i just came across this techcrunch post that attempts to refine the way in which social networks are valued. in a nutshell:
- ignore visitors and page views
- calculate internet spend per online user and multiply that by the number of users in a given network to obtain a valuation for the network
- compare these valuations to the value actually paid in recent transactions
- apply these multiples to the other networks you want valued to cross-check against the valuation obtained above
read the techcrunch post in detail if you like, then come back.
i think this methodology is a good effort and definitely much much better than valuing eyeballs alone BUT:
- the value of a network grows non-linearly as the number of nodes increases. we all know this. yet this methodology simply calculates values for various networks in a linear fashion
- this model doesn't account for cost structures, only taking (potential) revenues into consideration. which may seem a persnickety objection to raise, but if we're talking about spending hundreds of millions of dollars yearly for the larger networks just to stay alive, then it doesn't seem so inconsequential after all. (besides, costs can be more non-linear than revenues.) however, they're right in the sense that the number they come up with can be considered the revenue potential (not valuation) of a given network
- what someone else paid for a network... it's either a bit artificial/circular or completely irrelevant
- artificial because the logic runs like this: someone paid $x for that network, therefore i shall pay $y for this one and someone else will follow my lead and pay $z for that third one, and so on. but what if that first $x number was 'wrong'? shouldn't valuations be intrinsic to networks? how quickly can you say dotcombust?
- ...and possibly irrelevant because how i value something could be completely different from how someone else values it. this doesn't matter in an efficient market (~= public markets) where there are several investors both long and short waiting to pounce on mispriced assets but DOES matter in an inefficient market or one that's non-existent (i.e., private companies)
but what's the conclusion? is this exercise completely useless? not quite, but it's not the be all and end all either. it's just another data point to be considered.
in fact, there is one way to improve the methodology: apply it to public companies, compare the derived valuation to the valuation assigned by the markets and test the efficacy of this model that way (whether the public market valuation is itself 'correct' is a different debate altogether).
so which social network is public today? surprising but true: google. obviously, google isn't a 'social network' in the sense that most people mean it, but to all intents and purposes, this model is quite applicable to them. what do they do? they serve ads. what does this model do? it calculates the worth of online ad companies. besides, google owns youtube. guess what youtube is in the business of.
