I’m a designer on the Pages team at LinkedIn. Pages are the LinkedIn profiles of organizations such as companies and schools.
Problem & opportunity
How might we increase discoverability of relevant pages for LinkedIn members?
There are 2 sides to this problem:
- Members have a hard time discovering new content on LinkedIn. They are always trying to stay up to date in their industry by finding professionally relevant content. Pages play a crucial role in providing that type of content.
- Page admins want to find more ways to engage new followers. We have observed that members typically only follow well known brands. While smaller businesses account for 80% of LinkedIn pages, they find it exceedingly difficult to stand out.
The scope of this project was small, so we focused on designing one feature to test. After considering a few options, we decided on a drawer module on a page itself.
- Phase 1: 3 weeks for mobile designs
- Phase 2: If proven successful from test, desktop designs will be built
Members visiting pages (browsing content, work-related, job seeking)
- Increase in page follows
- Increase in page visits
I did light competitor analysis since this was a test and we didn’t have scope to do extensive research. I gathered screenshots of similar features on other platforms and noted down their characteristics/pros/cons. Some important factors I noticed:
- Choice of metadata displayed in the profile cards
- Flow — whether it’s permanently on the page or only appears through relevant action
- Relevance algorithm — how accurate are the suggestions, and how many should be shown at once?
Currently the only place members can discover new pages is on the My Network tab. Not only is it hidden under a tab, one must scroll very far down the page to reach the page suggestions (most of the page is focused on profile suggestions in your network).
I also posted a question on LinkedIn to ask my connections why they would follow a page. Although this is not a conclusive research method, it at least provided some leads on what information would be most important to members.
Based on comments, I gathered the following points:
May follow for:
- job seeking
- interesting or employee/culture/industry-related content
- reputation of the company
- company to show up in their Interest section of profile
Will not follow:
- pages they don’t support
From my findings, I defined design principles to ground my thinking:
Clear and informative: It should be clear why this is shown to the user. How can we surface the right information to pique user interest?
Contextual, not intrusive: The experience should be timely, relevant, and not disrupt from the user’s intent. Height and space should also be considered.
Design explorations: Initial brainstorming
Early on, my PM and I both agreed that showing the suggestions contextually based on the user’s action would be the least intrusive. We wanted to focus the test on pages themselves, instead of altering anything existing on the My Network tab (since another team has their own initiatives to improve on that experience).
I sketched out many different formats for how the feature could appear — here are just a few of many models I explored:
The content in the card will influence users’ decisions to visit or follow a page, so I sorted metadata based on initial assumption of importance, and looked at how the cards could be formatted:
After weighing the pros and cons of each design, we honed in on patterns we felt strongest about:
- Carousels: It can surface a lot of suggestions without taking up more space, and is a familiar pattern
- A full page modal: An immersive experience in the mind set of discovering
- Reusable component: Can this feature be applied with any LinkedIn entity such as profiles, groups and events? (at LinkedIn, we can’t all be making our own version of everything!)
- Cover image: It’s actually quite nice seeing the cover image paired with the logo! This may also encourage more admins to upload one on their page to attract new followers
Design explorations: High fidelity
Below is a snippet of some higher fidelity design explorations:
My thinking on a flexible card that can house different amounts of information. I wanted to ensure this component could be leveraged by other teams as well:
To touch on some of the reasons I gathered at the beginning on why members might follow pages: I explored personalized metadata we could show (such as hashtags, new updates and relevant jobs to the member):
After rounds of design iterations and design reviews, we narrowed down to the following two concepts:
Quick research: Intercepts
Intercepts are great for gut-checking your designs, especially if you’re looking for something quick. At the LinkedIn SF office, we’re fortunate to have a bustling public lobby space. I wrote up a brief of questions I wanted to ask, got a few coffee gift cards, and approached people for their opinions.
I asked 4 participants (mix of students and professionals) to play with each prototype. Besides first impressions, the two main questions I wanted to know were:
- What metadata is most important in the entity cards describing a Page?
- Which design did they prefer, if they had a preference?
Participants who preferred Concept A liked that it was a quick way to be shown suggestions upfront that they would’ve not otherwise discovered. The experience of scrolling horizontally was also familiar for them.
Participants who preferred Concept B liked the experience of browsing the content in a larger space. However, they would only click into the banner if they had intent to, and they suggested to have more sorting options to know what to focus on since they’re shown a lot more pages at once.
- People are open to being given suggestions contextually and trust that LinkedIn will surface the right content
- The right metadata intrigues members to engage with new pages they are not familiar with. One example brought up was network connections, that matters when looking for new pages to check out.
- People want the flexibility of being able to weave in and out of discovery experiences
Results from the intercept enabled us to confidently decide on a final design, which was actually a hybrid between the two. We kept the upfront carousel while providing an option to discover content in a larger space.
Entry point: Contextual to a user’s action, the suggestions only appear after a user has shown intention of following. LinkedIn can then surface other related pages the user may be interested in following
Full page browse: With this hybrid design, it allows for a lighter weight experience, while allowing those with further intent to immerse into a bigger browsing area.
Entity cards: The cards you see in the full page experience is an existing component we have at LinkedIn. We always try to align and use the same patterns as much as possible with other teams. However in this case, the cards are too tall and wouldn’t completely be above the fold on some smaller mobile devices. This is why we deviated and created a card shorter in height. Unfortunately this meant sacrificing the cover image, but that is something we can incorporate and test when building the desktop web version, which will have a lot more room.
Relevance: Another project was underway to improving our relevance, but in the meanwhile with our limitations, we wanted to make it clear how these suggestions were being generated. It’s why we titled it “Pages people also viewed”
Metadata: We decided to try A/B testing two variants to see which would perform better. Our hypothesis was that connections get more engagement, since we heard it from intercept. If this proved true, then it would also benefit smaller businesses to not have their number of followers be the defining factor of why someone should visit their page.
Accessibility & localization: At LinkedIn, we care a lot about inclusive design — our specs always consider colour-blind and screen reader users. And because our platform is used internationally, we always have to consider how text grows in long languages, right to left languages and any potential offensive icons.
After ramping the suggestion drawer, it drove +3.04% page follows and 0.29 million page follows per week. We communicate with our page admins regularly, and some expressed that they’ve noticed their pages getting more traffic! From our A/B test, the connections metadata performed slightly better than the number of followers metadata.
Later in the year, we designed and built the desktop version, which drove +2.6% SWI page follows and +2.14% SWI productive follows. Based on result of the A/B test we did on mobile, we stuck with connections as the metadata line.
The pattern is in progress of expanding to all entities to use on LinkedIn (profiles, groups, events, etc) as a way of surfacing relevant suggestions for members. This initiative is being carried on by another team
Overall, this project which thought to be small, ended up driving exciting impact for our admins, as well as creating simple way for members to discover new pages and content.