Online Performance-based Marketing Overview – Part I

(Originally published June 24, 2007 on the DoublePositive Blog)

The year was 1995 and a new technology startup called Netscape had a wildly successful IPO. Their product, of course, was the first widely adopted web browser and the Internet as we know it today was born.

In the early days of the Internet, the web was largely static brochure-ware (meaningful and consistent media feeds came significantly later). There was very little interactivity beyond the clicking of Internet users’ mouse on hyperlinks. Internet Search was not yet a useful tool. Internet users were driven more by curiosity than by intent or objectives. In other words, the spirit of the Internet users in 1995 was essentially “what’s out there,” as opposed to “I know it’s out there, help me find it.”

Perhaps the most useful tool for exploring the web at this time was a site built from the acronyms Yet Another Hierarchical Outline Organization, aka Yahoo! Yahoo used human editors in an attempt to crawl through all the nascent web pages across the Internet and appropriately categorize and link to them within their outline structure. This passive browsing activity of the early Internet users was more like the flipping of pages through a massive magazine (with Yahoo and others providing the table of contents) than it did a brave new interactive medium. Thus, it is not surprising that the owners of the web sites were labeled “publishers.” It did not take long for advertisers to identify the billions of web pages as attractive real estate for their advertisements. However, there were no standards as of yet on how to buy and sell advertisements online. Thus, advertisers and publishers looked to the offline models of buying and selling, specifically those of print/publishing, and broadcast mediums.

For decades, the buying and selling economics of offline advertising revolved around reach and frequency of an advertising channel. A daily newspaper read by 1 Million subscribers could command much higher rates than a monthly magazine read by 25,000. The common denominator by which reach and frequency was evaluated distilled down to the acronyms CPM, which stands for Cost-per-Thousand impressions (the letter “M” actually represents the Roman numeral M, which equals 1,000). Internet website publishers translated this offline CPM model into a system for charging advertisers for every 1,000 times their advertisement were displayed on a web page. Also as in the offline world, web publishers would charge more for larger advertisements and could also command a premium for more demographically-targeted audiences, but these value adjustments all distilled down to CPM.

While this porting of the offline pricing model to the online world was a quick and convenient path for the buying and selling of Internet advertising, it did not take long before the advertisers realized that this model was not taking full-advantage of the interactive nature of the Internet. The key driver to this realization was that the web (unlike offline mediums) allowed for advertisements to be hyperlinked to a different web page, one that was ostensibly maintained and controlled by the advertiser. Thus, the general voice of the advertisers began to repeat “I don’t just want my ad to be SEEN, I want my ad to be CLICKED so that I can present additional information to those users who were interested in my offer on my site.”

A little-known company with the name TeknoSurf from Baltimore, MD is often given credit for being the first to respond to the new demands of Internet advertisers in 1998. The company introduced a new method by which advertising could be bought and sold – advertisers would ONLY be charged for the CLICKS of users on the advertisements, not for each individual impression. This new delivery model was given the acronyms CPC, which stands for Cost-per-Click. With CPC, performance-based marketing was born. TeknoSurf went on to change its name to in 1999, was acquired by AOL in 2004 for $435 Million.

Performance-base models shift some or all of the risk away from the advertiser and to the publisher. With a CPM model, the publisher makes money with every ad view of every user. However, with CPC, the advertiser only pays for the users who not only see the advertisement, but are interested enough to engage the advertisement by clicking on it. Paying only for clicks provided advertisers a much more quantifiable method of determining the true value of advertising placements.

Put yourself in the shoes of an Internet advertiser in 1998. You may be considering buying advertising on and You may expect some degree of higher performance from an advertising placement on a brand-name website such as but how much more? Twice as much? Ten times as much? What if the results are counter-intuitive and the results turn out to be the reverse? Paying only for clicks from both sites cuts through the concerns of ad-placement performance variations. In general, web publishers much prefer the non-performance-based CPM model. What if the advertisement is poorly designed or advertises an extremely low-interest/low-demand product or service (i.e. an unattractive creative advertising a bulldozer repair kit)? In these situations, the web publisher feels as though they have wasted their advertising inventory (impressions) without be compensated, not because the traffic was not valuable, but rather because the offer was generally not attractive from an economic standpoint.

In order to ensure the highest and best use of their advertising inventory, publishers that agree to sell advertising on a performance-basis such as CPC will do the math to determine how much they are effectively earning on a CPM basis (often referred to as eCPM). The bulldozer repair kit advertisement may pay $1 per click but it takes 1,000 page views to generate one click, thus the eCPM is $1. However, an advertisement to enter a sweepstakes may only pay $0.05 per click, but the same 1,000 page views generates 40 clicks, thus the eCPM is $2 and a better deal overall for the publisher. Pulling the economics down to the lowest common denominator of eCPM allows publishers to determine which performance-based advertisements are monetizing their inventory the best. Despite the appeal to advertisers of performance-based CPC, publishers were reluctant to absorb some of the risk and were slow to offer CPC pricing, still favoring the risk-free and easier-to-measure comfort of CPM pricing. The market soon reconciled these conflicting desires between advertisers and publishers based upon supply and demand. Inventory with the perceived highest value (i.e. high-volume traffic of a highly-targeted, high-value demographic such as the finance section of had the leverage to continue to demand a CPM pricing model from advertisers. While advertisers preferred a performance-based CPC, their only option for advertising to the high-value audience on was CPM pricing and they grudgingly complied. However, only sites with the highest quality traffic could hold the line with CPM. Other publishers with less-proven inventory found it too difficult to attract advertiser’s dollars with CPM pricing. The inventory sold by, for example, would not be considered nearly as attractive as the inventory, thus the publisher of would grudgingly agree to a performance-based CPC in order to attract advertiser dollars.

By and large, advertising dollars command most of the leverage within the market. Thus, only a small group of premiere publishers (and a small percentage of advertising inventories) continued to command CPM pricing. The shift from CPM to a performance-based CPC took place between 1998-2001 and was limited to display advertising (banner advertisements on web pages and in emails). During this time, the rate at which Internet users would click on advertisements dropped precipitously. This phenomenon was highly correlated to the shift in spirit of the Internet user away from “what’s out there” curiosity and more towards an objective-driven or intention-based “I know it’s out there and just need to get it or get to it.” Banner advertisements enjoyed a relatively short prime era while they were aligned with the “what’s out there” curiosity of the Internet user. As the curiosity was gradually replaced with familiarity, advertisements needed a significantly higher degree of relevance to an Internet user in order to be rewarded with a click of the user’s mouse. Moreover, it did not take long for advertisers to realize that not all clicks are created equal.

By the year 2000, the Internet advertisers were split into two categories, brand advertisers and direct marketers. Both groups preferred to buy advertising on a performance-based CPC basis, but direct marketers simply used the click as a means to an end. The ultimate goal of the direct marketers was to convert the Internet user that clicked on their advertisement into a sale (i.e. buying a product on or into a lead (whereby the Internet user would provide the advertiser with personal contact information). Direct marketers have very specific targets as to how much they are willing to spending in advertising dollars in order to complete a sale or generate a lead, so the conversion rate from a click relative to the cost of a click (CPC) was the ultimate measure. Direct marketers that had previously enjoyed the comfort of buying advertising on a performance-based CPC basis quickly learned that the value of clicks derived from one advertising placement could vary wildly from the value of a click from a different advertising placement. A $0.10 click from a advertisement could represent a 25% conversion rate while a $0.10 click from a advertisement could represent a 5% conversion rate. Thus, even on a performance-based CPC basis, advertisers were reluctant to buy a broad range of advertisements in order to test and optimize the CPC to achieve their targets.

It was an unattractive solution as it would require significant time, resources, and testing budgets. Rather, the general voice of the advertisers again began to speak. This time, their general voice repeated “I don’t just want my ad to be CLICKED on, I want the user that clicked on my ad to convert!” Not surprisingly, publishers were not initially enthusiastic about pricing their inventory on a higher performance-basis than CPC, but the market again reconciled the conflict relative to supply and demand leverage. The highest value inventory could continue to command CPM pricing, middle-tier inventory could continue to command CPC pricing, but lower-value inventory needed a new pricing model in order to convince the advertisers to spend additional dollars. The new higher-performance pricing model was given the acronyms CPA and/or CPL, which stand for Cost-per-Action and Cost-per-Lead respectively.

At the time, the theory was that this new delivery model of CPA/CPL represented the ULTIMATE in performance basis, where the advertiser absorbed virtually no risk and all the risk was shifted to the publisher. Tracking was relatively simple – a single hidden pixel would load on the “Confirmation Page” or the “Thank You Page” that worked like a counter since the page would only be displayed if the user had submitted a lead or taken any other sales-related action. This method continues to be the standard tracking method today (notwithstanding some javascript-based alternatives).

Just as with CPC, publishers continued to pull down actual results to an eCPM measurement so as to compare all their CPM, CPC, and CPA/CPL deals side-by-side at the lowest common denominator. However, publishers understood that their eCPM is affected by the likelihood of CPA/CPL conversion as much as the actual amount paid per action or per lead. However, unlike a click, there was no standard definition of a “lead” or of a users’ “action.” The exact meaning of these terms needed to be determined in advance and case-by-case before a target CPA or CPL price could be determined. For CPL, publishers would examine the general appeal of the advertisers offer relative to the expected users’ demographics (entering a sweepstakes vs. enrolling for a newsletter vs. requesting a call from a sales professional, etc.). Also for CPL, publishers would examine the number of (required) fields necessary for the lead to be considered eligible for a payout (advertisers typically prefer to demand many fields, but each additional field increases the friction and decreases conversions).

For CPA, typically, the user action was defined as an online e-commerce purchase. Items that affect the publishers’ eCPM would include the price of the item being sold (typically, lower-cost items convert better than higher-ticket items), the general appeal relative the expected users’ demographics (i.e. selling an MP3 player instead of a Bulldozer Repair Kit to teenagers), etc. While this post identifies a clear distinction between CPA and CPL, it should be noted that a significant part of the industry merely lumped everything into the CPA category (arguing that completing a online lead form is still a user action). However, the reverse did not occur – the industry never referred to pricing based upon a completed online e-commerce transaction as CPL.

For the direct-marketer advertisers that were indeed trying to sell a product or service where the sale could be completed online, CPA represented the ultimate performance-basis. The advertiser could simply identify the amount or percentage of a sale they were comfortable in allocating towards advertising and they would know in advance – even before the sale was completed. Increasing the target CPA would ultimately increase the publishers’ eCPM, which would typically increase the traffic (by increasing the number of publishers willing to run the offer) and increase sales. Decreasing the target would have the reverse effect. It became an elegant supply and demand model where the advertiser could essentially determine their own results as established by their own targets. For sales that could be completed online, performance-based marketing had fully-matured with CPA. As a result of CPA pricing, many popular and successful “Affiliate Programs” were launched across the web in 1998 and 1999, including those of and eBay.

E-commerce CPA ultimately had to be completed on the advertisers’ site, so the publishers had to completely accept the advertisers’ tracking and reporting system, which often left publishers feeling dubious about the lack of transparency. But how else could complete a transaction but on It was a necessary condition of this type of CPA e-commerce model. Thus, the “trust factor” became a major element in the “Affiliate Programs” across the web early on. However, CPL did not face the same advertiser-specific-site requirement. Advertisers in search of leads were much more accepting of leads generated by users on the publishers sites.

In most cases, the advertisers preferred the online form be hosted on their own site, but there was not tremendous resistance to the form being hosted on the publishers’ site (especially when the form was properly branded with the advertisers identity). This acceptance by advertisers eliminated the publishers’ “trust factor,” as they could use their own tracking system to ensure accuracy. The acceptance of leads being captured by the forms on the publishers’ sites began a significant paradigm shift and the introduction of the “lead aggregators.” Lead aggregators essentially disintermediated the advertisers’ brand from the user UNTIL the lead was sold to the advertiser. The earliest example of this paradigm is demonstrated by Lending Tree when they hit the scene in 1998 with the slogan “when banks compete, you win.” In this case, the advertiser may have been Countrywide, Wells Fargo, or Bank of America, but the user did not know that until they had responded to a LendingTree-branded advertisement and completed a form on

LendingTree was not selling banner/display ads (CPM) to Countrywide, Wells Fargo, or Bank of America, nor were they selling clicks (CPC) to their respective sites. Rather, LendingTree was selling these advertisers leads (CPL), but the user experience revolved around the LendingTree brand until the lead was handed off to the advertisers. Advertisers did not necessarily prefer this approach. I’m sure the advertisers would much rather have shown users their own branded advertisements and captured the leads on their own branded sites, but in most cases, the advertisers lacked the expertise to buy publishers’ inventory on either a CPC or CPM basis and translate the results into a cost-effective CPL. Thus advertisers would grudgingly buy leads on a CPL basis from lead aggregators because it was the most cost-effective option. With the introduction of CPL into the marketplace in 1998, many advertisers quickly shifted their advertising dollars to this new higher-performance pricing model. Over the next several years, the web became the supreme lead generator – B2B, B2C, product, or service – it made no difference. It became clear to advertisers that people were going online in en masse looking for information, and this represented an outstanding time to reach out and ask them for their contact information using a controlled-cost method.

Stay tuned for Part II of Online Performance-based Marketing Overview where I will describe how higher-performance delivery models are evolving for the transactions that cannot be completed online.

I seldom proactively solicit comments on blogs posts, but given the amount of territory covered on this one, it is probably appropriate to solicit feedback from industry veterans.