August 29, 2003

On Business Models

Don Park and Tim Oren are engaged in an interesting discussion of business models in the wifi world. Don kindly makes a suggestion to have Access Point hardware vendors subsidise the price of the wifi Access Point (AP) by bundling it with a services oriented business model.  It's like a blast of deja vu to 2001, back when Sputnik was getting started, and our original business model

Don lays out the basic ideas behind Sputnik's original model pretty succinctly:
  1. Bob, a store owner, buys Sputnik at 1/4 of the price, plugs it in at his store, and use the installation software to register the AP with Sputnik Network.
    • The AP is configured so that only Sputnik Network members can use it. 
    • Administration, security, and account management is all handled by Sputnik Network.
  2. James, a Wi-Fi user, subscribes to World-wide Sputnik Network service for $10 per month, enabling him to use any Sputnik Network AP around the world.
    • Sputnik client software running on his laptop automatically handles authentication with each AP.
  3. AP usage is metered so Bob might receive a check each month if his AP gets a lot of traffic.
In late 2001, Sputnik released its Sputnik Community Gateway to the world, which would turn any old PC with a wireless card into an AP that authenticated users onto the Sputnik Network, a centralized authentication service.  Lots of people downloaded the code, used the gateway, and people joined the Sputnik Network.  But we decided that we were pursuing the wrong business model, and changed our plans. Here's why:

  1. Revenue split.  Each subscriber is paying $10 a month in Don's example.  Some of that money is going to go to Bob, the store owner who has installed the subsidised AP.  The revenue split needs to be compelling enough to make it interesting for Bob.  But there are other folks in the mix here too - like the ISP, see below, the VAR or SI who installed the AP, the location owner, and possibly the roaming agreement provider (like iPass or Boingo) .  So now that $10/month is split with even more people.  That's a lot of ways to split $10, so your service margins get pretty thin, even as it is. Now add in the fact that James is calling customer support because he's getting unreliable service (see Customer Service Headaches below), and it becomes nearly impossible to make money.
  2. Legal issues.  Most residential broadband connections come with a pretty strict Terms Of Service and Acceptable Use Policy which prohibit the sharing of broadband connections.  Of those that do allow sharing, most only allow for sharing within a single household, not reselling of service.  One way around this is to cut the ISP into the revenue split, which would hopefully provide an incentive to them, and cover their costs of extra data traffic passing over their backbone as well as (potentially) James as a lost customer - why should he buy a new DSL or Cable modem if he can surf on his neighbor's connection?
  3. Customer Service headaches.  If we presuppose that our aforementioned wi-fi user, James, is paying a monthly subscription fee, then he's going to demand some kind of service level, otherwise he's going to feel like he's wasting his money.  The problem is that Sputnik has no control over how James is getting his access - for example, what if his neighbor, who is providing him with wifi access decides to move?  Or if he unplugs his AP when going on vacation?  James doesn't know about this, and the Network provider has no control over James' neighbor - we can't go over to his house and turn his AP back on.  James' percieved value of the service drops precipitously, and he gets puzzled, or even angry.  Then the support calls begin - Since he was getting service just fine the day before, he is going to try to figure out why the service isn't working now.  Now James starts calling the Sputnik call center, trying to diagnose the problem with "his computer".
  4. The rise of "free" networks.  A significant number of businesses are giving wifi away for free - as an incentive to get butts in seats, who then order coffee, or happy meals, or whatever.  Other businesses are using an advertising-based support model - watch an ad when you log in, and you get free access for the day.  Others are using wifi as a customer affinity program, or CRM system - why go to the cashier when you can order your food and drinks while at your seat - "oh, and can we sell you a new CD with that, sir?"  Some businesses just want to get a better idea of customer demographic.  The point is that a traditional for-fee network isn't necessarily the right business model for all occastions or for all locations.  
Believe me, we looked long and hard at the business model and tried to find ways to make it work.  What we realized is that wifi is not a one-size fits all service model.  Sometimes a per-minute or per-day for fee network is the way to go.  Sometimes it isn't.  What we realized, is that by creating an architecture that supported Don's idea allowed us to let our customers figure out the business model that was right for them.  In addition, the further commoditization of wifi hardware means that it is going to become more and more ubiquitous - so the number of potential wifi customers will increase, but hardware profit margins will decrease.  So we embarked on a new strategy:
  1. Give AP manufacturers something to differentiate themselves.  We shrunk our codesize so that it now fits onto the standard flash sizes of inexpensive APs.  Our core code only takes about 150KB of space, which means that there is no need to change hardware designs or increase hardware costs.  At the same time, optional components allow for manufacturers to add additional value through combinations of hardware of software, like VPN accelerators, group policy, bandwidth shaping and throttling, and more.  Licensing is very affordable, and it allows the AP manufacturers to increase their margins by selling differentiated products.
  2. Make money as a software company.  Sputnik's business model is based on selling the management system that lets you control and manage all of those inexpensive APs in a centralized manner.
  3. Let our customers decide on the right business model for them.  We built the Sputnik Central Control system using a set of open interfaces - so that our customers could use different billing systems, settlement systems, and authentication systems.  Because their capital expenditure is reduced by using inexpensive APs, and their operational expense is reduced by using the Sputnik Central Control management system, our customers are free to deply wifi in interesting ways, and to experiment with different service business models.  At $895 for Sputnik Central Control, it is also 1/4 to 1/10th the price of competitive systems.
  4. Be backwards-compatable.  Our products don't use any proprietary new radio encoding method, or even require special client software - all you need at a minimum is an SSL-capable web browser.  That means that all the major operating systems, all the major handhelds are immediately able to authenticate to a Sputnik-powered network.  Of course, client software can make things easier and more functional, but it is not a requirement for the system.  And IT directors can rest easy knowing that they don't have to add a single new piece of software to their standard builds.
  5. Be forwards-compatable.  Wifi is constantly changing - new speeds, new radios, new encryption methods are coming out all the time.  There's a lot of innovation going on in the space.  This is good and bad - you don't want to get locked in to buying a system that will be incompatable with tomorrow's standard. There is one body that everybody looks towards: the IEEE.  802.11 is the name of the IEEE working group that covers this whole area - and all the vendors work on and respect the standards coming out of the working group.  For example, when WEP was broken, the IEEE embarked on a new standard for encryption, based on AES, which is being hammered out by the 802.11i task group.  It's not ready yet.  When it is, Sputnik products will support it.  Until then, we're not getting into the crypto debate or muddying the waters with some proprietary crypto scheme.  A proprietary scheme ends up locking in a subset of customers, but it also ends up fragmenting the market, hurting everybody, especially the hapless souls who are now locked in.
I still love the "change the world" aspect of Don's idea, and ideas like his that build on network effects can certainly create economies of scale and competitive advantage.  In fact, I want to encourage Don to go out and build it and turn it into a gold mine.  Along the way, we're happy to sell him the picks and shovels he'll need to mine that gold.

August 25, 2003

Technorati Downtime

I just got off the phone with my colo provider who seems to be having a major unplanned outage. The long and short of it is that Technorati is down right now, and I don't have a firm ETA on when things will be fixed.

Hopefully soon.

UPDATE, 1 minute after posting this entry: Amazing, things are back up and running. I got a call right after posting this entry, and the systems appear to be back up and running. Hooray!

August 8, 2003

Technorati Tutorial, Part 1

Lilia Efimova at Mathemagenic asked an interesting question about Technorati on her weblog today, and I popped by (thanks to my watchlist) and answered her questions. Given the interest, I thought I'd republish my response here, along with a few elaborations.

Lilia asked,
Does anyone knows how Technorati works? Do they process blog homepages only? Or only items in RSS feeds? Or only things "not older than ..."?

I wonder because I usually observe some fluctuations in numbers of inbound blogs and inbould links. E.g. yesterday I had 100+ inbound blogs and today it's 80+. It would be interesting to know why these things change. I tried Technorati site and weblog of David Sifry with no luck.

I guess this is a quite typical question that user has about systems that digest information: what are the criteria that are used?
Some basics about Technorati

1) We spider weblogs, and correlate each weblog's outbound links to any page on your blog/site
2) Technorati works on any URL - not just URLs for weblogs. For example, you can see what people are saying about an interesting article or favorite company, and get an instant read on the conversations going on around that article or site.
3) The simplest way get your weblog included in the Technorati index is to ping us whenever you update your weblog. That puts you in the high-priority queue for indexing. You can save the page as a bookmark, or you can program your weblog software to do it automatically.
4) To calculate the inbound blog list, we use the outbound links from the blog homepage, not from the archives
5) We do process RSS feeds an other metadata, but that doesn't affect your inbound blog stats. As long as you produce HTML, you're OK.
6) Nightly, we go through the database and re-calculate the number of inbound blogs and links to every weblog we track, which helps us double-check our work and also allows us to create the interesting newcomers list, the interesting recent blogs list, etc.

We strive to be accurate all the time. Sometimes things slip through. For example, one of the reasons why your inbound blog count may be smaller today is because we were doing maintenance of the database last night to remove duplicate blogs - for example, Radio Userland has an obnoxious habit of sending pings to www.weblogs.com for each weblog "category" if you use multiple categories on your blog. Same information, same author, just link spam, basically. So, last night we cleaned out a bunch of that stuff. If you were linked from a bunch of people's blog categories, then you lost those inbound blogs. Then again, so did everyone else. :-)

The last thing to remember is that while we strive for accuracy and completeness, we still do have bugs and have to fix things. If you notice something strange, please don't hesitate to send us feedback (feedback@technorati.com) and let us know.

August 7, 2003

Looking for dietbloggers?

I got pinged today by a reporter looking for people who are using blogs in innovative ways to help them diet. I know that I was inspired by folks like Cory, Glenn and Doc when I got started doing Atkins, but I wouldn't say that I was using my blog to help motivate me or keep track of progress or anything. It does seem that lots of other people are using blogs to talk about their diets, however. Searching through Technorati, I found 592 weblogs that mention "Weight Watchers" and 740 that mention "Atkins", for example. Seems natural, given that most of us feel comfortable to talk about our lives, and that's a part of our lives.

I remember seeing some group weblogs where people were reporting regular diet progress, and egging each other on. I know I saw some folks start a public bet with each other - $100 each in the kitty? - to see who could reach their goals. It is late though, and I can't seem to find the sites (yes, insert irony here). Can anyone send some pointers or links? Is the blogosphere seeing a new trend of dietbloggers? Got any great diet blogs or diet stories? Post 'em in the comments section, or drop me an email. BTW, I've dropped another 10 pounds since the last time I blogged about it. Feeling great.