Tuesday, November 27, 2007

Only a couple of weeks ago...

I didn't realize I had posted only about a couple of weeks back. It seems a lot has happened in these two weeks.

Thanks to the success of some of our apps on Facebook, we ran into scaling issues the last few days. We went from one box setup to two and then to three very quickly. This was the one of those times when years of dealing with large scale problems at Amazon comes in handy. Above all, we were particularly pleased with the fact that our component based distributed system design allowed us to scale out with little headaches, and it gave us considerable flexibility in optimizing our systems.

Quite a few friends are now in the midst of their own startups, and when I share my lessons learned, I actually don't harp much on building scalable systems anymore. Product and monetization comes first, and I still believe if you need to build things to scale you'll be able to handle it. Just make sure you have the engineering talent for it. That said, if you are building applications on platforms like Facebook where, just maybe, your applications can ramp up to 100,000+ daily active users in a few days, you'd better plan for scaling upfront. I think, it would have been much more costly for us to scale had we not built our systems well the first time around. But even then, every such problem is unique, and you will have to handle new problems when you do run into scaling issues.

In the midst of this I managed to sneak away to Hawaii for a short trip. If you haven't been to the big island of Hawaii, I suggest you pack a bag and go. Snorkeling on the west side -- particularly the Cook monument -- was probably the best we've experienced in the state of Hawaii.

Tuesday, November 06, 2007

Facebook ads: More hype?

It's no secret we've been developing apps on Facebook for Jambool. And where the more involved apps have had a hard time taking off, the brain dead ones are rocketing up. In the last one week we added more than 50,000 users without spending a penny on advertising. Now there's got to be some value in that.

Facebook announced today that they are coming out with "Facebook Ads." Techrunch covered the announcement in New York. This stuff has been speculated for a long time. Facebook already has their ad network -- it's called Facebook Flyers. And going by the ads I see on them ocnsistently, my guess is that they are not doing that well. We even tried using them sometime ago, and ad clickthrough as well as conversion from these ads was abysmal. So are things set to change with the new release?

I don't think so.

I'm trying to understand what they are offering, and here's my understanding. Facebook does provide advertisers the ability to target ads for specific demographics and user attributes (age, location etc). But this is what they do with Facebook flyers -- and afaik it is not a successful program. I think the important factor in ads that helps convert users is not as much just the targeting but the context that the ad is displayed in as well. This is where Facebook seems to be missing out. They talk about beacons and insights -- that people can be walking billboards for brands they love, and spread the virus into their network. So you can publish your purchases etc to your friends, advertising the web site you bought it from. While useful, I would wait to get some numbers on this data to see if this is really worth anything. Because my friends bought "something" at Ebay doesn't make me jump on and go to ebay. It reeks of advertising, not a recommendation. Besides being a privacy concern, I am not convinced that users will want to have their purchases get published to their friends.

The talk about "Insights" seems to be wishful thinking. They "will be able" to track how much people are talking about the brands in forums, which forums and brand pages the user goes to, and target the ads better. I suspect that there is no data that they have to back up their claims -- What is the actual impact they can make for a brand? It isn't even anything new. Companies like doubleclick and Google have cookiess in just about every browse out there. They know which sites you go to, which brands you visit, etc etc. Or at least they are in the position to. They may not know the specific age and sex and favorite color and pet peeves of every user, but when you have enough data about the user through the kind of sites he/she visits, this data starts to become irrelevant -- especially if the end goal is to just target the ads for that user. Age and sex be damned, just show the ads to the user that he or she is most likely to click on. That's the best service you can do to your advertisers and publishers.

It also seems that Facebook is not really providing application developers a new avenue to make money. They seem to be intent on being the publisher for all ads, and attracting the advertisers to put ads -- including application developers. At least I did not read anywhere that they are providing a means for developers to monetize through Facebook ads. This, I think, is where they are missing out. There is more money to be made by sharing the pie with developers than with getting the developers to pay for getting traffic. This would mean, that developers will continue to stick with the existing ad networks to get ads for their pages, and as a result, advertisers will also continue to advertise on these networks.

So my reading of the Facebook ads is that it is more hype than substance. Life's going to go on just as it was.