by Mostafa Hossam

1) Product Case

Improving LinkedIn’s Search Results

Problem and Discovery:

Recently, I got an interview in the field of identity and access management, which is a field in cybersecurity. I was interested to read what people are posting about this field and try to learn more about it and find career related tips, so I typed ‘identity access management’ in the search bar and chose to see ‘posts’. Here are the problems that I faced:

• I got spammed by the same post 12 times in the first results page, but this issue got fixed once I wrote ‘identity and access management’ instead of ‘identity access management’.

• posts can be grouped into: people announcing they are starting a new job or leaving a Job in this field or gained a certificate in this field, people sharing a job opportunity available in this field, company updates in the field/advertising their product, sharing informative articles (what is identity and access management, terminologies, strategy) and discussion topics I was most interested in the last group, but had to skim through a lot of posts to identity which posts I really am interested in.
• I wanted to try and sort posts by most interacted with (comments/likes) to get to the most interesting or valuable posts, but this feature wasn’t available
• Using hashtags to filter for the topic was also a bit confusing, since there are many hashtags about the same topic and you can’t filter for multiple hashtags at once. Every page for a hashtag has then the same problems mentioned above

Should we solve this problem?

LinkedIn is probably interested to make users spend more time as possible on their website, as this would let more companies use their website as a platform for advertising and more talents being available on the website which will give also a boost for their talent finding solutions. So making users happier or enhancing their experience in a way that will make them more engaged on the website will help the business overall.

It would be interesting to test first how many users actually use the search bar in general to filter for posts and hashtags and if it is currently worth working on its improvement or work on something else would have more impact. It is also important to identify what they are looking for when they are searching. For example, people might use hashtags when looking for a certain career event that is happening or a trending hashtag, but not for general technical topics. Also most people might be interested in job offerings when searching for posts and not technical stuff, as they would use Google to know more about technical stuff. Doing a couple of user interviews/surveys would be helpful here to know more about peoples intentions when using the search bar for posts and for hashtags, which will help us what solution would have the most benefit for a large variety of users.

What can/should we build to solve the problems mentioned above?

1. allow searching for multiple hashtags → easy to build, low risk, low impact

2. group similar hashtags → complicated to build and high risk by irrelevant grouping which can create confusion, low/medium impact (searching for or following multiple hashtags could also accomplish same search results and is easier to implement )

3. search for key words instead of an identical phrase or work on diversifying the first posts that appear in search results, like the problem by ‘identity and access management’ and ‘identity access management’ → medium difficulty to build, medium risk, medium impact (could improve search results, but could not be directly noticeable by users, also by this problem didn’t occur by searching for other topics like ‘product data analysis’, could be an edge case that I have just found)

4. allow to sort results by number of likes/comments/views → easy to build, medium risk, low/medium impact (probably already implemented in hashtag results by choosing ‘top’ instead of ‘latest’, could have also a positive impact on search results, but could also have a side effect of boosting already seen top posts and undermining new ones)

5. categorize posts using key words. Categories could be (career update, job posting, promoting product, opinion, informative, …) → difficult to build, low/medium risk, high impact (could improve search results with more relative posts for the user, could also allow user to pick categories he is interested in for a specific hashtag improving his newsfeed experience, could in general give more insights about users interests to provide him with more relevant content)