The mean value of an Online Ad

Reading about the layoffs on Techmeme this morning, two things struck me that are fairly key to any startup business model

Firstly, TechCrunch reporting on SocNet Hi5 laying off 15% of its staff, despite a glowing writeup on the blog re Fortune mag - that sort of thing doesn't help the cred of corporate blogging at alll.

Secondly, Ad agency Adbrite laying off 40% of its staff - but what hit me here was this:

- Adbrite has revenues of c $32m
- They serve c 1.3 billion Ad impressions a day

Which allows a rapid calculation that their mean revenue per 1,000 Ads served is c 6.7c. Assuming that they take the full revenue before passing it on, that implies the mean CPM across all of AdBrite's customers is of the order of 7c. That's probably a good indication of the mean value any startup should assume.

So, assume you want to build that $100m business - at 7c CPM, say 30 impressions per visitor per month (c 1 visit/page read per day) and 2 Ads per page means that you need c 20 million unique users. If you can get them to visit say 3x a day, and read more pages (say 2 per visit) then it brings down the need to a mere 3.3 million uniques

Update - I see while I was writing this that Alex van Elsas saw the same thing:

A staggering 475Bln times a year people are exposed to advertisement so that Adbrite can earn 32Mln a year. A perfect example where the transaction costs to display advertisement have dropped to zero, but the accompanying value that is being created too.

That got me thinking, as the transaction costs are not zero - I'm not sure what they are for AdBrite, but assume bandwidth costs c $ 0.50 per Gb and each served Ad is 10,000 kb (c 1,000 kB) that implies c 5c per 1000 Ads served - and thats just the fully variable bandwidth costs, before staff etc.