Improving Member Productivity with In-Product Help
December 27, 2018
LinkedIn’s Help Center boasts a personalized experience aimed at getting our members the help they need, finding the answers to their questions, and empowering them to be more productive and successful.
Despite having a wealth of customized content, members are still faced with a problem when they need help: they must stop what they’re doing and go to the Help Center to find an answer. To address this problem, we gave ourselves a challenge: Can we bring the Help Center to members instead of bringing members to the Help Center?
More specifically, we asked ourselves these questions:
Can we show the right content based on context? For instance, if a member is viewing their feed, can we help them post their best content?
Can we avoid disrupting a member's flow? If possible, can we avoid taking a member out of context to go to the Help Center?
Can we make Help Center content easily searchable from within other LinkedIn applications?
By leveraging the Help Center's pool of content and combining it with faceted search, we built the answer to these questions into a service called "In-Product Help."
Using Search as a Service
Help Center has a powerful search capacity built on top of Galene, our Search-as-a-Service (SeaS) infrastructure. Additionally, the article content present in the Help Center is classified according to which product it is for. For example, an article may be relevant to both "LinkedIn" as well as "Recruiter." By combining search with these classifications, we can surface articles specific to a product. However, we wanted to take this further and display help content specific to the task a member may be performing.
To facilitate this goal, we form a “help context” by combining a "product" and a "topic." For example, if you're viewing the LinkedIn feed, the help context would be (linkedin,feed); if you were using the search functionality inside of Recruiter, the context would be (recruiter,search). Since Galene supports faceted searching, we use the help context as a facet and display very specific, targeted content.
This approach allows each application to define what its help contexts are. Our content team can tag articles for different products in different contexts, which allows for great flexibility. Help Center search supports live updates, so as soon as an article is updated, so is In-Product Help.
In-Product Help widget
We developed a drop-in widget that enables the In-Product Help feature for any frontend application at LinkedIn regardless of whether it is implemented on our Dust or Ember stack.
The widget is built on top of many high-quality open source projects and consists of a small client library and a UI layer—all written in TypeScript. For Ember clients, it is packaged as an Ember Addon and leverages ember-cli-typescript for TypeScript integration.
Packaging the widget in an Ember Addon allows us to share the client library and the UI code easily. It really is little more than a drop-in! Embedding the widget into an Ember application looks like this:
The topic can be dynamic and is defined by the application that is rendering the widget. This allows application developers to set the topic contextually—for instance, if a certain view is visible or a particular feature is being used.
Our first target integration for In-Product Help was the Flagship desktop experience, which is Ember-based. At present, we have ramped this feature to all members who have their language set to English.
Future rollouts include extending the feature to other languages and to native mobile platforms, as well as to other LinkedIn products such as Recruiter and Sales Navigator.
In-Product Help is a simple solution to a difficult problem. The widget that we built allows members to do all these things without leaving an application:
- Get contextual help content instantly.
- View summarized, "bite-sized" help articles.
- Rate an article's helpfulness.
- Perform a custom text-based search for help content.
Additionally, we have built In-Product Help with scale in mind, such that it can be adopted for use across the LinkedIn ecosystem. Moving forward, we will continue to work with partner teams to integrate In-Product Help into their products, in addition to adding international support.
In-Product Help is the fruit of many people’s labor. Special thanks to Chris White, Stephanie Lucas, James Gatenby, Sanmin Liu, Cindi Jordan, Joe Villafuerte, Aris Harutyunyan, and Zhou Jin.