This is the subhead for the blog post

When you look at the evolution of online marketing, you see an almost linear march toward performance-based marketing. For example, in the beginning – say, 1996 – most advertising spots on the Internet were sold on a subscription-basis. For $1000 a month, you could buy banner placement on a top Web site.

This quickly morphed into CPM (cost per thousand) advertising. Now, instead of just plunking down some cash and hoping that someone would see your ad, you at least got a guarantee of a certain number of impressions. Then came Overture and Google and CPC. Now impressions weren’t enough. Now you were guaranteed clicks (or at least, you wouldn’t pay until you got clicks).

And we then saw the rise of CPA – cost per acquisition. Companies like Adteractive, Azoogle, Quinstreet, and Nextag have created massive businesses simply by selling leads – the advertiser doesn’t pay until he actually gets a completed “request for information” form, or even a sale.

This is not to say that CPA has taken over online marketing. Far from it. I would say that CPC has or will soon surpass CPM as the dominant form of online advertising. CPA is growing very quickly, but it is probably at least three years away from generating the same type of revenue currently generated by CPC.

Nonetheless, even if we aren’t yet in a CPA world, the result of this evolution has been two-fold. First, it has resulted in shifting risk from the advertiser to the publisher. Second, it has made metrics and analysis critical for all parties involved.

The first point is fairly obvious, so I won’t spend too much time on it. In a nutshell, if a publisher doesn’t get paid until an advertiser gets a lead or a click, that publisher is essentially betting the farm that their users are interested in their advertisers’ products. Compare this to the old days of subscriptions (the publisher didn’t even have to guarantee a certain number of impressions or page views) and you can see how much higher the risk-profile is.

For that reason, publishers need to become hyper-sensitive to metrics. Letting a non-performing CPA offer run on your site or network for too long can be the difference between a profitable and unprofitable month. Similarly, signing up for a CPC distribution network that doesn’t have a good algorithm to match your content with the right CPC advertisers is also disastrous (this point was made clear to me by readers when I discussed threats to AdSense from YPN and Amazon. Many readers pointed out that AdSense is so far ahead from an algorithmic perspective, that these aren’t true competitors yet).

In fact, the metrics begin long before the offer runs; these days, publishers must model the likely performance of an offer before signing the contract in the first place. I should note that by “publisher”, I am basically talking about lead gen companies. To date, the only major Internet player I know of that has accepted CPA advertising is MSN.

What keeps me up at night (other than Iowa losing in the first round of the NCAAs . . .) is trying to figure out what comes next after CPA. In my view, the logical path is a revenue share model.

Revenue share – where the publisher gets a percentage of the advertiser’s sales – already exists. Amazon’s Associate program is often credited with creating the revenue share model online, and that program has been around for 6 or 7 years. For that matter, if you login to Commission Junction or Linkshare, you’ll find that about 90% of the offers available to publishers are revenue share offers, the rest being CPA.

You may be asking, then, how is this evolutionary if the revenue share model is already thriving? The evolution I see is that revenue share expands beyond the mom and pop’s and professional affiliates using Linkshare, and into the lead generation companies and even major online publishers.

The hardest job advertisers have when considering a CPA deal is measuring potential ROI. Buying leads for $10 each is well and good, but if none of those leads convert into paying customers, you might as well buy CPM or CPC.

For this reason, more and more top advertisers are negotiating with lead gen companies on two axis: cost per lead, and quality per lead. “Lead quality” is essentially a nice way of saying “did the lead convert into a sale.” When you combine lead quality and cost per lead, you end up with cost per sale. At that point, it is difficult to argue that the advertiser is buying leads, since the price being paid is actually a percentage of the sales the advertiser is receiving (this, by the way, is analogous to calling Google AdWords a “cost per click model. Since AdWords combines CPC and CTR, it is more appropriately called an “effective CPM” model, but most people haven’t figured this truth out yet).

The problem with this sort of model is that an advertiser’s success rate really becomes a black box to the publisher and is open to manipulation. Let’s say that you have a widget advertiser who is willing to pay you $20 for every lead from someone interested in widgets. The campaign starts and one month later, the advertiser storms into your office complaining that only 5% of the leads you sent him actually converted into paying customers. He demands that the cost per lead be reduced to $10 or he’ll cancel the campaign. Because the conversion takes place after you send the lead, you have no idea whether this is accurate or not.

Contrast this situation to a classic CPM or CPC sale. In either case, the advertiser can be presented with detailed reports about CTR, keywords, CPC, placement, and so on. If the advertiser has a complaint about anything (for example, click fraud), the black box is on the publisher side, leaving the advertiser at his mercy.

The problem, then, with a revenue share model is essentially one of trust and transparency. Publishers have already taken tremendous risk by shifting from CPM to CPC or CPA; the shift to revenue share is currently too much for most publishers to handle. How can this be solved?

My thought is that there are two solutions – market conditions and end-to-end tracking. By market conditions, I mean that over time an advertiser that lies about the actual value of sales will eventually be pushed out of prime positions by advertisers who are more honest. If widget seller A claims he is only get a 5% conversion rate on leads and widget seller B claims a 50% conversion rate, it’s only a matter of time before a publisher will stop running seller A’s leads, simply because the rev share is so much higher for seller B.

Of course, the problem with this argument is that it assumes an efficient market, where all (or a critical mass) of the possible advertisers are vying for the same position on a publisher’s page. If you only have one advertiser who wants to play with you, you have no choice but to accept what he says as fact. Still, if a Yahoo or MSN wanted to really get into rev share advertising, they could easily aggregate enough advertisers to create incentives for the advertisers to be brutally honest about conversion rates (and thus give MSN a nice chunk of rev share).

One possible solution for publishers who are dependant on a small number of advertisers is point two, end-to-end tracking. Under this model, the advertiser basically has to let the publisher track or host the entire customer experience – from the moment a user sees an ad, to the check out, and maybe even through future visits to the advertiser’s site. In this way, the publisher gains transparency into what the advertiser is actually making from his site, and can thus be assured that the advertiser isn’t lying about conversion rates in order to increase margins

With all the future predictions I make on this site, it’s always nice to have an example of someone actually doing something similar to my ideas today. And that company is Snap was started by none other than Bill Gross, creator of the Overture CPC model and the god-send for Google. I guess you could say he is the “father of pay-per-click.”

So Bill has decided that CPC is yesterday’s news, and now he’s built a search engine that allows advertisers to bid on a CPA basis. As the Web site “About Us” page proudly proclaims: “Finally, a search engine offering true pay-per-performance advertising. Snap’s cost-per-action advertising options mean that you pay only when you get the action you want from the customer, giving you maximum flexibility (a customer action can be a purchase, download, registration, subscription, lead, or just a click). You decide what makes the most sense for your business. The result is that you have ZERO RISK and an INCREASED RETURN-ON-INVESTMENT.” (Hey, stop yelling at me!).

Right now, as far as I can tell, Snap isn’t a roaring success, mainly because it has fallen into the GoTo trap. GoTo – the name for Yahoo Search Marketing before it was Overture – tried to exist as a destination search engine that just happened to sell ads on a CPC basis. That didn’t work too well because, frankly, most consumers are set in their ways and are either going to use Yahoo, AOL, Google, or MSN. So GoTo got smart and distribution their CPC leads across big publishers – in particular Yahoo and MSN.

Assumedly, this is the plan for (unless Bill hasn’t learned any lessons at all, which could be the case, considering the rumor that he is broke). And if you think about it, there’s no reason why Snap couldn’t have the same successful distribution that GoTo once did. After all, from a publisher perspective, everything comes down to eCPM – effective cost per thousand impressions). Whether that eCPM comes from subscriptions, CPM, CPC, CPA, or revenue share really doesn’t matter.

At the end of the day, a revenue share model provides increased transparency for both the publisher and the advertiser, and this where Internet marketing is heading (publishers have been in to revenue transparency for a long time, the point is that advertisers are now starting to pick up on this too!). Over time, I suspect this will reduce margins somewhat for publishers, simply because advertisers won’t be wooed by sales pitches to overspend for worthless inventory.

I do think, however, that the drive towards transparency will end up benefiting Internet marketing enormously, at the expense of offline marketing. Just think about it – you could buy advertisements online that virtually (no pun intended) guarantee you profit, or you could spend $1 million on a primetime TV ad that will likely be Tivo’d over anyway. So while the margins might be slimmer, the overall media spend will only increase with transparency of value.

As I’ve noted before, the biggest losers here are the Madison Avenue agencies and the big corporate marketing heads who really like those boondoggles with network execs. Revenue share deals are the worst thing that could happen to these bloated industries. Terms like “lift” and “buzz” are thrown around to justify 25% margins and minimal correlation between media spend and ROI.

When smart advertisers start to spend more and more on online revenue share deals, the end result will be that dumb advertisers will lose market share, and the agencies and publishers (i.e. TV networks) that cater to these dumb advertisers will go down with the ship.

To conclude, then . . . Internet advertising is moving toward a revenue share model, and TV is dead. Allow myself to pat myself on the back for predicting the death of search engines, AdSense, and TV in a time period of less than three months! I think I need to get out more, this is becoming a bit too morbid.