How HelpHub search works
HelpHub’s goal is to show users relevant content whenever they open it up, but it is also designed to be a natural language entry point into all of your content.
This doc explains how HelpHub’s search works across different types of content.
Out-of-the-box, HelpHub uses natural language, or “semantic” search (for the big nerds out there). That means when users search for something, HelpHub will show results that match the meaning of what they searched, not just the words themselves.
For example, let’s say a user searches for “how to add seats” but you don’t use the word “seats” anywhere in your docs. HelpHub is smart enough to know that “seats” refers to the concept of adding users, so will return any results that relate to adding users, upgrading your plan to add more users, etc.
When you add content to HelpHub, CommandBar automatically breaks up your content into passages (roughly speaking, paragraphs). This allows our search algorithm to look for the passages that are most relevant to your users’ query. The top document result in a HelpHub search will be the document that contains the passage that is most relevant to the user’s query.
HelpHub searches over videos too. What does that even mean? When you add a video to CommandBar, the platform automatically transcribes the video. That transcript is then available for HelpHub to search, like any other written content.
If HelpHub finds a section of a video that is relevant to a user’s query, it will return that result in the response list to the user. To make sure the user gets the information they need, HelpHub will begin playing video results at the point in the video where the relevant information begins.
We have found that this approach makes video content much more useful to users. Instead of having to sit through a full 4-minute video to get to the 20-second clip that is relevant to them, users can begin watching at the highest-signal point in the video.