relevant results. It might flatout fail to work at all at times.
Many of the issues are likely to be ironed out as the model improves and Zendesk iterates on the feature, but jumping on this train too early increases the risk of mistakes.
2. Accuracy of information
Responses generated by AI may not always be 100% accurate or relevant. While your support agents can proofread and fact check any Generative Search answers within the Agent Workspace, customers using Generative Search in your help center don’t have the safety net.
3. Availability and access
Currently, Generative AI results in the help center are only available for logged-in users during the Early Access Program (EAP). The Zendesk team is reviewing the pricing structure and may consider extending this feature to non-authenticated users in the future, but this certainly isn’t guaranteed given OpenAI’s pricing.
By keeping these considerations in mind, you can better navigate the early stages of Zendesk Generative Search while deciding if it’s right for your team and your customers.
How to get the best results with Zendesk Generative Search
Simply enabling Generative Search isn’t enough to guarantee success. You need to optimize your content and equip the AI model with robust resources to achieve the best performance.
Here are key strategies to maximize the benefits of Zendesk Generative Search for your organization.
1. Optimize your knowledge base
The best way to ensure quality AI answers is to maintain a well-structured, up-to-date knowledge base.
Zendesk Generative Search relies on the content of your help center to produce answers, so the quality of the answers can only be as good as the quality of your content.
The answers are based on the top three results returned by the help center search, so optimizing your content for the general search experience will significantly improve the generative search results as well. Also, consider structuring your help content for generative AI chatbots. Key recommendations include:
Clear titles. State the product and feature name clearly in the title and word title as closely to the user question as possible. For example, “How do I add additional team members to my Slack account?” will likely perform better than “Adding seats”.
One article, one message: Avoid long reads covering multiple themes and ideas. Instead, split them into separate articles grouped into one help center category or section. This makes navigation easier for both humans and AI. For example, a long list of FAQs can be separated into individual articles grouped under the same FAQ category in your help center.
Main idea in the first paragraph: Ensure that articles start with the main point, instead of using personal stories or illustrations. While we can’t say exactly how Zendesk’s AI model works, AI models typically look at the first ~100 words to identify the key idea of an article.
Following these tips will significantly improve navigation around your help center and search relevancy, positively impacting the results you’ll see with Zendesk’s Generative Search.
2. Review no-results queries
If there are no existing articles or posts matching a search query, an answer obviously will not be generated. Flag these queries to your knowledge base team so they can create new content or adjust terms, tags, and labels on existing articles to improve search visibility and relevance.
Tools like Help Center Analytics allow you to proactively review no-results queries, enabling you to see what users are searching for without waiting for someone from your team to uncover a no-results search query.
3. Provide feedback for generated results
Feedback is essential for improving the accuracy and relevance of AI-generated responses. You can and should upvote or downvote answers to provide feedback and enhance the quality of answers produced by the generative search over time.
To indicate a helpful answer, click on the thumbs-up icon. Similarly, click the thumbs-down icon to indicate a poor quality answer.
Zendesk will then prompt you to select the reason for your rating, which will also contribute to making answers more relevant over time.
4. Avoid one-word search queries
For the most accurate results, structure your search as a phrase, sentence, or question, and advise your agents and customers to do the same.
If there isn’t enough information for Zendesk’s AI to generate a quick answer, you'll receive instructions to rephrase your search. For example, the search query "How do I reset my password?" is more likely to generate an answer than the single term "reset".
How to enable Zendesk’s Generative Search
Generative search for both Agent Workspace and your Zendesk help center are not available by default, as these features are brand new and have not been rolled out to all Zendesk workspaces yet.
To get early access and activate Generative Search on your Zendesk account, you first need to sign up for the Zendesk Early Access Program (EAP) using this form.
Once the EAP is activated for your account, you can enable generative search for the Agent Workspace and help center by following these steps.
Activating Generative Search for Agent Workspace
Navigate to your Admin Center
Click on Workspaces in the sidebar and go to Agent tools > Context panel.
In the Knowledge section, locate the Quick answers section and tick the box to Show a quick answer before search results. Then, click Save.
Activating Generative Search for help center
In Guide admin, click the Settings cogwheel icon in the sidebar and go to Search settings.
Click Manage on the Quick answers and check the Show quick answers for search queries box. Then, click Save.
If you’re using the Copenhagen theme or a custom theme, you’ll also need to take a few extra steps.
What's Next for Zendesk’s Generative Search?
With Generative Search, navigating through help center content becomes more intuitive and less time-consuming. It’s likely going to transform the way we search for information, making the process faster, more efficient, and user-friendly.
Of course, it’s still early days for Zendesk Generative Search, and you can expect improvements such as better reporting, options to remove specific articles from the generated results, and more. The Zendesk team encourages early access users to share feedback and report bugs about Generative Search for Agent Workspace and the help center via their community platform, which will help prioritize future improvements.
As you experiment with these features and work to enhance the underlying content for generative search, you may need tools to streamline your help center workflow.
Swifteq's Help Center apps, specifically designed for teams running on Zendesk, can help with a range of tasks to optimize your Zendesk help center – from managing articles to AI-powered translation and comprehensive analytics.
Book a free demo with us today to see how it works!
Written by Maryna Paryvai Maryna is a results-driven CX executive passionate about efficient and human-centric customer support. She firmly believes that exceptional customer experiences lie at the heart of every successful business. |