Max Ventilla - head zookeeper, Aardvark
Brynn Evans — digital anthropologist. Bolt | Peters
Scott Prindle - Crispin Porter
Ash Rust — OneRiot. real time search engine. director of search relevance.
Brynn Evans
* Not thinking about search just as a question in a box
* What happens when you just have a question? How can our friends answer that question?
* Search happens over time. Interactions with friends at any point in your search task timeline
* 3 flavors:
** collective — gathering trends from a crowd
** friend-filtered — how it sounds. looking for info your friends have shared.
** collaborative — working with someone to answer a question, or asking a friend a question.
* Two main strategies:
** Ask the network — Certain people actually scared of Google — but feel comfortable asking their friends
** Embark alone — some people WANT to try alone. if they can’t find an answer, they might turn to friends.
* Need to design for both strategies
Max Ventilla
* web search is great for objective questions, but subjective questions generate the majority of search revenues
* friends can answer subjective questions, but…
** unreliable — small number of friends available to you in the moment
** hard to keep up with what your network knows about
** social cost of asking for a favor
* IM, email, mobile, twitter, web — utilizing the existing social networks for communication
* people want personalized responses to questions
* most content is still locked in peoples’ heads
* avg query length is 16 words (<3 on Google)
* What we learned:
** Intimacy (more than authority) facilitates trust
** Social context is different than social graph, and is frequently sufficient
** Speakers want to know who they are addressing
** People do not need artificial incentives to be helpful if there’s no friction involved
** People do not like receiving random questions but they don’t actually know what’s in their profiles
Ash Rust
* realtime search -> realtime advertising -> API
Scott Prindle
* Give customers something good to talk about in social media, and they will talk.
* …and that conversation becomes content for social search, helping to drive additional traffic and conversation.
* Enable customers to find the experts within your organization
Aardvark and OneRiot — two different approaches
* Ventilla: context is important. social should make a value add if you direct that question to the right person amongst a large group of possible answers
* Rust: what’s going on, right now? not just from a single source
Is google still relevant? Will social overtake?
* 20-40% of search results can be answered socially
* Evans: will NOT overtake. different use-cases for social search. google may give you the answer, but social search is complimentary.
* Rust: user authority is important to their rank algorithms. Brynn: do you index relationships? Rust: yes, to some degree, but how far do you go?
What questions did you have to get right to solve this problem?
* Brynn: to solve social search, you have to break it down: who is an authority? what is relevant? VERY difficult problems to solve. people naturally want to ask friends for help — don’t have to retrain them. Over half will prefer to ask friends.
* Ventilla: more and more sites will open up data — too much noise. Build a great UX first — saw how the user’s wanted to get data, THEN put the backend in place to service this desire.
How do you make information relevant to people? By topic, by previous interaction with the site?
* OneRiot: you tell us what you’re interested in and we give you everything around it — different approach. Also, if no one is talking about it, you’re not going to get much back.
* Over 50% of people say “yes” to the question: “would you like the sponsored answer?” wow.
* “Thrashing” —> seeing that people are slightly modifying their same queries over and over again. Perhaps inform them that they can ask their friends?