How Quora and StackOverflow scale.
A lot of Q&A communities have been mushrooming lately. Why the sudden surge in Q&A communities?
Q&A dates back to the early internet days when forums abounded. Forums are based on an outdated model which suffers from a lack of identity of participants and results in trolling and noise (and hence, poor navigation and discoverability of the right answer). For a long time, Yahoo Answers was the best thing when it came to Q&A, which got a lot of things right including the reputation platform though it continued to suffer from the identity issue. Now that Facebook and FB Connect have solved the problem of identity and single sign-on across the web, several Q&A sites have emerged based on real identity and have made fundamental improvements in the model.
Typically, in two-sided networks like this one, one side is much easier to get than the other. In this case, it is easier to get people to ask questions than have people answer them. So the question boils down to:
This question actually involves five parts:
Kickass Discoverability: This is where a platform like Quora or StackOverflow is SO different from the traditional forum model. They do an incredible job with SEO. Every question has its unique, juicy, keyword-intensive permalink. It is an amazing model to harvest long tail demand from web search.
Piggybacking on other networks: Tweet it to get the word out. Yes, we’ve all heard that people in your social graph are not the best to answer your questions because they aren’t necessarily experts. But many questions do not require experts, they just need a discussion, a free-for-all opinion throw-in.
Incentives: Any participation site can perform better with the right incentives. In a Q&A site, the incentives are largely of 3 types:
A – Greater recognition in the community: Even forums got this right, and so did Y! Answers.
B – Greater megaphone-power: What’s that? The ability to be heard more if you participate more. Quora does this through credits. The more you participate, the more votes you can potentially get and the more credits you have to spread the word in the community about future posts and answers.
C – Monetary: In my mind, the least effective unless structured well.
All this sounds very intuitive, but companies like LinkedIn have got it wrong. LinkedIn had a problem of misaligned incentives because of which its Answers product is so sub-optimal. LinkedIn is a destination for people to promote themselves, not necessarily create a body of knowledge. As a result, all posts end up being self-promotional until the only people left posting are the self-promoters themselves.
Again, Quora has some really good tactics implemented to make the match-making work.
Pull: The person who asks the question is given suggestions for potential ‘experts’ based on their past track record at answering. The ‘expert’ algorithm could, of course, depend on many things including answers on topics, community reception of those answers, etc.
Push: This has to be a multi-channel tactic. News feed based on topics a user follows, related questions on a question the user is browsing as well as tactics to maximize return usage like a daily biggest mail of the most relevant topics based on the user’s interest, have all proven to be successful for various Q&A sites.
Quora and StackOverflow solve this brilliantly by combining discovery and creation workflows. Users on Quora or StackOverflow looking for a question are offered several recommendations of similar questions. If none of the recommendations seem relevant, the user is offered an option to “Add a Question”. This becomes a seed to increase production on the platform.
Fake it till you make it: Initial seeding of community is almost always editorial. Editorial seeding prepares the way for the crowd. Reddit did it posting links through fake accounts. Quora did it with moderators asking and answering questions. This is typically what it takes in the early dark days of no Users.
Focus on a micro-universe: This is probably the fundamental principle of creating any two-sided network. Facebook focused on Harvard; Foursquare focused on New York, and StackOverflow focused on programmers. In fact, StackOverflow’s growth and moving on to the cooking category from the programming category is a great case study. Apparently, programmers love to cook a lot!
While a lot of the points above apply to Q&A in particular, almost all of them apply to any two-sided network based around user-generated content.
Why the future of SAAS isn’t SAAS, it’s network effects.
Interaction failure and multihoming costs will determine winners in the war for the next big platform