Making Paid Search Pay Off

Take Coke vs. Pepsi, translate the concept into the world of paid search marketing, and you`ve got Google and Yahoo. Like the two cola giants, the two engines are the dominant brands and they`ve saturated their marketplace. And like Coke and Pepsi, at first pass they may seem little different--at least in terms of their audience and their ability to do what pay-per-click search engines do: connect advertisers with customers.

But subtle differences exist and the question for retailers is: How much impact does choosing one over the other have on a search engine marketing campaign? More specifically, are those differences ever sufficient to justify betting the whole farm on one engine over another for pay-per-click campaigns?

In what`s essentially a two-horse race (because paid search results on MSN.com, the third major search portal, are provided by Yahoo), the short answer from search marketing consultants is: No. Marketers want to be on every engine they can. But a more complete answer to search engine strategy is more complex and nuanced than that, and it stands to become even more so as the search engine marketplace expands.

It`s both

"Now, any major advertiser is crazy not to be on both Google and Yahoo, because there`s just no disadvantage," says Nico Brooks, director of search technology at digital marketing technologies provider Atlas, a unit of aQuantive. "On a campaign of 1,000 viewers and 10 keywords, you`re better off halving the size of the campaign and taking the five best-performing campaigns to both engines than picking one engine over another."

In fact, most big pay-per-click marketers play on both engines, and often with the same keywords. Though there`s significant overlap in the users reached by the two engines, there are also subtle differences in audience demo- graphics, psychographics, ad position determination and other factors that account for the fact that the same keywords don`t perform identically on different engines.

One of the most significant differences is in how Google and Yahoo rank paid ad listings on the page. Yahoo`s model allows a straight bid for position. The way to be in the top position under Yahoo`s sponsored listings for a given keyword is to pay the highest cost per click on that word. Some PPC marketers prefer this model because they like to know exactly which position they are getting for what they are spending and because it gives them the power to obtain a particular rank on the list, if they are willing to pay for it.

Google uses a relevancy algorithm that blends what the marketer pays per click with the number of clicks a listing gets from users to determine the ranking of paid results listing for a given keyword. The result is that, unlike on Yahoo, on Google marketers can`t control whether they`re in the top positions simply by boosting their payment per click.

"We treat click-through rate as a vote by the end user," says John McAteer, head of Google Retail. "By clicking, users are telling the advertiser what is most important to them." McAteer says it`s Google`s belief that ad ranking delivers more meaningful results to searchers. "The proof would have to come from advertisers working with both Google and Yahoo to see what works best for them," he says.

Check out the difference

Director of the retail search category at Yahoo Search Marketing Diane Rinaldo points out that Google`s model favors marketers that have a strong click-through rate. "With our model, they bid for position, but then, those marketers with strong click-through rates don`t have the same advantage they have on Google," she says. As a result, she adds, some marketers view Yahoo`s model as an advantage, some as a disadvantage. "It`s the same with AdRank," she says. "Some love it, some hate it."

Beyond the difference in how they position ads, experts say some audience differences among the engines are worth calling out. For one, Google gets the most searchers. In June, for example, Google`s total audience in unique visitors for paid and unpaid search across Google.com and the sites to which it distributes search results was 79.96 million, compared with 64.03 million at Yahoo sites and 47.77 million at MSN sites, according to comScore Networks Inc. AOL had 34.06 million searchers, while Ask Jeeves had 39.82 million.

Google`s greater traffic volume is a key part of how it positions itself against other engines. "Google pushes its relevancy and sheer traffic numbers. They do still have the majority market share and they want to tell you about how many people you are going to reach and how much you are going to be out there," says Tim Kauffold, director of client services at search engine marketing company OneUpWeb LLC.

The engine audiences differ slightly in gender composition. Among the top five engines, the search audiences at MSN, AOL and Ask Jeeves skew slightly toward female users. In June, 51% of MSN search users were women, as were 51.9% of searchers at AOL and 53.7% at Ask Jeeves. At Google, however, the audience skews towards males: 51.5% in June, according to comScore. Yahoo`s search audience was split almost exactly between men and women.

"The differences aren`t huge, but with 48.5% of Google users being female versus 52% at AOL, that equates to a pretty big delta when you`re talking about the entire Internet population," says James Lamberti, vice president at comScore. That may have implications for marketers targeting a male or a female audience.

Buying power

Lamberti notes other differences among engines found in comScore`s data. ComScore`s buying power index--the average of any given demographic group`s propensity to buy online--shows that in June, users of MSN search were 48% more likely than the average Internet user to buy online, versus Google search users, 42% more likely to buy; and Yahoo search users, 31% more likely. Ask Jeeves search users were 17% more likely to buy than the average Internet users, and AOL Search users, 3% more likely. "That`s important to showing the value of the consumer and the fact that they have money in their pockets," Lamberti says.

Lamberti also says comScore sees some suggestion of different consumer intent at different engines. At the portals--AOL, MSN, and Yahoo-- comScore sees less challenging search activity and the use of simpler search terms. By contrast, Google brings in more complex search queries. He adds, "What that suggests to us is that when consumers have a more challenging research task, they may be thinking about Google as a way to get that done."

Data and anecdotal information from search engine marketing companies point up other subtle differences among engines. With a few exceptions, "We see better conversion rates for our b2b clients in Google`s PPC search ads," says iProspect.com Inc. CEO Frederick Marckini. By contrast, he estimates two-thirds of his company`s b2c retail clients perform better in Yahoo. However, iProspect`s data, gathered from analytics attached to its automated bid management agent, iSEBA, show that the difference in conversion rates is slight, perhaps half a percent. A spread of that size is meaningful or not depending on the marketer`s going conversion rate. "If your conversion rate is half a percent, it means a lot to double it," says Ben Perry, manager of paid search at iProspect.

More than one way to measure

And even where differences in direct conversion are identified among engines, they`re not the only way to measure ROI, Lamberti points out. The more specific the search term, the more likely it will result in a direct conversion. But many consumers use only broad terms to start their research online, eventually taking their search offline and buying in a store or from a catalog. And whether the purchase eventually is transacted online or off, it may be weeks or months after the initial click occurred, representing a latent ROI attached to the earlier, broader search term that first introduced the buyer to the product or brand. "In search, a lot of ROI is left on the table," says Lamberti.

It`s a dynamic that`s equally true for search marketing campaigns across all engines. That highlights the view, shared by many search marketing professionals, that from a marketing value perspective, what the top-tier engines have in common-- a platform that reaches millions of potential customers who declare their interest by what they choose to click on--is more important than any differences. It`s what strategy and not which engine that wins or loses, they say.

"You always want to spread your money and find the best pieces of both engines," says Kevin Lee, CEO of search engine marketing firm Did-It.com LLC and chairman of the recently-formed Search Engine Marketing Professional Organization. How to find where those pieces are? Trying out and measuring what produces clicks and sales and what doesn`t, across keywords, landing pages, and ad creative, are the best ways to gain an understanding of the potential of a keyword on a particular engine, say the experts.

For example, Google`s paid listings format doesn`t have a lot of space for copy, and users must click on the ad to read the full text after seeing a few initial words. By contrast, Yahoo`s format allows room for more copy, giving marketers the opportunity to add words that could entice a click, such as "free shipping," for instance. That`s one way that what`s in a listing, rather than audience demographic or psychographic attributes of the engine it runs on, makes the difference.

IRCE 2008 Conference Multi-Media CD-ROM

Core keywords

Lee advises clients to start a paid search program by determining a core set of keywords, then figuring out the metric to be used to measure campaign success, such as cost per order, ROI, or venue per dollar spent. Then, set a benchmark that represents the tipping point at which the keyword campaign works or doesn`t work. Next, marketers should consider what creative their customers want to see. One Did-It client, Haband, makes men`s trousers that customers think of and search for as "slacks" rather than "pants." That shows that knowing how to write the creative that customers associate with the brand or product makes a difference between campaign success and failure.

Developing a testing strategy is core to search campaign management, on any engine, Lee says. "Until you go live and create a feedback loop that tells you what is working and what is not, you won`t know whether `red cashmere sweater` is worth more on Google or Yahoo," says Lee. "Put campaign analytics in place and you will see that for every thousand visitors that came off `red cashmere sweater` in Google and every thousand that came off Yahoo the conversion rates were different."

Handling paid search across two major engines is already a handful, but the logistical demands of handling paid search campaigns across the marketplace are about to change, because the PPC marketplace itself is changing. MSN.com is expected to launch its own paid search product, eventually replacing its paid search feed from Yahoo. Though MSN.com declined an interview request, most industry observers believe that will occur within the next few months. Some believe it could be in time for the peak of the holiday shopping season.

Ask Jeeves, the fourth largest search engine according to June figures from comScore Networks, announced in August it will launch its own paid advertising network, using a keyword auction model similar to Google`s and Yahoo`s. Ask Jeeves will run its own paid listings alongside those it currently gets from Google under a revenue sharing contract with Google that runs through 2007. Eventually, most industry observers believe, Ask Jeeves` paid search offering will stand alone.

The management headache

In addition to listings on Google.com itself, Google has another big distribution partner for paid search in AOL.com. Given what`s poised to happen at MSN.com and what`s already happening at Ask Jeeves, will AOL at some point go in the same direction and split off a paid search offering of its own? That`s speculative, but "It would make sense that most of the major publishers will start selling their own search ads because they don`t want to give someone else a piece of the pie," says Brooks.

It`s a scenario that could soon leave online marketers facing the prospect of managing separate paid search campaigns on at least four of the top five engines, territory they could cover previously by dealing with two engines. And that doesn`t even count smaller engines looking to differentiate themselves with strategies such as ad delivery by user segment.

Though pay-for-performance search marketers needn`t pay up unless there`s a click, the sheer labor of managing campaigns across more engines stands to create a potential resource issue. For some marketers not already automating campaign management with bid management tools and services, it could accelerate a move in that direction. "You`re not going to be able to do this with a spreadsheet," says Marckini. "There are still people with significant budgets trying to manage their campaigns manually."

The engines themselves are aware that managing campaigns across more search engines and the different proprietary technologies attached could mean more competition for marketers` budgets. "Resources aren`t unlimited," says McAteer. But McAteer and other search engine operators also know what will win in any resource squeeze, and online marketers will find that answer in their own experience. "Building a better search tool, having more relevant ads showing up against more relevant searches, and continuing to drive ROI for advertisers--that`s really the only differentiator that has value to a marketer," McAteer says. l

mary@verticalwebmedia.com

Desktops v. notebooks at HPshopping.com

For HP Direct Inc.`s b2c e-commerce site, HPshopping.com, the demographics of the customers it acquires through paid search campaigns at Google and at Yahoo are much the same, with little difference apparent across basic audience attributes such as income level, gender and age. On certain keywords, however, what are generally very similar campaign results across both engines start to diverge. The buyers that come in through Yahoo, for example, tend to be more frequent purchasers of customizable desktop computers, while Google buyers skew toward customizable notebooks.

HPshopping.com director of consumer marketing Catherine Paschkewitz can`t prove precisely why that`s the case, but she and her team have developed a pretty good theory. Snap on the filter of a different kind of audience attribute--in this case, an affinity for technology--apply it to particular keywords, and the picture starts to become clearer. Noting recent Jupiter Research findings that Google is the preferred engine among the more sophisticated users of online search, Paschkewitz postulates a match between those users and users more likely to buy notebooks online than through other channels.

"We see that the people who buy notebooks from HPshopping online tend to be more technologically savvy--they tend to buy their products more on the web than through the call center, for example, versus what we see on desktops," she says. To support the theory, Paschkewitz checked it against keywords associated with other high-tech products such as handhelds and iPods, and found similar results; in fact, finding even higher conversion rates and average order size from those keywords on Google than on Yahoo.

And that`s information HPshopping, which uses the search engine marketing services of Performics Inc., taps to guide its search engine strategy. Managing its paid search campaign across engines to an overall cost per acquisition target, it already dials bidding on different keywords and different engines up or down, based on their performance, so as to stay within that target range. Where data such as the better performance of Google on keywords that attract the tech-savvy comes in especially handy is when HPshopping can harness it to additional business objectives.

"Say we suddenly needed to find additional growth in a particular category; notebooks, for instance," says Paschkewitz. "What we`d potentially look at is, if we have additional dollars to put on the table, should we invest them in Google versus Yahoo because we tend to drive more online notebook sales from Google?"

A search engine strategy geared to another difference between the engines--the bid model and how it affects the performance of different types of keywords--might even help find those extra dollars. Yahoo`s keyword auctions are straight bid-for- position; Google`s AdRank formula incorporates relevancy as determined by the volume of user click-throughs in how ads are ranked in results listings. The net effect for HPshopping is that while both engines produce similar results on brand keywords in terms of conversion and average order size, brand keywords are a more efficient spend on Google.

"On Google, you don`t necessarily have to bid as high to get that better position because of the relevance factor. On our own brand terms we are always going be very relevant," Paschkewitz notes. By spending less for a high position on Google than on Yahoo when it comes to brand keywords, she adds, HPshopping frees up budget it can use to bid more aggressively on other keywords or categories.

As HPshopping is finding, identifying differences between search engines--even at the seemingly small level of a difference in the performance of individual keywords--can give search marketers more flexibility in campaign planning, and the opportunity to drive more return out of the same search budget. "A total CPA across our interactive program is what we look for," says Paschkewtiz.