As some of you may know I worked for an online advertising company - Bravenet Web Services and its affiliate Bravenet Media Network (which has since dropped the redundant Network).
My job at the outset was centered around maintaining spreadsheets which monitored the revenue and profitability of what amounted to a proof of concept that was put into full-time service with three dedicated staffers including the senior-most employee.
Reading the O’Reilly radar reminded me of the simple, and bleak, economic reality that seemed to be either ignored or denied - generating serious revenue is damned hard exclusively in online marketing.
Tim O’Reilly references a blog maintained by a successful VC called Lightspeed Venture Capitals regarding the economics of online marketing and furthers it with his own comments regarding some basic Web economics phenomena.
The premise of Jeremy Liew’s article is answering the question, “What kind of resources does it take to generate serious cash in online marketing?” Serious cash to Jeremy Liew is $50 million USD in annual revenue.
Okay - that is serious cash, no argument here. But most businesses of say 45 - 65 employees in the tech sector would be glad to take home $4 or $5 million in profit. So either Mr. Liew has very VERY high operating costs or just expects very serious revenue.
$50 million, however, is the magic number for IPO and for corporate acquisition (assuming they don’t want your assets.) As any student of Benjamin Graham would know 
In any case - Liew realisitically asserts that a company generating a whopping $1 RPM (that’s revenue per thousand views, through whatever mechanism you like - Cost Per Action (CPA), Cost Per View (CPV), Cost Per Impression (CPM) etc.) Which are usually general interest, general community sites (MySpaceMonster, etc) communications tools (UndeadMail, and JiggaMail) with no real demographic focus would need just over 4 billion page views a month - which is a stretch even for Godgle. Getting $1 CPM average on general traffic is tough. Getting above a quarter was tough a year ago and I imagine it’s just gotten worse from there.
So, next rung we have the slightly more premium but still consumer focused site - the vertically organized portal or *cringe*… vortal(”vortal” & “incentivise” are two non-words that should be stricken from humanity - does incent not have enough syllables?) These topical sites can usually get in the neighbourhood of $5 RPM. Making for 800 million page views a month your minimum at that rate.
The next rung up are the premium sites, these are usually either highly concentrated (read: captive or known-professional) audience or more commonly sites about topics with deep marketing pockets (movies, cars, travel, wangs and plastic surgery) and can get into the $20 RPM territory especially if they can launch parallel campaigns or about 200 thousand page views.
Mr. Liew leaves it at that to make us consternated as contemporary Venture Capitalists and to reassess the interlogistics of our current vision to leverage our millions on some upstart interpipe web media network .com
The reality is that although it is hard to make serious cash it’s not as hard as Mr. Liew describes it.
Firstly he cites those RPMs, those are accurate as far as total revenue an average page view retains. A good chunk of my and others’ jobs at Bravenet was figuring out whether 1 ad was twice as effective of each of 2 parallel ads, and so on. So keeping up on optimizing that might garner you an extra 8 - 13%
So let’s focus on the second tier, as in actuality this is the easier of the tiers to develop. The first usually requires a) expensive hardware to handle those kinds of traffic volumes and b) development of some sort of tool system for people to use to generate your content (a profile system, dating system, webmaster tools etc). These are costly, cumbersome and further delay your time to market. Not to mention your profit centre (traffic) also becomes your primary cost driver (server load and tool expansion).
The third tier is difficult for a number of reasons. Firstly, the market is reasonably saturated. There’s a tonne of movie sites already out there (though I haven’t seem my MovieMatchMaker idea yet) there’s buckets of travelling sites and others. Secondly, such sites are usually difficult to develop because they are complicated by things like entangling copyright laws, content generation costs, and often require domain expertise your average Mum-and-Pop-Webmaster lacks. Third tier sites are usually the products of established corporate developers with the resources and manpower to invest in what is essentially a moneysink. Tier one sites are usually the product of reclusive teenagers who spend far too much time tinkering with Coldfusion and desperately need to get laid so they develop software to help them meet chicks.
So, $5 RPM - 800 thousand page views a month. That’s a challenge for a single site, but not a challenge for say, 12 sites. Assuming a 8% reduction in time-to-market per iteration, with say a 15% inefficiency each iteration and an initial development time of 5 months. You could have your 800 thousand page views in 4 and a quarter years, and you’d be making revenue in 4 months.
Now, I could whip out the calculus regarding acceleration of web development, release of each site, acceleration of site traffic - adjust for probability of random site failure or downage etc (all of those figures are easily and accurately obtained) but the reality is, assuming you start on Jan 1st of the year, and your first site is up and running June 1st, you can expect to be getting that 66 000 pages view (1/12th of 800 thousand) within the calendar year assuming you’ve picked your demographic properly and don’t offer garbage (and even then, you might succeed despite yourself).
So the numbers aren’t as scary as Mr. Liew points out. What he fails to acknowledge, which O’Reilly points is that the long-tail effect is a good thing for publishers and a pain in the ass for marketers and Venture Capitalists.
2nd tier sites don’t require massive infusions of cash (read: shares handed over to VCs) to get started because their staffing is minimal (5 webmonkeys, 3 content developers, 1 manager/owner, 1 sys-admin, 1 sales person and a KPMG agent walk into a bar and someone says “Hey, there goes a demographically targeted web site development crew!”)
However, 2nd tier sites fail very VERY frequently. One must treat them much like books (Gawker media does that by buying into established bloggers and fostering new ones, hoping the law of averages will give them more hits than misses while their revenue drivers keep up the bottom-line) and they’re always a gamble - but they’re a SMALL gamble.
So, if you’re interested in starting an online media co. it’s not rocket surgery but don’t quit your day job yet.