The 3-tiered revenue model for hosted web services

For years, pundits have been saying that software as a service will replace licensed software.  But it’s a tough business. For a hosted web service (especially one tuned for business problems), think about having three tiers in your revenue model.

  1. Free, ad-supported version
    This is what 99% of your users will choose, and where maybe 5% of your revenue will come from.  Sounds like a bum deal! But we know the threshold where a customer will part with their money for software is high (just look in the mirror). You need to have them use your stuff, understand you, believe in your value, trust you, and rely on you before it’s worth their while to pay you.  On the bright side, the cost of providing this free service largely serves as your marketing budget.  This is your cold-call salesforce.  These are your demo/sample units. And nothing grows mindshare as quickly as the right service, offered at an irresistible price.
  2. Premium ad-free subscription
    Maybe one in every few hundred users will find your service useful enough, important enough, that they want to get serious and lay down their dollars to get more.  Get rid of the nonsense (advertising).  Know that you have their back today, and tomorrow (service level agreements).  And provide some extra features to give them their money’s worth. This might be where 25% of your revenue comes from.
  3. Licensing
    As a customer, it’s wonderful to have someone else hosting your software — not having to worry about installing, maintaining, etc.  And you don’t have those big, up front licensing fees. But it’s like giving your car keys to a stranger.  It’s a no-no having corporate data floating around on someone else’s servers, on the Internet, and out of your control.  You have a firewall, and you want your corporate data to stay within it.  So the hosted web service is a great way to play with the software and see what it can do for you.  But at that point, you want to be able to say “sell me your backend software, so I can deploy it on my corporate network”.  Or, perhaps better, make it an appliance or a VM.  And this is where 70% of your revenue as a “hosted services software company” will come from.  Back to the future.

Companies that use this 3-tier approach are everywhere.  How does VA Linux (LNUX) make money on sourceforge?  Why does Google sell an appliance search box?  But there are still plenty of companies that stop at tier 1 or 2, and they may be leaving most of their opportunity on the table.

Selecting a Project: Instinct or Analysis?

Why choose to go after this need for web charting?

There’s no magic analysis. From a lean perspective, it’s a matter of triangulating at least 3 things:

  1. What are the market opportunities/needs
  2. How close is the available technology to filling them
  3. When can our group fill the gap vs. our competition

For a lean-style micro-isv, we would hope for projects that can launch in weeks, not years. In my case, it’s been 7 years since I’ve done this kind of development. I’m starting from zero on nearly all the technologies I’ll be using to create these web sites and web services: ruby, rails, lots of gems, css, subversion, new editor, etc. I know It’ll be slow going on the first projects. And as I experiment with openly blogging about this project as I work on it, the time window of my head start will be short.

For these reasons — and all the other reasons why short iterations, feedback, and adjustment are important — I have to be especially cautious about scale to keep the project managably short.

Personal interest in the space is also an important factor — even with trying to keep each project as small as possible, it’ll still be a large chunk of life working on it.

Data visualization has gotten my attention in recent years. I’ve spent a bunch of time on large software projects. Large-scale software development is an amazing challenge. Microsoft, which does a lot of it, has been pursuing a strategy of collecting and analyzing a ton of data for the purposes of managing their projects. So much data that it’s simply overwhelming. If you can’t summarize it with charts, teams can’t really digest it. Both inside Microsoft (where they can throw people at the problem), and on the wider Internet — we don’t have tools yet to make this visualization easy enough.

There is competition in this space — one can search for companies offering charting, graphing, visualization web services. This is a good thing, as long as there is breathing room for technical or business model innovation. is one example, that offers 1000 graphs for $12.95 a month (with lots of other pricing options) with a free trial that requires registration.

There are interesting new ideas around for doing things better, like Joe Gregoria’s sparklines service 2-year P/L for single project. This is a wonderful example of taking a underused form of visualization, and making it more approachable with a very well designed, simple site for creating the images, and hosting so the images can be easily included on other pages. I’d like to see something similar which makes it as easy to create larger visualizations.

And there are some exciting technologies to build on. Geoffrey Grosenbach’s awesome Gruff library,’s beautiful SWF/XML charts, and web frameworks like Ruby on Rails to speed development.

So, out of a long list of potential ideas, this one seemed to stand out as one where the need, and the technology gap to fill it, seemed to be a match. Instinct backed up by a bit of analysis.

And, in typical lean fashion, I shouldn’t be betting the farm on this one project. I should complete several projects in a year — and expect that I’ll have several failures and a few successes. It is through that feedback loop with the market that leancode as a business can succeed.