Is there any categorization or in-depth analysis of business models based on the Internet?

I'm not sure it answers your question, but one "framework" that I use to analyse business positions is the Five Forces model of Michael Porter.

The five forces specifically give you a framework for deciding whether the business you're looking has some ability to be successful.

As with any framework, it has downsides, but I find any order you can put on an analysis helps to clarify the analysis (provided it's not completely antithetical to what you're trying to achieve).

Not really bounty-full, but here's some extra information:

• The paper by Weill et al. is generic for the top 1000 US companies (so not specifically web-related), but it comes up with the conclusion:

  selling the right to use assets is more profitable and more highly valued by the market than selling ownership of assets.

  The paper goes on to talk about the four primary types of business model: Creator, Distributor, Landlord or Broker, and further subdivides these four by the type of asset involved: Financial, Physical, Intangible or Human.

  That gives a total of 16 business types:

• The paper by Dubosson-Torbay et al. is more specifically about eBusiness, and takes quite a different approach: it breaks down a business into four "processes" (my wording) : Product Innovation, Customer Relationship, Infrastructure Management and Financial Aspects. It then investigates various examples of businesses, and looks at what they do in each "process" category.

• Interestingly, one of the co-authors of the previous paper, Yves Pigneur, has an iPad app for investigating business models that seems kind of cute.

I don't have too much more time to chase this, but I thought it would be interesting to cross-pollinate the two ideas from the papers above. The second paper has some example businesses: auctioneers, retailers and a CIO group.

They fit (my guess) into the framework from the first paper here:

And the way the second paper proposes measuring each of these three business areas is as below.

Product Innovation

Customer Relationship

Infrastructure Management

Financial Aspects

Conclusion

There's lots of white space in that spreadsheet, and I don't really have time to prognosticate more. I've uploaded it to Google Docs, so you can download it for yourself if that'll help.

What'd I'd like to do next is to see what Porter says about the three examples I've filled in... but that's for another time.

Well "conducted through the internet" ... is pretty broad statement.

• Google is all about the number of times they can get people to click on an ad link.
• Facebook are similar but their success metric is the amount of time they hold peoples attention (google only have you a few mins at a time, facebook is hours at a time).
• Evernote have subscribers volume with a 10% conversion to paid clients, given they have millions of users then its "dedicated user base" which they can work with long term.

So metric that you could select from when building an internent business

• Impressions per day. If your collecting money through advertising impressions then this is a good one.
• Unique Visitors per day. If you listen to This Week in Startups then you will hear this metric come up A LOT.
• Returning visitors per day/week/month. For long term sustainablity of a lot of web properties this is key (like Facebook or Linkedin)
• New Visitors VS returns. This is a good one to see if you have a high level of interest in the site. The stack exchange will probably use this as one of their key metrics
• Severs per volume. How effecient are you, how fast can you scale. 1 server per 100 customers or 1 server per million customers is a big difference in your bottom line and your "need to rearchitect your system" point.
• Customer issues resolution rate. Open VS closed. This is crticial in many businesses for long term survival ... in the online world it is more so because your customers are likely to discuss your performance good or bad.
• Points per sprint. For development how many additional function points can you get through per release of your system.
• Technical debt backlog. Tracking this is critical as is making sure it stays fairly low. Too low and you won't move forward fast enough, too high and you will hit a brick wall and risk a major failure "somewhere".
• Staff turnover rate. Average months served is critical for long term consistancy especially on the technical side. A revolving door means nobody knows what is happening or where the technical debt is ... basically all your code base becomes technical debt.
• Mentions in the press. Old fashioned desicrption but concept still applies. This translates to how many refereneces there are either to as links or to you as mentions. CAn be facebook, twitter, other poeople blogs, etc. There are tools which track this for you now.
• Your google rank on your keywords.

General metrics for most businesses:

• Profitablity. Every company has this one and its a cheap shot :)
• Cost of customer aquisition VS value from customer. This says how effective your being at extracting money from the customer. Get this ratio wrong and your out of business quick.
• Life time value of a customer. Do they buy your product once so everything is stacked against the first sale or do they come back / get subscription in which case the first sale can be given away in order to get the long term value from the customer.

There are heaps more once I get into the swing of them but most people have probably stopped reading by now ...

So for those that have made it this far the key thing you need to do is identify and track the ones that are important to you. Setup metrics / goals around each one so you have something to compare them to (is 50,000 a good number or not?) ... then the key magic bit is to normilise them somehow.

The easiest way is to setup ranges of expecatation per variable you are tracking:

Variable 1: (say unique visitors per day)

• 1 to 100 is a 1
• 101 to 200 is a 2
• 201 to 5000 is a 3
• 5001 to 50000 is a 4
• 50001 to 500000 is a 5

etc.

Have about 5 - 10 buckets anymore starts to get meaningless (unless you use 100 which works well too). So now you have a common measure for all of your variables:

So you can then say

• Profitablity is 2/5
• Issue Resolution is 3/5
• Average Time on site is 4/5

And you end up with my company is 3.5 on my magic "going well odometer"

ADDITIONAL BIT:

And to finish that thought off properly ... Instead of one overall metric I would break "internet companies" down into a series of subcatagories and then rank them against each other.

It will be interesting to see if there are any concise and well-researched resources for this. I would start by looking at a few dimensions:

• Product or service (what you provide and/or how you provide it)
• One-time payment or recurring (how you're paid)
• Free (promotional), ad-supported, freemium, premium, or exclusive (the pricing level)

Based on various combinations of these you could categorize business models, such as a recurring subscription to receive exclusive (high-margin) products. There's probably some variation in success rates but that could be more about the people running the business than the business model. For example many people scared to charge full-price for something fail on the low-end of the pricing range because they aren't good at figuring out a business model, not because lower prices are a bad business model.

The success factors definitely will vary. With exclusive pricing your challenge is to make people want something no one else has (sometimes this could mean reducing volume). With low pricing your challenge is to get high volumes with optimized marketing. With one-time sales your challenge is to keep new people coming in; with recurring sales you challenge is to keep customers interested. If you have a lot of cheap labor services are efficient; if you have repeatable technological advantages products are efficient.

That said there are business models that have a very high chance of failure, like an unknown person getting extremely popular on youtube in order to launch a real-world entertainment brand.

Just before posting this I remembered Alex Osterwalder's Business Model Canvas: http://en.wikipedia.org/wiki/Business_Model_Canvas. It's more of a tool than a reference but it's a good way to break things down.

The one I like the best and is quite new is called Startup Genome.

They have done surveys and indepth analysis of what makes startups successful. They have a very interesting report you can download.

They allow you to do a detailed Benchmark of your startup, and it will place you into one of it's four startup types, and give you detailed comparisons of your startup to the others in your category.

The four categories of Startup Personality Types are:

• Type 1 - The Automizer
• Type 1N - The Social Transformer
• Type 2 - The Integrator
• Type 3 - The Challenger

After I took the test, I found I was a Type 1 Automiser.

I have not read it but time and again when I ask about this topic, people recommend this book: http://www.businessmodelgeneration.com/