The holy grail of advertising is to have a prospect interested in your product, and have an ad targeted to that prospect just at the moment he is researching the product. Put another way, suppose you decide you need a new toothbrush. On the way to the drugstore, you hear a radio ad or see a billboard specifically extolling one brand of toothbrush. This is potentially powerful stuff, and something that the relevance engines behind CPC programs make easy. After all, when you hit a page of information about toothbrushes after searching for them online, Google knows you are interested in toothbrushes and can easily serve appropriate ads.
The basis of CPC advertising is having visitors to your website click the links presented to them by ads. People are likely to notice ads, and click on ad links, only if the content of the ad is relevant to their current interests. This leads to the notion of contextually relevant advertising (usually referred to simply as contextual advertising) often being confused with CPC advertising, even though the two are not the same.
Contextual advertising, which as I’ve noted need not be CPC, but often is, hopes to target a specific individual visiting a website (or page within a website). More accurately, it is the web page itself that is being targeted. A contextual advertising system scans the text of a website for keywords and returns advertisements to the web page based on the website (see Figure 6-2). To some degree, it is possible that the advertising system knows something about this specific visitor, but most likely the analysis is simply based on a general evaluation of the page.
There are a number of forms that contextual ads can take. Besides text ads with links and banner displays, contextual advertisements may be displayed as pop ups, rollovers, and so on. Note, once again, that contextually relevant ads need not be paid for on a CPC basis, but usually are because of the encouraging CTR.
Note
While CPC ads don’t have to be contextually relevant, totally unrelated ads look odd on a web page and are unlikely to be clicked. How often would you interrupt your train of thought to click on an ad for “contact lens cleaner” when you are reading a site about C# programming? As a website publisher, you are better off sticking with CPC programs that do, in fact, provide reasonably good contextual placement.
For example, if someone is visiting a website that covers digital photography, they are likely to see contextually relevant ads for cameras, memory cards, processing services, and so on. More professional-oriented digital photography sites might have contextual relevance for portfolio reviews and other services aimed at advanced photographers.
Note
Contextually relevant ads show up on search results pages, not just on web content pages. In fact, these ads have somewhat better CTR on search results pages than they do on content pages—because you know the viewer actively engaged in a search for the term that is used as the basis for contextual relevance.
How good is contextual analysis? The short answer is that most of the time it is pretty good, but when it goofs it can be a lulu.
It’s completely appropriate to display an ad for a Microsoft developer’s site on a web page devoted to C# programming (C# is one of Microsoft’s languages for developers).
However, sometimes the engine that figures out contextual relevance is just way off base. Most often, the cause is reliance on a keyword that doesn’t really represent the content of a page. Here’s an actual example: in a web page about digital photography, one photograph (of many shown) was captioned to indicate that it showed a child in a bathtub. The Google AdSense contextual ads for the page were for plumbing supplies, claw-foot bathtubs, and so on—all utterly unrelated to the content of the page (and ignoring the real context of the mention of the word bathtub, which was about the photograph, not the tub).
Another—and somewhat notorious—example: consider the news media account of a gruesome crime spree along the Pacific Northwest coast that led to severed feet being washed up on beaches. Some news agencies that ran the story also carried contextual banner ads for a moving company, PutUpYourFeet.com. The only connection between the content and the ad were the keywords “foot” and “feet.”
From the viewpoint of cost, depending on whether the PutUpYourFeet.com ad was CPM or CPC, this ad placement didn’t cost much (if it was CPC, it wouldn’t have cost anything to the advertiser, or paid anything to the site, because one assumes no one genuine would click through). However, it’s very damaging to the reputation of both the advertisers and the site to run such an inappropriate ad.
Technically, the PutUpYourFeet.com ad was an error because it was out of context. But what about an ad that is in context but tasteless—for example, a strident life insurance ad on a story about a prominent death? (This has actually happened.)
The contextual engines may have gotten the life insurance ad on the celebrity obituary technically right—the connection is the keyword “death” and related phrases. But, depending somewhat on the ad, the placement will strike most people as tasteless. Whether or not it leads to clicks through, it is a placement that damages the reputation of the website and the advertiser.
The mechanisms for intervention that might stop something like this from going to press in a newspaper are essentially not present when automated ads are served on the Web. Human beings don’t monitor the ad placements and software analysis is not subtle enough to pick up on nuances that form the boundary between effective and offensive to human beings.
Premium Google AdSense accounts, meaning those with more than 5
million page views per month, have some mechanisms for controlling ad
subject matter, but short of having this kind of clout, the best way to
direct the content of ads on your page is to construct your web pages
following the guidelines explained in Chapter 4 and to provide a contextually
accurate list of alt tags for images and
meta tags for pages. In addition, most CPC programs
(including AdSense) have a mechanism for filtering out competitive ads
(see Chapter 8). You can use a competitive
ad filter mechanism to help keep advertising appropriate. But the
reality is that if you publish a lot of content, sometimes contextual
analysis gets it wrong and there’s not much you can do about it.
A related issue comes up when your web pages feature controversial material and ads are placed by opponents of your viewpoint. For example, I wrote a series of blog posts blasting the rhetoric around “intelligent design.” These posts kept drawing ads from antievolutionists; in other words, people and organizations diametrically opposed to my viewpoint.
Although, as I mentioned, AdSense and other contextual programs do allow you to block specific web domains from advertising on your site (this is supposed to be used to keep competitors off your site), there’s nothing much you can do about advertisers with ideology you don’t like, except grin about the fact that if anyone clicks, the advertiser is paying for a context that will most likely be counterproductive for its ads. So maybe straying from contextual relevancy is its own best reward after all.
How are ads placed on contextually relevant pages? As you can imagine, it isn’t done by a roomful of gnomes scanning web pages and deciding what ads should go on them. There are simply too many websites and web pages, and their content changes too quickly, for this to work (even if you had plenty of gnomes).
Note
Even with print media, such as newspapers and magazines, it makes sense to position advertising near contextually relevant editorial material (although it is not always possible). The manual mechanism that I mock (albeit using humans instead of gnomes) has always been used to try to come up with the best placement of ads, called ad imposition, in traditional print media.
Obviously, software, not gnomes, is used to automatically analyze the content of a web page to determine its content and which contextually relevant ads should be placed on it. This is fortunate for Google (and other search engine companies) because determining the content of a site for contextual ad relevancy is essentially the same task as determining the content of a site to match with the keywords used in a search.
Note
To make it easier for Google’s AdSense and other contextual ad programs to accurately determine the content of your web pages, it’s important that you follow the guidelines for website and page optimization explained in Chapter 4.
Here’s how the process of placing a contextually relevant CPC ad on your site works in a little more detail than I previously provided:
A block of code on your web page calls a script on the server provided by the contextual ad program. This code also contains your tracking ID (discussed in Chapter 8) so the ad program knows who to pay when the ads are clicked.
This code is activated when users load your web page in their web browser.
The activated code invokes the script on the server. The content of your page has been analyzed. This process usually involves some time delay the first time; generic ads may appear until the server has had time to look at your web page. Once your page has been analyzed, the server doesn’t have to parse it again (unless the page content changes, in which case there may be a time lag before the contextual ads have caught up with the new content).
The software decides which ads to serve based on its analysis of your web page content.
The designated ads are generated as HTML and served on your page.
These logistics behind serving contextual ads were shown earlier in Figure 6-2.
There’s a separate piece of this system that these steps don’t describe: how the system on the server determines which ads go with particular pages or sites.
Taking Google as an example, Google has a specific web crawler, the MediaBot, which downloads web pages. The job of the software behind the crawler is to analyze the pages and reduce them to prioritized keyword lists. When the ad unit code on a specific page is loaded, the software compares the keyword list associated with the page with a ranked list of ads to serve for those keywords. This scheme is shown in Figure 6-3.
The amount of money generated by an individual click on a contextual ad is a highly variable and murky business. In the case of Google and other contextual programs, the amount that is paid for an ad depends on an automated bidding process where advertisers bid for keywords. So the amount paid for a contextual ad varies day by day and even hour by hour.
Note
Google’s Terms of Service (TOS) forbids the disclosure of any financial information about how much is made off the AdSense contextual ad program except an accurate account of the total paid to a publisher.
Essentially, a contextual ad program such as Google is acting as an agent or a broker. In old-fashioned language, the role is an automated advertising agency.
Ads are sold by keyword to advertisers, and publishers are paid on the basis of click throughs (see Chapter 7 for the details of Google’s role as a broker). But Google doesn’t say what percentage commission it takes, or what share of the pie it leaves for you, the web publisher. If you sign up for Google’s AdSense, the leading contextual ad program, you just have to take what Google gives you, and trust Google.
Furthermore, the percentage of the fee paid by the advertisers isn’t generally disclosed by most contextual ad agencies. It usually ranges between 30 percent and 70 percent. (I know; that’s a pretty big range!)
An individual click on a contextual ad is not going to make you rich. The net value to a web publisher of a click ranges from one or two pennies to a few dollars at the very high end. The average click is probably worth about $0.20. Combine this statistic with the fact that the CTR (the clicks an ad gets divided by the number of times it is served) are in the low single-digit percentages—2 percent is a quite respectable CTR—and you come up with the fact that to make good money from CPC, you need a lot of traffic. A six-figure income from CPC contextual advertising is not unheard of, but it takes monthly page views in the millions. Doing the math, this page-view volume implies either a broad site or some highly trafficked pages (or both).
Supposing you have a great deal of content—say 10,000 pages—it’s possible to reach the target volume of more than 100,000 page views per month with an average of 10 page views per page per month—ambitious, in terms of the amount of content, but not impossible, and quite modest in terms of the traffic per page.
These 100,000 page views per month might theoretically give you an income of around $400 per month calculated by multiplying 100,000 times the 2 percent CTR, times the $0.20 average fee per click.
Of course, most sites are created to fulfill multiple needs and use a variety of mechanisms to generate income. But if your only consideration is creating revenue using a contextual CPC model, you should consider whether it makes more sense to get to 100,000 page views and beyond by creating relatively fewer heavily visited pages—the shallow-site approach—or a great many less trafficked pages—the broad-site approach. Putting the same question in a slightly different way: is it better to draw niche traffic, which may be worth more because it is highly targeted, or general traffic, which is likely to be less categorizable and therefore less valuable. The answer may, of course, depend on the niche you target.
If you take the broad-site approach, then you’ll need to figure out a low-cost way to generate the content. The calculation that shows a $400 monthly revenue for a 10,000-page site—a broad site that is not heavily trafficked—implies a revenue stream of about $0.50 per page per year. For this to make sense as a business proposition, it can’t cost you very much to generate this content (perhaps because you create it yourself and even because—fates forfend!—some of the content generation is automated).
Business modeling is only as good as its assumptions. If the hypothetical 10,000-page site averages 100 views per page rather than 10 views per page—not at all unreasonable for a worthwhile site—then the economics shift radically in favor of the site publisher. The site is now probably generating $50,000 or so annually in CPC contextual revenue (each page contributes an average of $5).
The point here is to understand the implications of site and advertising metrics on the economics of your site so you can take the steps required to meet your financial goals.
Site metrics, CPC advertising, and profitability in the context of Google’s AdSense program are discussed further in Chapter 9.