SXSW Live: Social Search: A Little Help From My Friends
Though our wide-ranging crepe-based discussion panel on Saturday morning was a highlight, I’m seated in the cavernous ballroom A for Brynn “No relation to Will or Bob” Evans is sitting on this panel, so I really wanted to hear it.
Brynn is kicking things off. She is concerned with the interaction design of search and how to integrate social networks and search.
We want to frame this problem as not making search as relying on google, but how can our friends help us? You have to think of search as a process over time. If you can reframe how you think of search is done, you could make use of friends at any time during that path, it’s not just google.
There is no one definition of social search but three distinct types:
1) collective
gathering trends from a crowd. Looking at many people’s activities and how can we make sense of that
2) friend-filtered
3) Collaborative
Like aardvark does.
What are people’s social strategies? How do people want to interact with friends when they search?
Two main strategies: ask the network, or embark alone. People ask their networks first, they’re kind of scared of venturing into google without help from their network.
People sometimes want to embark alone on google first, can’t find what they want or get confused, and then go back to their social networks for help.
Now it’s Max Ventilla from Aardvark.
They started Aardvard from a opp they saw in queries — subjective questions about how people spend their time and money — this is the biggest source of searches.
People are still turning to friends and co-workers for info, but it’s generally unreliable. It’s hard to keep up with what they know about, and unreliable to tap all your friends. There’s also a social cost to it. Generally, you don’t want to bother someone by asking.
At Aardvark, you ask a question, Aardvark identifies the people in your network that can best answer that question, and connects you to them. It connects you with information that isn’t published anywhere.
Social intimacy makes information actionable. They now answer 85% of questions asked to the system in the last year. 45% of questions lead to crosstalk.
About 50% of users answer a question at some point. This is incredible participation rate.
Intimacy is more valuable than trust
Social context is more important than social graph
Speakers know who they’re addressing
You don’t need a leaderboard or metric counts — people continue to participate regardless.
Ash Rust now up from OneRiot — ranks and delivers search results from social networks
Realtime search – they help people find what people are talking about now
Realtime ads
Realtime API – to make it movable
Now Scott Pridle from CPB who leads innovation talking about the money side of the equation.
Everything that CPB does factors in social. Now a few cases showing work that uses aardvark and one riot models.
Give customers something good to talk about in social media and they will talk.
Right now with Old Navy, you go online and create a mannequin like yourself, get people to vote on it. The winner gets 100,000 and a mannequin that looks like them.
What happens is that people pass it on Twitter, then picked up by One Riot. If you search OR for Old Navy, you won’t get the store site, you’ll see all the conversation. And that seeds the virality. Also has a facebook app/page.
Where it related to aarvark model, large brands are trying to figure out how they talk to people. With Best Buy, they made a platform for in-store experts to communicate with people. There was an opportunoity to pull all 2000 blueshirts together and give people responses customized, and quickly.
It used to be that agencies would create TV and then create interactive around that. With Best Buy, it was create the interactive platform first and then develop TV around that ability as a brand value add with consumers — even if they’re not on Twitter. They using it as social proof.
Now going to discussion time….
Brynn going after Rust to try and understand how they look at social graphs, relevance, and their methodologies to evaluate value, and he did a good job of stepping around it. But he did say that they’re breaking it down into smaller groups of value to try and return value.
Brynn has found that it’s not about retraining people on querying, but about how to include their friends. They pick the experts in their community to turn to. But how do you index that? How do you return good information?
Related posts:
