Written by Ebrahim Ezzy and edited by Richard MacManus. Ebrahim is lead developer of search product Qube and authored the Search 2.0 series, which featured recently on Read/WriteWeb.

This article explores the new online advertising model of CPA (Cost per Action/Acquisition) and determines whether it will be The Next Big Thing on the Internet.

The CPA feasibility test

Let's dive straight into testing whether CPA is good for publishers, then we'll explain the theory behind it and explore whether CPA is really the next stage in online advertising.

The following CPA feasibility test enables publishers to calculate which form of online advertising is best for them - CPA, CPC, or CPM? With the help of past performance data, publishers can develop benchmarks to determine whether CPA will work for them. It's a simple 3-step calculation to check if your web property can bear the risk:

1. Finding pricing benchmark; Numbers of impressions (to be sold or from the past data) / 1000 x CPM = Pricing benchmark 
2. Finding expected click yield; Numbers of impressions (to be sold or from the past data) x % CTR = Expected click yield
3. Determining Target cost per click; Pricing benchmark / Expected click yield = Target CPC

If the Target CPC is any lower than your Average CPC, which is most likely for an average publisher, then it is a risk. 

Beyond CPC - the emerging CPA world

Over the years, search has evolved to become a full-fledged marketing & advertising channel. Internet advertising revenues reached a new high of $3.9 billion for the first quarter of 2006, which continued the trend of record setting quarters. It is abundantly clear that advertisers are seeing a compelling opportunity to leverage the Internet as a powerful medium that drives both branding and sales results.

Google is currently monopolizing the online advertising industry. Internet Advertising or Online Advertising is commonly associated with Google's Adsense/Adwords (and its derivatives), since it revolutionized web publishing by turning personal websites, blogs and forums into potentially lucrative ventures - by setting a new standard for affordable, targeted and effective advertising.

However, the recent issues of click fraud, click attack, ad tracking discrepancies, etc., are starting to cause cracks in online advertising. It makes one wonder whether or not the CPC (Cost-Per-Click) advertising model that Google, Yahoo, MSN and others espouse actually works. Click fraud and other critically contentious issues are a big threat to the CPC model.

Also in these days of budget cuts and profit woes, advertisers are struggling to link a clear return on investment by investing in various online (and offline) advertising platforms. 

So let's look beyond CPC to explore the opportunities offered by the CPA (Cost per Action/Acquisition) model. All the big search engines, including Google, are considering adopting CPA.

CPA - Cost per Action/Acquisition

In contrast to the CPC Model, which seeks to drive a high volume of traffic to the advertiser, the still-emerging CPA model provides action/acquisition opportunities by offering financial incentives - usually in the form of a revenue share percentage - to publishers. Incentives are solely based on actions such as acquiring qualified database entrants (e.g. opt-in email), driving sign-ups, downloads, inquiries or ultimately acquiring paying customers.

CPA is considered to be an optimal form of advertising from the advertiser's point of view, since the entire burden of responsibility is on the publisher. The commissions that publishers receive are dependant on good conversion rates from the ads. 

Indeed it could be argued that CPA, being virtually risk-free for the advertiser, discourages marketing innovation and diligence. If publishers overexpose their users to CPA advertising, it could have a long-term detrimental effect on the industry (not to mention annoy readers!).

The CPA Model - Google is key

To address problems posed by the CPC model, Google recently acknowledged that it's testing CPA. Some niche players (Snap, ValueClick, Jellyfish, etc.) have already adopted CPA, but Google's foray into the market would be a huge step to broaden acceptance of this concept.

Google's competitive advantage is its network size. With thousands of advertisers and publishers, grouping is relatively easy. Depth and variety can compensate for any flaws in the grouping algorithm. So a substantial user base is important - even ad networks with outstanding technology can still perform miserably without a large user base. 

Also eBay, which is in the same business of connecting people - and can be said to be nothing more than a glorified classified advertising business - has launched its AdContext program. This should make CPA more popular, because it will lure publishers by promising them a cut of auction sales made from the ads.

Problems with CPA

• Difficult to manage; balancing the risks of advertisers and publishers against each other makes the CPA model the most difficult to manage. Publishers should be responsible for influencing the consumer, not closing a deal too.
• Not immune to gaming; CPA might prevent click fraud, a major issue with CPC model currently. But soon enough, there will be people will who will figure out ways to cheat CPA too.
• Suitable only for medium/large businesses, because small businesses can't afford to pay relatively high CPA premiums to publishers (see this post by Seeking Alpha for more context).
• Ignores the value of brand building.
• Immature tracking technology; the CPA model may yet be unsuccessful too, because the tracking technology hasn't yet matured into something so robust and reliable that it would enable a vast number of affiliates to be utilized.

Reasons why publishers dislike the CPA Model 

While CPA is ideal for advertisers, publishers won't prefer this model for the following reasons:

• Most financial risk; if an advertising campaign fails and generates no response, the publisher receives no remuneration for displaying the advertisement.
• Low Earnings; in most cases, the revenue generated through a CPA deal works out to be the equivalent of a very low CPM.
• Tracking discrepancies; publishers are often reliant on the advertiser's figures, to calculate the number of actions generated from the campaign. 

Which Ad model is best?

For advertisers: CPA would be best, because ad cost is directly proportional to action conversion rates. Which means the advertiser only pays for actions which contribute to their bottom line.

For publishers: CPM (Cost per mille/thousand) is still the best option. CPM is an advertising model in which the costing is based on the number of times an ad is displayed to the user, regardless of the user's subsequent action. Most rich-media advertising, and nearly all offline advertising (Radio, Television and Print) costs are determined on CPM basis.

CPC is, and will remain, the most cost-effective, affordable and balanced ad model for both advertiser and publisher - IF its underlying problems can be addressed (click fraud, etc.), CPC can provide a radical improvement to overall ad performance and result in sustained productivity growth. Also, increases in productivity - which we've seen a lot of in recent years - is what most drives economic growth without causing inflation. 

To summarize: CPA is best for advertisers, CPM is best for publishers, and CPC is the best for both - as long as the underlying issues, like click fraud, are resolved.

Conclusion 

It is obvious that CPA alone is not the ultimate panacea for advertisers hoping to deliver campaigns that are branded or lead-generation focused. This is evidenced by the fact that many advertisers regularly shift dollars between several types of campaigns - including CPM, CPC and CPA. 

Therefore CPA, in my opinion, is not the holy grail of advertising. But I'm really hoping that Google will pleasantly surprise us, once again!

---------

Note: Ebrahim Ezzy is working on a Qube module called AdRoll, which will enable "free advertising". More details on AdRoll will be published in the coming months on the Qelix blog.

Photos: G0SUB and labnol