Towards the end of last year, my colleague Esteban announced the full roll out of our latest people search tool – “Faceted Search”. I’d like to give you an inside look at the process, principles, and decisions that went into designing that feature. Our overall goal was to create an inspiring and engaging experience on LinkedIn for finding people you already know, as well as discovering others you may find helpful.
I’d like to talk about a couple of key philosophies that drove our design approach on Faceted Search.
Data-driven approach
We strive to design based on a strong understanding of how people use the site. For example, we found that the average number of results per search on LinkedIn was well over 10,000, and a large portion of our searches were refinements on existing, broader searches. Looking at this data, we saw a huge opportunity to assist people in finding who they’re looking for, faster.
User-centered approach
We also conducted user interviews, lab tests, focus groups, and surveys to provide a strong picture of our users’ search habits and pain points when trying to find people on LinkedIn and elsewhere on the web.
These were a few of the initial observations we gleaned from those conversations. For e.g:
- People skip past navigation at the top of their page, in search of their results. Once they pass it, they don’t usually come back. Therefore we placed our refinements down the side of the page.
- Check boxes are more discoverable and easy to understand than links or other user interface selection elements because they provide a clear call to action, and enable easy and intuitive experimentation (including undo).
- Unchecked boxes better met peoples’ expectations of selecting what they want, as opposed to deselecting what they don’t.
- Users had a more intuitive and enjoyable experience when the results updated on the fly with each click, instead of being forced to click a submit button or wait for a page refresh.
This feedback then drove some core design principles for the project:
- Need for speed: The default experience now minimizes search calls, brings back facets faster
- A seamless user experience: We also don’t refresh a facet the user is currently interacting with, nor do we disable the in-use facet while refreshing results.
- Subtle user delight: Make sure the tool is discoverable, but not overwhelming (show a sample of functionality, with easy access to see more).
- Enable recognition instead of forcing recall: We introduced the ability to quickly create advanced searches directly from your search page last November. But we know it’s much easier for people to recognize what they’re looking for when it’s shown to them.
Facets not only help you filter your search results, but also create interesting “results” in themselves. They provide a summary of the search results, and bring forward the top ranking values for each facet, guiding further investigation and giving strong cues for further refinement. As you can see, we approached Faceted Search as an opportunity to enable a much more inspired and serendipitous exploration.
And finally, the value of people search lies in relationships that matter. So, we made “relationship” a prominent facet, allowing people to easily refine along many dimensions such as company, school or location, but then further narrow that down to people you know or could be easily introduced to (from your second degree of connections).
The response to faceted search has been tremendous, and we’ve seen significant increases in the number of searches by members since the release. We truly appreciate all the feedback we’re received from you both while testing faceted search as well as the user feedback you’ve provided after launch.
And we’re not done iterating. Please continue to send us your ideas and suggestions. There are three ways to do that: either as a comment on this post, through the feedback links at the bottom of the people search page, or @linkedin us on Twitter.
- Topics:
- Design
Comments
Sarah, thanks for giving these in-depth insights! I think you did a fantastic job. The details (like not taking away the fact the user just used, although 0 results) take so much effort to accomplish. Glad you could push it that far.
I am really surprised by your findings that checkboxes work better than links. First of all, links don’t help with multiple selection, but I thought they were the common denominator in faceted search which everybody understands and likes.
Linkedin search is POOR.
You should be able to use wildcards and partial names in company search.
Today if you type in “Acme” – you get get forced to select from list of some companies – but the list it too long – and doesn’t have all companies.
You should be able to type in Acme and get all hits from Acme Tool, Acme Computer, etc
If you select “Companies” from the drop down menu next to the search box in the header, you can can search across the over 500,000 companies in the LinkedIn directory. That should help you with searching for items like “Acme”. It will give you a list of all companies that match the query.
We do not support wildcard search (yet), but it’s something we’ve discussed internally. Appreciate the feedback that this would be useful for you.
Take care,
Adam
As a group moderator supporting a conference, I’ve been watching your development of search carefully. My greatest wish is for the Keywords free-form box in Advanced Search to support the Basic Boolean operator of searching for a two word phrase with either () or “”.
In my industry, most areas of interest — especially within Specialties on the Profile — are at least two-word phrases, if not three.
Is that on your development roadmap any time, soon.
I have to agree with the faceted design and the both the data driven and user centered approach, as it helped me recognize the features implement them to gain an increase in my connections by at least 35% in 2 weeks, Thank you Linkedin !
It would be really nice to have a function that allows you to search jobs posted in My Groups.
Sarah’s information in the faceted search was very informative. I like the changes, by the way.
What I would appreciate is someone taking the time to discuss beyond the “how” searches are performed from a user interface perspective, into “why” search results are what they are. I understand from articles I have read that relationship relevance weighs stronger in search results, versus just the level of connection to the person conducting the search. My question is “What factors are considered ‘relevant’ and given what level of priority in the search routines”?
For example, my girlfriend and I live in the same area, belong to the same industry, share more than half of our connections, went to the same schools, worked for the same employers, and belong to virtually all the same groups. However, if I do a data-driven keyword search for something like her job title, her profile ranks 23rd among the search results. I think I would be hard pressed to find anyone else on LinkedIn with more relationship relevance than her, so what other factors are coming into play here?
I really like and appreciate the new search features. It’s enabled me as a premium user to create some useful saved searches.
What would make it better? Regular Expressions baby. Let’s get technical.
David Sealey
The ability to refine searches is awesome!
One addition that would make it even better if I can refine by tags as well: you allow us to tag our connections, but not search by those tags today.
My current use case for this is to look for all my connections that are headhunters/recruiters.
Thanks. Usefull info.
Thank you, it was really intersting to read.
Search would be much better if you could filter/exclude terms. For instance, when job searching I use VIDEO as a keyword – sadly, my results are usually cluttered with dozens of jobs at Google and YouTube. If I could use Boolean operators like [Video NOT Google], my results would be cleaner and more useful.
@Bruce- We do allow you to exclusions in our search. This information on our Learning Center should help you out. http://bit.ly/aQeuAI
Derek!
Got it – perfect – many thanks!
Sarah, great job.Linkedin has really good search options, better than any other social networking site.
Hi Sarah,
In Advanced people search, I would really love to see more options for the Company option. Say I want to find people who worked in company A in the past and are working in company B in the present, I can’t search for this with the current Advanced people search. Would love to see some improvement to this feature.
Regards,
Kamala
Interesting post!
You mentioned lab tests — can you tell us more? Did these include any public-facing A/B tests?
–Gabe
That was godd informa grasiaz.
Linkedin is a good search option and is more effective than any other social networking site.I can’t believe it’s ability to refine searches!This is really a fresh search feature!Thank you for this useful information!
This is really the high tech age!No doubt about that. It’s ability and capability to give good search options is amazing! This post is really useful, thank you.
The fact that it updates as you select or deselect is fabulous. It makes for an enjoyable and efficient search. I can focus on getting the results I want rather than focusing on pressing buttons.
Well done!
Stephen Hart
I love you site