Visibility

An excerpt from the upcoming book, “The Ultimate Guide to Pay-Per-Click Advertising”

By Richard Stokes

When you buy advertising placement on the search engines, you specify the maximum price that you are willing to pay when someone clicks your ads. This is known as your maximum bid.

This maximum bid is the price that you are willing to pay for a visitor to your site. It is not usually the actual price you pay per visitor, the cost-per-click (CPC).

The actual formulas used to calculate CPC are well-guarded secrets and change often. This makes it impossible to predict with a high degree of accuracy. Nevertheless, by studying the history of the pay-per-click (PPC) algorithms in conjunction with a bit of computer simulation, we can develop a reasonable idea of how it works and even some insights to help us gain an advantage over our competitors.

The Evolution of Pay-Per-Click

In the late 1990s, the PPC advertising model ran a distant second to the pay-per-impression (CPM) model which dominated at the time. Although a number of ad networks had been experimenting with PPC since at least 1996, it was not widely accepted due to rampant click fraud.

The attractive thing about PPC was that advertisers paid only when people clicked on their ads. This greatly reduced waste and improved the ROI of internet advertising. For those who knew how to use the medium, the return on search engine advertising was far in excess of the traditional gold standard, email marketing (back then, you could buy traffic for a penny).

In the early Overture/Yahoo! days, the CPC you paid was pretty straightforward. You specified a price that you were willing to pay (with no minimums!) for a specific position and that was that. Advertisers’ ads would appear in the order of bids, much like an eBay auction:

This simple model left a lot of risk on advertisers’ plates. As you can see from the example above, the top advertiser was paying $25.00 per click, while the advertiser below was paying only $1.77. Obviously, the advertiser paying a higher price would exhaust their budget more quickly and receive a lower return on their spend (if they made a profit at those prices at all).

In response to this, both Google and Yahoo! tweaked their bidding algorithms so that advertisers would only pay a penny more than the nearest competitor below them. Even if advertisers specified a maximum bid that was far too high, the system would automatically drop their bids to just one cent more than the next advertiser:

Another important consequence of this auction-based model is that it led to the development of defensive and offensive tactics for interfering with competitors’ campaigns. For instance, one of these tactics, bid jamming, consisted of bidding one penny below the top bidder. This forced them to pay their maximum bid. So if advertiser B was bid jamming, they could bid $24.99 and drain advertiser A’s budget:

This “pay for position” model was modestly successful and straightforward, but it led to a proliferation of low quality ads. Search engine users tended to despise these ads due to their poor relevance, and clickthrough rates suffered. Had this trend towards untargeted ads gone unchecked, most people would have trained themselves to ignore the ads, and search engine marketing would probably never have matured into the billion dollar industry it is today.

Yahoo! recognized this problem and attempted to clean up their listings through a manual review process. Advertisers would submit ads and then be forced to wait up to five days for a customer service representative to approve them. This improved the quality of the ads being shown and led to higher clickthrough rates, far in excess of those available from banner advertising.

However, the delays introduced by this cumbersome process prevented advertisers from reaching the maximum profit potential of their campaigns. It was difficult to split test ads, so advertisers took few steps to improve their clickthrough rates. And of course, this slow process made it difficult to add thousands of highly-specific, niche keywords necessary to create a world-class PPC campaign  (Yahoo!’s cumbersome user interface at the time didn’t help things any either).

It was primarily because of these obstacles that search engine advertising remained the realm of a few thousand savvy webmasters for so long. It wasn’t until 2003, when Google launched their “do-it-yourself” advertising platform, that PPC advertising finally caught on with advertisers.

Advertisers Go Gaga for Google

Google entered the online ad scene in 2000[1] with their AdWords Select program. At the time, it was only open to big budget advertisers and, ad buys were sold face-to-face by sales reps, in much the same manner as traditional ad buys. In 2003, they opened up the platform so that any advertiser could participate with nothing more than a website and a credit card.[2]

Google added an important new twist to the Yahoo! model: ads would be awarded placement not only by the maximum bid price advertisers were willing to pay, but also by their clickthrough rate. This not only led to better quality ads (and higher clickthrough rates); it also allowed Google to get rid of the manual ad approval process. As a result, advertisers could manage their campaigns far more easily than before, and Google quickly picked up small and medium customers that Yahoo! had missed.

However, this change removed much of the transparency of how ads were priced because popularity was now being factored into the pricing equation. Specifically, if one ad was twice as effective as another, the first ad would be ranked as if its maximum bid was double what the advertiser actually set. The advertiser would, however, pay only their original price.

Sound complicated? It was simple in comparison with what was yet to come. The algorithms were tweaked and even sometimes overhauled over the next few years to address a variety of issues ranging from quality control to deliberate attempts by spammers to bypass the checks and balances of the system.

Today’s AdWords Pricing Algorithm

Fast forward to 2009, where the CPC pricing model is far more complicated. Google now calculates a quality score for every ad on its system.[3] Many factors go into calculating quality score, but to understand this discussion, you simply need to know that the quality score ranges from zero to ten, where a higher number indicates a better ad (at least in Google’s opinion).

In any given keyword, the quality score is multiplied by the advertiser’s maximum bid price to determine the ad rank:

The ads are then sorted in order by their ad rank to determine the order in which they appear:

To find out how much each advertiser actually has to pay for a click, Google calculates the minimum amount necessary to retain their position. They do this by comparing the prices and quality scores for each adjacent ad.

For example, to find the price for Advertiser C above, we compare it to Advertiser B using the following equation:

You then repeat this process for each of the remaining advertisers. The lowest ranked advertiser pays the minimum bid, which can be as little as just a few cents.

As you can see from this example, paying more does not guarantee you the top ad spot. Rather, the price you pay is determined by the quality of your ads in relation to the advertiser just below you. Mathematically, this is expressed as:

This formula explains why we need to spend so much time on conversion optimization: Improving your conversion rate affords you the opportunity to raise your maximum bid. This directly impacts your ad rank, resulting in higher placement for your ads.

We also manage our quality score on several fronts, including optimizing ad copy, reducing landing page load time, and (perhaps most importantly) monitoring competitors’ paid search efforts. Getting even a single point or two of quality score advantage over them allows us to pay less per click than they are willing to and provides a powerful defense against bid jamming.