Was “Classic” Coke so much better than “New” Coke? It sure was in the real world. Why mess with a good thing? A favorite formulation turned out to be, well, still a favorite in spite of a more modern, “scientific” formulation.
Google creates products that are a lot more complex than a soft drink. When all is said and done, I think the latest-generation product development protocols are going to have to lead to significant shifts in how advertisers interact with online ad platforms like AdWords. In the short term, for example, many of us have preferred “classic” Google AdWords Content Targeting to the newer Site Targeting flavor of content targeting. But that’s probably temporary. In short, sentimentality means little in the face of changing economic trends and the impatient needs of today’s sophisticated online actors. Google ain’t a soft drink.
New, improved, more scientific formulations are likely to make us ever more impatient with some of the “classic” versions of the ad auction platforms. I no more think that the world will continue to put up with some of the quirks of first-generation ad auction systems than I think I’m likely to go back to using a retail stock broker who sends me paper confirmations in the mail that tell me I was lucky enough to pay a $78 commission on that small stock buy that I researched myself.
Google content targeting already showing its age
If you’re like many search advertisers, you fell into the so-called “content targeting” add-on to paid search platforms mainly through inertia. As you reached the limit of search click inventory, you enabled content targeting (or “contextual” ads). You then began to manage this carefully to ensure you weren’t getting tripped up by the quirky economics and distribution patterns.
If you were one of those unfortunate ones who enabled a new AdWords account without looking before you leapt, I’m afraid your budgetary mix was even more exogenously determined than that. (“Sorry, newb.”) Basically what would have happened if you weren’t paying enough attention was that you would have launched your campaign by opting into content targeting, even if you may have been only dimly aware of what it was. Worse, you bid the same as you bid on search, even though almost no one does this across the board due to the often sharply different ROI’s on the two streams.
At a certain point, in either case, you decided to take more control of the situation: tracking results more precisely, bidding better, and now, reading Placement Performance reports within the Adwords interface to determine what referring sources were responsible for the majority of your clicks, and whether they converted or not. (Such reporting has long been available with good third party analytics tools, if customized properly.)
On the whole, Google’s content targeting program now works well, in the very practical sense that more inventory is now available, and that steps have been taken to improve both targeting and pricing, so that the ROI is improving. (One of these initiatives is called “smart pricing” – Google’s algorithm looks at the projected quality of publisher pages in terms of the likelihood of behaving like pages that typically convert to sales or leads for advertisers.) Instances of zero conversions on substantial numbers of clicks are less frequent now in the content program.
But what’s striking is that most of the “improvements” to Google’s contextual ad program have come in the form of “fixes” to an initially flawed product that was released to a skeptical response from advertisers. The “fixes” were relentless, and admirable Google’s overall revenue from contextual ads dropped from over 50% to below 40% over a two-year period as a result; you can bet they are now hoping to grow this back to well over 50%, given the volume ceiling on search inventory proper. But the reality is these were still band-aids applied to a clumsy model whose black-boxishness is legendary. To exaggerate for effect: neither advertisers nor publishers know what the heck is going on with the program! Meanwhile, Google gets paid handsomely for acting as the middleman. Google has been playing the part of one heckuva crafty mutual fund salesman.
Google: today’s sophisticated product development strategies
The latest generation of Google, the diversified web products/services company, has taken the world by storm. The world caught wind of the momentum with the release of things like Google Maps, Google Earth, and of course GMail. But most pertinent to the present discussions, and to search marketers, has been the release of the latest version of Google Analytics, and the rollout of Google Website Optimizer. Inside Google AdWords, to boot, there have been a number of innovations that look like they fall into a similar category or generation as the above products.
Systematic planning efforts go into these products now, with increasing degrees of input from directly affected communities and blue-chip panels or beta testers getting more involved. Another clue to Google’s growing maturity is the establishment of formal reseller programs that allow collaboration for mutual benefit.
The fact that Google is now a mature company that creates and evaluates product development in a sound fashion that reflects input from relevant experts in the immediate ecosystem, as well as users, puts the design of the platforms that generate its core revenue streams at a crossroads.
The current paradigms that AdWords and AdSense (content targeting) are built upon must satisfy a number of criteria if they’re to stand up to the next phases of growth and competition:
- They need to scale
- They need to integrate well with the efforts of editorial, sales, and customer support staff, and be well understood by them
- They need to adapt to different communities of users
- They need to make economic sense to advertisers and publishers
- They need to be transparent enough that they don’t appear to be one-sided in Google’s favor
- They shouldn’t be overly complex
- They shouldn’t be just plain stupid, or seem stupid to the average user
Some of what I’ve stated seems obvious. But these are pretty daunting criteria in some senses, depending on what the market comes to expect.
Content targeting, the product: pros and cons
In this case I’ll just look at content targeting. From the beginning, the program was flawed in a number of senses. I’ll tip my hat to the benefits, which I’ve always tried to highlight. The automated methods of matching content to the keywords and ads in your account allow advertisers to add relevant exposure with no research and often make fewer mistakes than human media buyers. The system also (now) allows for control in the form of separate bidding, site exclusion, and Placement Performance reports.
But there have been numerous downsides. First, it seemed manipulative: newbies are opted into content, at the same high bid as they pay for search. It’s all fine and good for customer support to “recommend” content bids and other niceties, after the fact, but designing the platform to be misused in the first place sounds like the kind of “you had every right to…” formalism that has no place in a discussion of real world usability.
In theory, content targeting works. The “reverse usability” of the program, to anyone watching carefully, makes it obvious that in fact there was too much one-sidedness in Google’s favor built into the very design.
On complexity: there is plenty. Many advertisers simply do not understand how the content targeting program’s matching works. Those of us who do understand that our ad might show up on relevant pages of any type around the web, based on a proprietary matching technology and our bid, still really do not clearly understand exactly what’s going on, and have a limited grasp of the overall size and shape of the ad network. We definitely have no clue as to what sites our ads will show up on next. That’s uncertainty combined with complexity, especially for the direct-response advertiser who counts on ads to convert, and doesn’t like to put money in the pockets of untried publishers.
But it’s not as easy as it sounds to come up with an alternative.
Radical simplicity? Brilliant hacks, design philosophy, and markets
The potential solution to that leads to another dilemma, however. The obvious answer, to some, would be to remove complexity in the way that Apple does: build such a compelling, intuitive outer layer that nobody seems to care that there’s a crazy amount of the same old sausage being made by tiny primary-colored leprechauns inside. It seems unlikely that Google can go to this extreme – by, say, sending you an “ad pod,” a lovely white plastic spheroid that blinks its red eye and tells you your ad spend and not much else (other than “Googlepod loves you!”) at the end of each day (or week, depending on whether you order the “red eye ad pod” or the “blue eye ad pod”). We’ve been too conditioned to drill down on details of user behavior and ad performance to go for this. And by “we,” I mean us OCAJ’s – Overly Conscientious AdWords Junkies. Many ordinary advertisers out there would actually go for my “ad pod” plan, especially if it doubled as a paperweight or a radar detector (the latter being a banned item for Google advertisers, but I digress).
It’s important to stress that Google has been impressive in its awareness of such issues, and in its willingness to experiment and to roll out new features that actually work in the real world and do emanate from principles like radical simplicity and from asking bold “what if” questions. Google knew classic content targeting was suboptimal, and asked: “What if we made it very advertiser friendly and priced every click in part based on the likelihood of that ad making money for that particular advertiser? It will be a self-learning system, and we’ll call it Smart Pricing.” Google actually did this – they pulled it off, and it worked pretty well. But due to circumstances, it just wasn’t the type of fundamental reorientation advertisers really needed, though it did markedly improve their economics. It turned out to be a brilliant technical solution grafted onto a flawed network. It also directly contradicted the principles of transparency and efficient markets, as radical simplicity and arbitrarily-designed pricing models inevitably do. Radical simplicity and arbitrary design works a lot better for a physical consumer product, let’s say, than it does for a complex system like a market.
Brilliant band-aids, but still band-aids
The open-ended nature of the ad network, and the way that a spike in content targeting clicks can not only alter overall campaign performance but “borrow” budget share from search (assuming a reasonable budget limit on a campaign), leaves advertisers vulnerable to the unknown. Case in point: a friend with an AdWords campaign that follows direct marketing criteria for its highly technical product recently received 100, then 500, then 10,000, then 50,000 clicks from a major social networking site, in a single ad group, in spite of a relatively low bid (40% of what was being bid on search), further discounted by Smart Pricing. 50,000 clicks at 17 cents is still $8,500 worth of unwanted advertising. Not a single click converted; the situation cropped up in such a way that the non-contributing inventory was not caught for awhile. Going forward, the advertiser can exclude this site, but they are left wondering when the next “incident” might occur. Advertisers want to plan proactively. They do not wish to play “rogue publisher whack-a-mole.”
All this incident proves is that our old friend, Content Targeting circa 2003, is still in place despite the many fixes.
The Site Targeting flavor of content targeting appears to be an interim solution only – a response to complaints about Classic Content Targeting, but not a particularly compelling design in its fundamentals. You can already tell Google was grasping for a good fit here by how many fixes to Site Targeting were implemented or contemplated. The pricing model was set at CPM instead of CPC, and then the minimum bid allowable kept dropping, as I chronicled with each price drop. At a later stage, beta tests of a CPC version of Site Targeting were introduced.
What none of these band-aid fixes seem to do is to provide the advertiser and publisher with a transparent marketplace that truly maximizes economic efficiency, as the initial search ad platforms did. And part of the reason for that is because of the low quality of the majority of the publisher inventory featured on the network. If nobody seems motivated to engage in bidding wars for particular publishers’ offerings, it’s because they fear it won’t be worth much to them. Thus, publishers never get the type of gratification they need to put their full confidence in AdSense. Whether it’s site targeting or content targeting flavor, Google content targeting seems to present itself as what some observers lament as “yet another ad network.”
The problem with ad networks
According to Jay Sears, SVP of Strategic Products and Business Development at ContextWeb, “most ad networks are remnant ad networks, built to favor direct response advertisers.” So by extension, a network that is not particularly friendly to publishers will likely elicit publisher behavior that is unfriendly to advertisers. “Publishers have no control on the economics,” argues Sears, so they typically provide only “remnant inventory” to ad networks.
This means they choose to monetize premium and middling content in other ways. So advertisers scouring, say, Google AdWords’ content targeting network for quality inventory, are going to be looking very hard indeed.
Wouldn’t it be better if advertisers were enticed to take the content program more seriously — as seriously as the search program — because it offers some really transformative level of functionality, and access to higher quality publishers and more information about the auction process?
Two years ago, when Google was making only incremental changes to its site targeting and content targeting programs, I started adding a slide to my presentations (at SES New York, etc.) arguing that a new type of platform was needed to help advertisers and publishers truly communicate with one another. (I alluded to this only briefly in blog posts — the need for advertisers and publishers to “transact based on something closer to perfect information.”) What I was getting at, without knowing it, was the emerging concept of a true ad exchange. Today, ad exchanges like Right Media, ContextWeb, adify, and DoubleClick Ad Exchange (which will be a Google offering in some form, if their acquisition of DoubleClick passes regulatory muster), are the talk of the industry – and for good reason.
When you look at the ad exchange model as the model for buying and selling targeted online advertising, suddenly the jerry-rigged Content Targeting and Site Targeting programs look creakier than ever.
Ad exchanges are so promising, they deserve separate treatment, so I’ll drill down on them in a n upcoming column. Suffice to say that with its DoubleClick acquisition, Google has acquired significant food for thought.
Opinions expressed in the article are those of the guest author and not necessarily Search Engine Land.