In 1996, I started writing about advertising price equilibrium on the Internet. (Yep, this is a yawner, folks.) 9 years later, I'm still talking about it, but more than that we are leveraging this within IHG E-Commerce. My theory revolves around the premise that advertising costs are a function of risk and timing combined with the value of the advertiser's desired outcome plus any branding value. Here is my first attempt to express this function mathematically:

Price of Advertisement = [(Risk/Uncertainty) X Timing] X Value of desired outcome + Branding value.

Risk describes which party - the advertiser or the publisher - is bearing the risks of the advertisement. If the advertiser is required to pay irrespective of performance metrics, his risk is greater. If the publisher is not paid unless a sale (or equivalent desired outcome) is achieved, she bears the risk. For the advertiser's cost model

Timing refers to when payment is made for the ad. This value ranges from prepayment to payment in arrears, with a lower value for payment terms similar to prepayment and a higher value for payment in arrears (inverted for the publisher's pricing model).

Uncertainty takes into account asymmetry of information between the advertiser and publisher. For the advertiser, this factor is highest in the first instance where a particular ad placement is purchased by an advertiser (so it would have a value between 0.0 and 1.0, with a lower value indicating greater uncertainty). For the publisher, this risk continues as long as they are not able to access income about the success rate in achieving the advertiser's desired outcome and the value placed on that outcome by the advertiser. Publisher's tend to deal with this uncertainty by under-pricing ads, so values here would be greater than 1.0. An absence of uncertainty (perfect information) would yield a value of exactly 1.0.

Value of desired outcome is entirely determined by the advertiser. This needs to be a real, fixed value in order for this theoretical model to actually be useful in pricing. For some firms, it may mean reaching an arbitrary value by consensus until a more mature and robust valuation model can be built.

Branding value, also determined largely by the advertiser, attempts to quantify what an impression is worth irrespective of the specific, trackable outcome that might otherwise be desired. (This one gets tricky not because in some cases the impression is the desirable outcome, but because most advertisers do not have this metric defined and this would vary from publisher to publisher or even placement to placement.)

I have a powerpoint chart which expresses this function graphically, and at some point I'll publish it to this forum in some way.

Increasingly, I find myself thinking that there might be a book in this somewhere. My friend Bill Nussey, author of The Quiet Revolution in Email Marketing advises me that this is a minimum 1,000 hour effort even with a professional writer.