A few months ago I laid a project to rest. It was part of a long list of things that I consider experiments – some of which have done well (like this blog) and many which have not (too many to list). Like any good experiment, I learned some valuable things. This is the postmortem of an idea. I hope you can learn something from it too.
It has been over a year since I first encountered Yahoo Answers. It's a nice service as long as you don't need to ask anything important. I played around with it, but quickly became tired of the trash I had to sift through. There were too many questions like "what do you peeps think of my avatar" and "which chick reading this is the hottest?" Give me a break. Time is the only thing in life I don't have enough of, and I have to waste it wading through this kind of junk? So like any good entrepreneur, I thought about ways to improve it.
During my ruminations on the Q&A process, I stumbled across what I thought was a really good idea – make the Q&A nearly instantaneous. Think about it this way… you probably have friends on IM that you are experts in certain fields, right? I have technical friends that I will shoot of messages to like "do you have a chunk of code that does _____?" It's a quick and easy way to search. But often those friends aren't available. Well, what if I could build a really really large network, and just throw that question out into the void? Could software algorithms determine the best place to route it based on knowledge profiles of the people currently connected to the network? If so, then in theory I could get an answer in just a couple of minutes. And I would pay for that. Think about it this way. I don't know much about databases, so I could spend two hours debugging a problem, or I could find a database expert who could do knew the answer immediately. Isn't that something worth paying for?
I told the idea to some people, who immediately said that's what forums are for. Well, have you ever tried to get help on open source software in a forum? You have about a 10% chance of getting your question answered.
…we realized we had no early incentive, so we had to come up with something. To do so, we played out possible future scenarios of how the business would evolve, and looked for places to give ownership to users. Users love owning things. We let them own an answer. We thought about the type of questions that would be asked, and realized that some questions would have direct, simple answers that would be unlikely to change. For example, someone may ask "when is Britney Spears' birthday?" That has a simple answer, and if our service grew popular, it may be asked hundreds, possibly thousands of times a year. There is no need to connect the asker to a different answerer every time that question is asked – we could just kick it out of a database. So if the asker paid $1.00 for that answer, we don't need to pay the $1.00 to a brand new answerer. Then the idea came – let's pay the person that put the answer in the database. So if someone asks a question and gets back a good answer, we dump that answer into a database. Every time the question is asked, we kick the answer out of the database and paid the answerer a a percentage of the question price. It's a residual. Answer the question once, get paid indefinitely.
We used this model and returned to the asker side of the equation. We could give our answerers incentives – but what about askers? Why would an asker pay to use our site instead of the competition? For one, we hoped that we could provide higher quality answers for certain types of questions. Back to my question about top ten Taiwanese OEMs – you can't get that answered on Yahoo, so it might be worth paying for. In addition, we hoped that people would pay more for near real-time answers. Consumers always pay more for convenience. But we needed more. So we decided to extend our residual policy to the askers. If you ask a question, and it gets answered, then ever time it is asked again, the original asker gets 30% of the fee and the answerer gets 70%. The pitch would be that if you asked your question on our site, you may make money on it eventually.
Now we had our model. We had a reason for askers to use our site. We had a reason for answerers to user out site. The next issue to tackle was marketing. How do we make them aware of it? We decided to use blogs. What better way to expand than to piggyback on an existing network? But back up a second – regular readers will know that I don't like the viral-marketing-by-blogs strategy. It rarely works. Everyone wants to do it, but it isn't easy to get bloggers to write about something. So we turned back to our ideas about web economics – we needed an incentive for bloggers to post the site on their blogs.
We had our programmers build a blog widget. It worked like this. I would go register at the Q&A site and I would specify the kinds of questions that should be directed to my blog. The questions would then scroll in the blog widget I added to my site, along with their reward fees. For example, if you went to the Q&A site and asked "how do I calculate Net Present Value" and you offered $3.25 for the answer, it might appear on Businesspundit.com in the Q&A window. A blog reader could then click on it, be taken to the page, offer an answer, and collect the fee. Then the blogger took a cut. But here's the best part – we could change the reward fee. Think about it. The answerer does not know what fee was offered, so we could bid it up slowly. For a question worth $3.00, we could first send it to a blog at $1.50. If no one answered in 30 seconds, we could send it to a set of different blogs at $1.75, then more blogs at $2.00, and so on until we hit the $3.00 maximum. That type of price discrimination would boost profit margins.
The next issue to tackle was how to get busy people to use it. That was where we came up with one of my favorite ideas. We built a little client app that could be set to filter questions. So if I'm busy, if I'm working on something important, I could set it to only show questions worth $15 or more, or only show questions about Warren Buffett, or whatever it is that I would be willing to stop work to answer. If I'm bored, or goofing off, show me everything that matches my knowledge profile.
The software moved forward, we got the plan in front of a few angel investors, and we even got some good ideas from the people we pitched. An employee at a major semiconductor company told me there were always looking for ways to connect with the successful entrepreneurs of the future, but they had no way of knowing who they were. She suggested that her employer may want to sponsor questions. If someone asks about a 8051 microcontroller, they don't pay because it gets filtered to an engineer at this company who would answer it in hopes of beginning to build a relationship. The company would pay us for the right to answer questions. I was ecstatic.
But the more we moved down the path, the more I realized the complexities involved with selling answers. Knowledge is a tricky thing to sell, because even experts disagree on some answers. What's worse, most people think they know more than they really do. Look at how many idiots think they know stocks, or programming, or even business. Nearly everyone thinks they can give good management tips. It is difficult to sell something so… confusing, and we realized it would lead to problems down the road. Yahoo, and most of the other sites, fix this by having people vote on the best answer, but we couldn't post answers in public because that would take away our residual incentives. And anyway, I'm not convinced in the "wisdom of crowds" for anything beyond general knowledge. It doesn't work for domain specific stuff.
We reached the point where we needed to pump in more money, and we did a tough analysis of the business. It was such a fun idea, and everyone who heard it was fascinated. But fascination doesn't always translate into cash, and I've done enough things in life that were fun, cool, and unprofitable. I began to realize that experts in most fields are usually pretty busy people, and with the exception of technical folks (who spend tremendous amounts of time in front of computers), most other people don't like sitting in front of one, and probably won't answer questions for a few measly bucks. It's a matter of incentives. If you make 100K a year and have a family, you probably wish for more time over more money – so is $7 really that enticing? Sure, some people would love this idea, but would millions?
I did some more market research, and in the end, with all the competition, the low barriers to entry, and the questionable economics, we scrapped the idea. It's better as an add-on to an existing business than as a stand-alone company. (We still have the code (in Rails), and if you are interested in it, my partners may be willing to let it go to the right person.)
The larger lesson here is that business is ultimately driven by economics, and economics are driven by incentives. Entrepreneurs, and in particular web entrepreneurs, are often driven by a myopia that says "if I like it, so will everyone else." That's what I thought for a while. Some people say go with your gut. Others say trust the numbers and the research. I say find an idea where the two line up.
I hope you can learn from the insight into my own thought processes throughout the project. I always seem to learn from reflecting on it. A key skill entrepreneurs need is the ability to distinguish between a temporary setback and the signs of an idea doomed to failure. I think we made the right decision.
Here's a list of the sites I'm aware of: Qunu, Guruza, Yahoo Answers, Live QnA, AnswerBag, Yedda, Wondir, Askville, Yedda, Helpshare, and I know there's more that I don't remember, but will add them as I think of them