Nội dung

CHAPTER 8: Writing Winning Ads This example uses the word easy, which tests out better than a few other adjectives describing your software. It says you’re selling a firewall, which is exactly what you’re selling and what that user typing firewall product reviews is probably looking for. And it introduces doubt about the quality of the industry leader (you have proof on your website). They happen to have 61% market share (you only have 4%, so comparing yourself with them is inevitable), but their product is rather expensive and difficult to use. This piques the interest of real prospects—enough to click on your ad—but because it’s fairly clear about what you’re selling (and mentioning BigCorp Terminator makes it even clearer, as it further “marks” this territory), you don’t attract the confused or the vaguely curious. Plain and simple is often the most effective ad copy to use. While every attempt may not be a winner, as long as you hit the tone right and write a clear ad, you’ll know you’re on the right track if your ad is attracting a strong CTR and, ultimately, if your post-click tracking shows these clicks converting to sales, leads, and registrations. Tracking users after they click is covered in detail in Chapter 10. Here’s another example that shows the difference in performance when you do a better job of understanding your target audience. The first ad uses a generic and somewhat hyped-sounding headline, with some impressive but unverifiable claims: Hot Stocks Uncovered Our portfolios were up 67%, 58%, & 34% for 2003. Free trial & report. www.FindProfit.com The second ad uses more specific language and touts the founder’s credibility: Short-Term Trading New site from Raging Bull founder RT commentary. Free report & trial www.FindProfit.com The first ad pulled only 0.5% CTR; almost certainly too low for the Quality Score gods’ taste. The second ad pulled a much healthier 1.3%. A third ad was a hybrid of the two, using the short-term trading headline but sticking with the claims about portfolio gains. As you might expect, it performed somewhere in the middle, at 0.8%. The underperforming ads were dropped, and a couple of additional tweaks were then tried with the winning ad. Tone, directness, and credibility helped this new information service find its audience. Of course, we only know this in hindsight. No one could have predicted in advance that Bill Martin’s and Matt Ragas’s past experience in creating investment-related content at financial discussion site RagingBull.com would resonate so concretely with the target audience. Bill and Matt gained some fame in the dot-com boom era as college students who founded a stock discussion site that grew rapidly from inception in 1997, attracted $22 million in investment from CMGI and CNET, and went on to be acquired by Terra Lycos in 2000. Many savvy online traders seem to know who Bill and Matt are because they remember trading stock information at Raging Bull. Based in large part on their paid search campaigns, the paying 219 220 Winning Results with Google AdWords subscriber base of FindProfit.com continued grow from 2004 through December 2007, when the publication merged with another in its field, the Bull Market Report. Credibility is a guiding principle of copywriting. Be careful not to obsess about ad copy so much that you change it when you perceive it to be stale. Remember the maxim that you should stick with a campaign as long as your accountant likes it.5 Who cares if you’re bored by a certain phrase or angle; who cares if your friends wonder if you’re ever going to put up a new ad? If it ain’t broke, don’t fix it! Continue to test new theories; you may find something that works better. Just don’t make changes simply for the sake of making changes. One of my European colleagues regularly tries to argue that ad “freshness” degrades, hence requiring that new and “fresh” ads be actively written. I agree that seasonality and changing trends might severely affect user response to some ads (fashion and counter-obsolescence imperatives abound in some industries), but degraded ad performance doesn’t occur because of “staleness” as some kind of general rule. There are many reasons for fluctuations in campaign performance, and isolating the impact of ad freshness in situations where it does seem to matter is nearly impossible. One approach that withstands the test of time is the appeal to your customer’s self-interest. Saving money, making money, winning at something, getting a deal, being able to make installment payments, alleviating an annoying headache,6 beating a competitor, making a friend—these are all reasons for users to decide that what you have to offer may be the ideal choice for them. It’s possible to go too far in analyzing the underlying motivations (money, sex, love, beauty, simplicity) that supposedly drive every purchase. At the extreme are the ads you see from time to time in technology magazines somehow trying to convince readers that supermodels will be impressed if they would just acquire the latest in encryption technology. It would be funny if it weren’t so prevalent. Surely there must be a better way to convey a product’s benefits. (And yes, I’m aware that domain registrar GoDaddy has become notorious through the use of a “sex sells” strategy, but let’s not assign too much causality to this in either GoDaddy’s, or your, business. They’re also benefiting from heavy investment in every type of media exposure, and the “winner take all” effect in new product categories devoid of household names, that used to be so prevalent in the television-industrial complex era. If Apple computer sales increase this year and its operating system gains market share, should we conclude that a folksy ad campaign that caricatures the uptight competitor, almost to the point of making the competitor seem lovable, “works”? Again, it would be tough to generalize based on this sequence of events, given that your competitor might not, like Microsoft, be a household name who released a new operating system that absolutely bombed.) In some industries, exaggerated promises work. For a brief shining second, some golfers appear willing to believe that hitting a ball two yards farther is akin to conquering the galaxy. In almost any business, appeals to customers’ rational side—to improve their communications skills by buying a book, tape, or hiring a coach, for example—often work. But what motivates someone to purchase seeds for a certain mossy perennial? They desire a beautiful garden, of CHAPTER 8: Writing Winning Ads course. Do we need to know why they want a beautiful garden? Probably not. I’m not sure I even want to know. Ideally, at a certain point, the process takes care of itself: users want what you have, and you sell it to them. At times, I think it may be a blessing that Google advertisers can’t gather detailed demographic information on Google users. Unlike some advertisers, deep down I don’t necessarily want access to the proverbial electrodes to the head that might reveal every motivation that drives purchase behavior. Having access to a narrower set of user data actually makes this job easier for the average business to deal with. Addressing Multiple Priorities Writing ads would be easier if you had a single goal in mind, but that isn’t the case here. You must weigh a number of priorities. The two primary goals of an AdWords ad actually compete with one another. Sound confusing? It isn’t really. On one hand, your goal is to encourage prospective customers to click your ad. On the other hand, you want to discourage inappropriate prospects from clicking on your ad so you don’t waste money on clicks that have a low probability of ever turning into sales. But if all advertisers became extremely good at this filtering process, Google would be showing ads that got lower CTRs across the board, which would hurt them in at least a couple of ways. First, the perceived relevancy of Google ads would be in jeopardy, but more importantly, Google would make less money per page of search results served, because fewer users would be clicking on ads. As a result, as I already emphasized in Chapter 5, Google designed the system to reward you for a higher CTR, along with some other relevancy factors. All else being equal, you should test your ad copy with a view towards generating higher CTRs on ads. If your highest-CTR ads turn out to perform worse from an ROI standpoint, then you’d need to revert to the one(s) that offered the optimal trade-off between CTR and ROI. That’s a matter for experimentation with the help of post-click tracking data, discussed in Chapter 10. Balancing CTR and ROI When it comes to gauging the performance of your ads, you may be thinking that a higher ROI is the holy grail you’re pursuing. But this is an oversimplification, given how vitally important CTRs are in allowing you to gain favorable ad positions with lower bids and ultimately improve ROI. Remember, you must encourage prospective customers to click your ad while discouraging inappropriate prospects from clicking your ad so that you don’t waste money on clicks that have a low probability of ever turning into sales. While ROI will be the ultimate arbiter of success, realize that in the early going, you’ll probably need to move CTR higher up your priority list than you might expect, or wish to, because of the strong incentive system built into AdWords that favors high-CTR ads. 221 222 Winning Results with Google AdWords Other goals for your ads might revolve around credibility or leadership. You might wish to use certain phrases or wordings to reinforce your company’s brand awareness, regardless of how well the ads perform over the short term. Management has every prerogative to sacrifice short-term sales numbers for long-term strategic goals if they believe that to be their mandate. Sometimes, then, ads that aren’t really about selling anything directly might make their way onto the user’s radar screen. Some companies wish to increase “mindshare.” One company I work with allocated $800,000 for 2008 as part of an effort to demonstrate “category leadership” for certain seasonal transactions. Such efforts are part of a multifaceted effort to shift customer perceptions of a brand over time. If turning a big company’s strategy around is a little like steering an ocean liner, then so is changing deep-seated customer perceptions of such companies. To be sure, a sampler of searches for dozens of terms will turn up few big-brand advertisers, and even large advertisers appear to be drawn to AdWords mostly as a highly measurable, direct-response medium. For the time being, most of the “presence and awareness”-style ads are from government and nonprofit organizations. On a search for pesticides, it’s fascinating to see several nonprofits and government organizations competing for awareness against companies (such as Orkin and HengDongChem7) who actually sell pesticides or pest control solutions. For the time being, “awareness”-type advertising is relatively rare on AdWords, but as high-profile politicians gain ink for their savvy use of online promotional opportunities, awareness of raising awareness through AdWords is growing. A fascinating study in opportunism comes through a search for trendy Democratic presidential nomination hopeful Barack Obama.8 The top-position ad, in premium position, is from BarackObama.com, the senator’s official site. The landing page, my.barackobama.com, is tailored to generate signups to be part of a list of people interested in the campaign, including notifications of rallies and events. It doesn’t stop there. Google itself, through its subsidiary, YouTube, has placed an ad suggesting that you “watch videos from each of the 2008 presidential candidates.” At the time of this writing, over 14,000 people had subscribed to the Barack Obama page on YouTube that collects videos of his speeches. (Just in case you were getting your hopes up about the depth of political engagement in America: by contrast, a video by Obama Girl, a young singer professing her love for Barack Obama, had been viewed over 4 million times. Obama Girl does say “Hey B., … I was just watching you on C-SPAN” on the video, so maybe there is hope for the fusion of politics, sex, synthetic drum tracks, and geekdom after all.) An opportunistic project, the Iowa Global Warming Campaign (supported by multiple interest groups and nonprofit organizations), weighs in with an ad exhorting users to “learn how Barack Obama proposes to solve global warming.” Finally, a retailer is promoting a “super soft Obama t-shirt” that says Obama is My Homeboy. This is also the title of the ad. The copy on the landing page says “Barack Obama is like a latter day JFK, but way more accessible.” Evidently, in the case of Barack Obama, direct marketing opportunities take a backseat to awareness-raising. But in the end, we see a fascinating blend of the two.9 Why don’t big companies do more with AdWords to raise general consumer awareness of their adaptations to new trends? The effort would be far less costly than all those TV campaigns. They’re starting to get it, but slowly. Some ads will have multiple roles, attempting to increase CHAPTER 8: Writing Winning Ads sales, but only indirectly, by changing consumer perceptions. What about a company like McDonald’s, which recently underwent a successful turnaround from a declining purveyor of supersized calorie-laden meals, to a solidly profitable company that can convince people to buy $6 salads at a drive-through where they once bought a $2 burger? (“Mickey Dee’s” is doing very well, financially speaking, these days. Its stock price has appreciated steadily over the past five years and it now pays a healthy dividend, still yielding 2.5% in spite of the healthy stock price). Many customers know about the salads now, but it wouldn’t cost that much to add to this awareness with a simple AdWords campaign. Online promotions by large companies have often been obsessed with complex schemes involving coupons, loyalty, contests, and so on. They require high-powered thinking. But given the low cost of a paid search campaign, a basic campaign merely alerting the public to the sea change in McDonald’s product mix—for example, adding awareness of the quality and health benefits of their new salads—would be worth every penny, even if no particular tracking scheme were built on the back end. For all the money companies spend on expensive television and billboard campaigns “getting their message out,” paid search seems like an incredibly cost-effective way to achieve that goal, one highly targeted customer at a time. And those kinds of avid customers create word-of-mouth advertising, which is free. A campaign of this nature wouldn’t need particularly clever ads—“learn more about our delicious new salads” or “50% off coupon to try our Mandarin Chicken salad—limited time offer” would be enough. It would be the power of the salad-related keywords and the fact that the clicks were so inexpensive to reach out to salad enthusiasts that would be the real driving force behind such a campaign, not necessarily the ad copy. To give you a better sense of the current economics on such a campaign, let’s say McDonald’s managed to garner a million ad impressions on their AdWords ad at ten cents per click, for a cost of $1,000. Their effective cost per thousand ad impressions, or CPM, would be $1. A million impressions for a comparable television ad would cost significantly more. According to recent press reports, CPM rates on 30-second ads on network prime time are running at about $16. Shorter commercials in off-peak hours might cost 40% of that, but ads on specialty channels can cost significantly more. At a relative “bargain” CPM of $7, a 15-second spot for McDonald’s would cost $7,000, or seven times more than the AdWords ad. This doesn’t factor in heavy production costs for TV advertising, nor does it measure its relative effectiveness. Search ads are significantly more effective and measurable even than other forms of online advertising, yet on a lot of highly targeted keyword inventory, you still don’t need to pay an outrageous premium. Keep in mind that those “impressions” for a McDonald’s spot are impressions of the ad by anyone with their TV tuned to that channel. Search ad impressions are restricted to people who have actively typed a query into a computer. The difference is difficult to quantify, but no matter how you slice it, the contrast is stark. Performance aside, the disparity in price between TV advertising and search advertising (7× in this example) is great enough that more big companies will find the opportunity worth investigating. Some time after pondering this example, I did spot a special McDonald’s promotion being tried in an AdWords ad for McDonald’s Canada only (triggered by a search on the keyword McDonald’s). The campaign didn’t make much sense to me—coupons for low-end sandwich meal specials for different days of the week—but I was intrigued to see larger companies beginning to experiment in this way. 223 224 Winning Results with Google AdWords Currently, forward-thinking campaigns by large corporations concerned about protecting their brand image or sending a general message to consumers are relatively rare in the paid search ad space, which leaves many opportunities for nimble smaller companies. Maintaining Accuracy One of the most important pieces of advice I can offer is, be meticulous. The image of your company as a provider of quality at every level of the operation certainly won’t be helped if you have misspellings in ads or use nonstandard punctuation. Some companies are willing to take small hits in short-term performance in order to maintain a certain image. For example, you might prove that capitalizing the first letter of every word in your body copy (now allowed by Google) slightly increases CTRs and ROI. Yet your CEO and/or shareholders might think this looks amateurish. So be it: there is nothing wrong with sticking to principles for long-term image purposes. (If that weren’t the case, every television spot would probably end with the directmarketing pitch, phone number information, and other calls to action favored by, well, direct marketers on late-night TV or infomercials.) Some advertisers write their Google ad copy as if no one is watching. Obviously, just the opposite is the case. Don’t make spelling and grammar mistakes and don’t write like a four-year-old! Accuracy is a guiding principle of copywriting. You should generally avoid false or unverifiable claims. Google Editorial might clamp down on you if you use unverifiable superlatives such as best, cheapest, or longest lasting anyway. But no matter what Google says, it’s up to you to decide whether it’s in your company’s best interest to shade the truth, misrepresent your product, or misreport your pricing. On the other hand, if you’re sincere, there is nothing particularly wrong with making true but difficult-to-verify claims, and Google doesn’t prohibit them all. Depending on the product or service you’re advertising, you may be under a closer microscope, remember. “Accidentally” inserting fresh for a frozen shrimp delivery can land you in regulatory hot water, for example. And you aren’t allowed to advertise insurance (for example) in a state where you aren’t licensed to offer insurance. But your customers are probably going to be dissatisfied soonest, regardless of the looming and generally toothless threat of federal regulation. Advertising fresh shrimp when it’s really frozen is literally a classic example of the improper information “scent” (discussed in Chapters 10 and 11) that can cause dissatisfaction among searchers, lower conversion rates, increase “bounce rates,” and cost you money. If you’re a consultant, or work in a company with a clear chain of command that requires approval of written materials, leave new ads paused until the appropriate person approves them. For smaller companies, this offers another advantage. While the big guys are waiting for approval, you can beat them to the punch and test new ads at will. CHAPTER 8: Writing Winning Ads Getting the Most Out of Your AdWords Ads Possibly the most consistently successful copywriting tactic is to write more ads, make them more personalized, and make sure that more of your ad titles and/or body copy match (or come close to matching) the keywords the user types in. This dovetails with the admonition to “organize, organize, organize” in Chapter 4. Clearly, if you sell a fairly diverse line of (let’s say) diamond jewelry, your performance won’t be as good if your whole campaign relies on a single generic ad: Diamond Jewelry Store Earrings, bracelets, watches, more Check out our weekly specials! This ad might be fine for those searchers who have specifically typed diamond jewelry, but if they typed diamond bracelets, it would probably be outdone by a more specific ad run by a competitor. As I’ve said before, users most often gravitate towards the ads that match more closely to what they’ve typed into the search box. To improve on the performance of this ad, you’d simply write ads that are more targeted, such as this one: Designer Diamond Bracelets Diamond bracelets by top designers Dazzling one-of-a-kind items! The effectiveness of an ad depends on a large variety of factors. As you ponder alternatives at the outset, don’t overthink. Just plunge in and write an ad or two that you believe will attract your target audience, and test and adjust from there. In this case, I mentioned one-of-a-kind items to reinforce the image of designer jewelry so as to attract a higher-spending clientele. A Technique for Ad Refinement in Stages To make this process a little more systematic, I’ve developed a rough two-stage process that is meant to get your initial “good” ads up to “very good” (step 1), followed by a data-intensive phase that can raise “very good” performance to “even better” (step 2). Step 1: Discover the Main Triggers You don’t want to start off aimlessly testing piddling little differences in your ad copy, just to prove to yourself that the ad rotating function does indeed work. What you’ll be looking for are significantly different approaches to getting your message out to prospective customers that can teach you something you can take away as you use the winning approach as a basis for another round of refinements. For example, you might not know in a certain context whether fast shipping, low prices, a big selection, or some other major element of your retail experience for customers is a particular trigger for them. Don’t expect to do exhaustive research here. In one case, I pitted three potential 225 226 Winning Results with Google AdWords triggers against one another: price, selection, and the ability to do research and access schematics on the website. Shipping offers became a permanent feature of our ads, as their efficacy had been demonstrated in separate experiments. We discovered that selection consistently (though not always, across all products) beat price and research as the best trigger, when combined with the shipping offer, which referred to the speed of shipping. Later, I tinkered with subtle wording variations: “big selection”; “huge selection”; “full selection.”10 When you do this for your own business, try to write three or four ads that are distinctive enough to return information to you along the lines of “what are they thinking?” and “what causes the right people to click and buy?” Step 2: Multivariate Testing You might want to try a couple more rounds of tests based on new ideas that fit in roughly with the Step 1 type of research. Once you’re armed with a lot more information about what triggers seem to be consistent, you enter the refinement stage. Multivariate testing is a systematic approach to testing ad variations. Let’s say there are four elements of the ad you want to test: the headline, a discount offer, a certain punctuation quirk, and the use of a brand name in the body copy. Let’s say you’ve got three versions of the headline, two versions of the discount offer, yes or no to the punctuation quirk (such as a single exclamation point), and yes or no to using the brand name. Doing what math whizzes refer to as a full factorial multivariate test—that is, a test of all possible variations of your ad—would give you: 3 × 2 × 2 × 2 = 24 ad variants Multivariate testing is becoming commonplace for landing page designs, but it’s possible to do it with ads, as well. With the above method, you can often generate amazing results testing 16, 24, or 32 ads as part of a single multivariate test. Part of the power of such testing lies in discovering unexpected variable interactions. Independently, a certain headline may not perform all that much better than the other two. But in one particular combination with two other elements of your ad, you might discover that headline #2’s performance is consistently better (let’s say it’s ad variant #19 that hit the jackpot in this way). You never find this stuff out unless you test all variants. The drawback to a full multivariate test is that it’s infeasible with limited click volume and sales volume. If you’re doing a good job of organizing your ad groups to be quite specific, any individual ad will not necessarily generate enough sales volume to generate meaningful feedback in a reasonable time frame. Enter a modified form of multivariate testing that allows you to generate results more quickly and with fewer test combinations, called fractional-factorial testing.11 Advanced mathematicians have developed shortcuts that create meaningful results with many fewer testing variations. One method, Taguchi testing, is quite common. It’s also quite controversial, because it might miss a truly winning combination, but I’m willing to try anything that will work reasonably well. One Taguchi-based ad-testing protocol I use takes a seven-element grid with two values for each variable (64 total variants) and provides a recipe that gives you 8 ads to test instead of 64. You can muddle through this “by hand,” as it were, on a useful site called Blair Gorman’s AdComparator (http://adcomparator.com). CHAPTER 8: Writing Winning Ads The beauty of multivariate testing is that you can answer several questions at once, often more accurately than through serial A/B tests of different theories over time. Implemented across an account, you’ll likely see patterns of performance of specific ad elements repeated time after time in different situations. That should be teaching you something. Ad testing is both art and science, requiring creative inputs and at least a rough experimental method. While you don’t need to be an advanced mathematician, the ability to apply some of the math-based tools available is a must for superior performance. Getting Help from the Experts Can we learn about what works from professional copywriters, for example, Bob Bly and Nick Usborne? Having reviewed their work, I’d say yes and no. Consider that many AdWords advertisers (myself included) stumbled into the process with no formal background in writing marketing copy (but plenty of general experience in thinking, writing, persuading, and testing ideas). By all means, read the works of such authors—especially Usborne’s Net Words: Creating High-Impact Online Copy (McGraw-Hill, 2001), which is particularly accessible. But keep in mind that some of their teachings about writing for the Web only apply to websites and e-newsletters, both of which provide considerably more room to develop your message. For other background materials that help achieve better bottom-line results from ads, more and more marketers are delving into the literature on “scientific advertising,” which begins with a very old book by Claude Hopkins entitled (not surprisingly) Scientific Advertising (Chelsea House Publishing, 1984). It includes various split-testing practices that have been adopted by many direct marketers (direct mailers) over the years. A more recent book by Seth Godin, Survival Is Not Enough: Zooming, Evolution, and the Future of Your Company (Free Press, 2002), helped me understand the evolutionary process of more rapid testing and improvement that seems to be required in modern companies. With so much material available, it’s easy to get carried away. Not only have social scientists been writing about this type of thing for centuries, but since the “Total Quality Management” movement, the market has been flooded with books about testing and improvement of manufacturing processes. This carried over to other fields such as web design. More recently, buzzwords from the manufacturing world seem to be leaking into general usage to cover almost any situation, but they don’t always apply. When your child’s T-ball coach starts talking about Six Sigma swing improvement, it’s probably time for a reality check. In any case, writings covering the rapid technological improvements at companies like Intel and Netscape—and now, Google—have been a staple of the business bookshelves and have filled magazines like Business 2.0, Wired, and Fast Company in recent years. It’s sexy to maximize performance and to “iteratively” improve your processes through a series of versions, just like chip manufacturers and software companies do. But even these companies understand that the point is not to improve on every single aspect of their operations, just the aspects that matter. 227 228 Winning Results with Google AdWords Six Rules for Better AdWords Copy While the two-step process described a bit earlier in “A Technique for Ad Refinement in Stages” narrates a plausible, staged approach to refinement, it’s also useful to keep general tips and best practices in the back of your mind. So far, I’ve discussed how principles of advertising and copywriting provide a foundation for writing AdWords ads, but one problem with following standard copywriting techniques is that Google AdWords are anything but standard. Since they are out of the mainstream of copywriting experience, a specific set of rules might be helpful as a kind of shorthand applicable to this domain. Here are my six rules for writing better AdWords copy: ■ ■ ■ ■ ■ Match the user’s query as closely as possible. Send appropriate cues by speaking the same language your target customer speaks. Filter out inappropriate prospects. Get the prospect’s “action motor” going by inserting a “call to action.” Based on CTR and conversion data, pick the winning ads among several you’ve been testing. Occasionally, introduce new ads so the testing process never stops. ■ Inject some flair or brand appeal in the process. Pick any example ad, and these six rules should be enough to get you thinking about how to write winning copy. Rule #1: Match Ad Titles to Searched Keywords The example about diamond bracelets gave you an idea of the typical process of improving on a single generic ad by writing many more specific ones so the ad titles match query keywords. (The example is drawn from a client campaign, but the type of jewelry has been changed to protect the confidentiality of their data. My tests consistently proved the power of creating a large number of more specific headlines as opposed to a smaller number of general ads about the jewelry store. In most cases, the more specific headlines had significantly higher CTRs.) This isn’t a rule that applies across the board, but in many cases where users are looking for a specific item that might be found in a retail catalog, such as a diamond brooch or a diamond pendant, ads that contain that exact item in the headline feel more personalized to the user, so they’re more likely to click. Working for clients like this, I’m often amazed at how well they can do against much larger competition simply because the other advertisers may have limited themselves to a generic ad covering, for example, diamonds or jewelry in general. Ads that have a “canned” feel to them— or more to the point, ads that don’t seem likely to take the user to a page describing exactly what he’s looking for—might not get clicked on quite as often. Perhaps you’re wondering if this tendency would continue if every ad on the page had exactly the same title. If everyone did this, and every ad contained “diamond brooch” in the headline for ads appearing next to searches for diamond brooch, it’s possible that a different sort of headline might stand out and therefore be clicked more often. For the time being, this is still a strong rule, but like any user tendency, it’s subject to change. CHAPTER 8: Writing Winning Ads Rule #2: Send Appropriate Cues to Your Target Audience A corollary to matching ad titles to searched keywords is appropriateness and sending cues to certain consumers, and that’s a bit subtler. You might not consider the mention of “oneof-a-kind” jewelry items to be filtering out bad prospects so much as it is a kind of cue that you hope will resonate with a very specific type of customer because it literally speaks their language. Some car buyers will respond to “240 hp” or “6-speed manual”; others will assume the horsepower is in line with comparable vehicles and will actually respond better to “zoom, zoom.” (I would like to state for the record that I am not one of those who would buy a car just because it goes “zoom, zoom.”) Consider using language and terms that will resonate with your target audience. Let’s say you sell housewares and your target customer is female, moderately affluent, and seeking the latest style. You’re not selling traditional dinnerware, and there’s simply no better way to convey that than just to tell it like it is, by using words like modern and contemporary in your ad copy. Clearly, different demographics require subtly different wording in ads, which can be achieved with an adjective here, an expression there. For example, even though I’m a nonexpert when it comes to martinis, and I don’t throw dinner parties, I knew that the ad copy “elegant martini misters for the discerning martini-meister” would work well for my client, KlinQ.com. I was speaking the lingua franca of the affluent, young, martini-drinking set. A light touch of tonguetwisting humor and a post-materialist (nonbasic, nondiscount) ethic were conveyed to give that demographic what they were looking for. The fact that customers quickly found what they needed, and felt that this retailer understood their needs (my ad was only a small part of that equation), quickly overcame any price sensitivity on the part of the bon vivants in the market for martini-ware. Rule #3: Filter Prospects The process of filtering out bad prospects begins with keyword selection and targeted phrasing. Make sure your keywords are not so broad as to cause your ad to appear in searches that are irrelevant to your product or service. For example, if you’re offering loans for cars and title your ad “Low APR Loans,” you’ll get people looking for personal loans, mortgages, home equity loans, recreational vehicle loans, and student loans, most of whom are not your target customers. By adding the word auto (“Low APR Auto Loans”) to your ad, you can eliminate a large percentage of those nonproductive clicks. Probably the most common problem pay-per-click advertisers face is the bargain hunter who clicks on an ad seeking an inexpensive or free product. Sometimes it just doesn’t pay to be subtle. One client from the UK was getting too many tenants applying for loans, even though the product was clearly targeting the home equity loan market and even though the ad made it seem reasonably clear that you’d need to be quite creditworthy. The problem was easily solved by making the first word in the ad copy: “Homeowners!” (A searcher would be hard pressed to misinterpret that.) As paid search advertisers, we need to be mindful that there are millions of noncustomers out there clicking away on our ads. We need to be particularly careful of those who are looking for something cheap or free. Unless you rely on a strategy of upselling people from an entry-level 229 230 Winning Results with Google AdWords product or free offer, you might need to filter the low end of the market. I realize that advice won’t apply to all of you, but the general principle of filtering does, and the “low end” (lowbudget buyers and tire-kickers) presents a formidable challenge for advertisers. Keep in mind, though, the difference between someone who is in the early phases of a normal buying cycle (just not ready yet) and someone who is simply a poor prospect for you. They won’t all buy immediately. The classic example that’s been given by companies like Google at seminars on this matter is the seller of graphics software who, due to poor ad copy and indiscriminate keyword selection, is inundated with clicks from people looking for free clip art. When I refer to the low end of the market, I’m really referring to a type of behavior that is endemic to search. Most of us constitute the low end (or “no end”!) of the market for advertisers on any given day, because we might be looking for free information, shareware, and so forth. If we click on someone’s ad listing when we’re in the “looking for free stuff” search mode, we usually wind up costing that advertiser money. Niche advertisers are looking for a relatively rare bird: a reasonably warm prospect who is likely to convert to a customer. Does this mean you should put your prices in your ads? You’re welcome to test it, but it doesn’t necessarily work. You can convey a sense of your positioning in a price range with verbal cues and brand names. Most folks have a general sense of where Pier 1, Holt Renfrew, Home Depot, Best Buy, McDonald’s, Outback Steakhouse, and Target stand in the price spectrum just from their well-known brand names. If you’re not well known, you can certainly choose your words carefully to convey a sense of price points. “Competitively priced data loggers” or simply “data loggers” might convey one message, whereas “industrial use data loggers” might convey another. Someone who is confused and seeking a speedometer gizmo of some sort for his bicycle, at the very least, is unlikely to be confused enough to click on the ad if you work in cues like “industrial.” On popular terms, sometimes even stronger language may be required to deter the mass market from clicking on your ad for a niche product. One simple word might not be enough. Though it sounds redundant, “data loggers for business and industry” might be needed to weed out the wrong sorts of browsers if data logging for personal uses like sporting activities were suddenly very popular. When too many eyes are flitting quickly across ads, it may not hurt to hit them with a couple of really obvious prequalifying words (such as both “business” and “industry” in this example). Note too that such examples are not cast in stone, because search behavior is fluid. Spikes in search activity on certain keywords can be driven by shifting trends, news items, and consumer tastes. These spikes in the frequency of certain keyword searches can cost your business, so you need to be attuned to changing search patterns and filter more aggressively if you believe that you’re paying for too many mass-market clicks that don’t genuinely want what you have to offer. I’ve worked with service-oriented clients, such as one in the web design business, who are upset that they receive too many low-budget inquiries from their pay-per-click ads. Unfortunately, though, shouting about a $5,000 minimum price tag up front is no answer to this dilemma. After all, why deter potentially good clients? Some relationships take time to develop. The process of sorting out qualified leads for a complex sale isn’t something you can expect to take place solely in your 95-character ad. While it’s sometimes useful to “cool off” nonprospects CHAPTER 8: Writing Winning Ads with restrictive-sounding messages, don’t you need to convey initial warmth (or at least relevance and an open-for-business attitude) to actual prospects? It’s all too easy to forget that there are many others competing for those customers (and advertising in the same space!), and that for these prospects, $0 is always the default amount that they plan to pay you until they get to know you a little better. When a potential customer is early in the buying cycle, you can hurt yourself by posting a price completely out of context, not least because a competitor can easily undercut you in the same space with a price that looks better than yours on the surface. Wholesale versus retail, and business-to-business versus consumer, are two other common filtering concerns. In wholesale and business-to-business, the number of target buyers is smaller. There should be no sugar-coating the fact that filtering out all the inappropriate prospects is difficult. Mostly, this should be done with keyword selection in your AdWords account. But when it comes to ad copy, you’ll want to experiment. In certain industries, terminological conventions tend to emerge, such that you can really catch people’s attention if you stay current and use the most current buzzwords. For example, in web hosting, an ad title like “Reseller Hosting” or “Co-Location Servers” would be fairly well understood by your target audience. You might soothe executive decision-makers with messaging around “redundancy” and “security.” Rule #4: Insert a Call to Action Google now recommends experimenting with calls to action, such as “download” or “shop and compare.” Believe it or not, in the early days Google was inclined to reject many common calls to action—including “buy now”! Editorially, today Google only rejects a small percentage of calls to action. (As stated in their written policies, they prohibit universal calls to action such as “click here.”) They’ve been a bit quieter about the related notion of an “offer,” which is a staple of ad copywriting. But that’s your job, not Google’s. Invent compelling offers if they might generate a response, and see what happens. One client, for example, tried this ad that included a bland mention of their company name along with a product benefit: ClearTone Acne Treatment Finnish acne cream now available! Clearer skin in as little as 7 days foracne.com A second ad contained a benefit and a free offer: Get Clear Skin in 7 Days European Acne Treatment now in the US. Act now & get one month free! foracne.com The first ad pulled a CTR of 0.8%. The second ad pulled 1.7%. By generating a higher CTR, we were able to keep more keywords enabled, and we were able to lower bids and significantly lower average CPC while maintaining a similar ad position. This did improve ROI, though ROI 231 232 Winning Results with Google AdWords could have gone down in the process of attracting more clicks. This is why it’s so important that Google typically rewards advertisers with higher ad position when they do increase CTR. We are hearing more now about testing calls to action such as “buy now” in pay-per-click ads, but what is not often discussed is the wide range of calls to action that is possible. Such call-to-action suggestions should connect closely with the nature of your offer or with a larger marketing strategy that you’re pursuing. One of my clients, a popular technology journalist, established a relationship with his audience by having them subscribe to a free newsletter about the foibles of Microsoft Windows as they affect the average computer user. It was his firm belief that the word subscribe sounded too daunting (he thought users would worry “Would it cost money? Would it mean an arduous sign-up process?”). He was adamant that we try the word get instead. As it turned out, those who saw the ad containing the word subscribe were slightly more likely to become subscribers than those who saw the ad containing the word get. The word get, evidently, did not explain to prospects what they’d be doing as well as the word subscribe did. Moreover, the word subscribe did not result in unsustainably low CTRs. The concern over this word turned out to be much ado about nothing. Rule #5: Run a Test, Keep the Winning Ads Unless your marketing campaign consists of nothing more than a single AdWords ad, you’ll need some method of measuring the effectiveness of your ads. With one ad, your sales either increase or stay the same. If the sales go up, your ad is effective. If the sales stay the same or go down, your ad is ineffective. If you’re running a number of ads, and you should be, the only way to determine the effectiveness of your campaign is to do some testing. After running a test, you’ll delete or pause the losing ads and keep the winners. Is it really that easy? Sometimes. But wait! There’s always a fun loophole or two to consider. For example, I’ve gone back and looked at ads a few days after I’ve paused them as “losing” ads, and due to latent sales that have come in after the ads were paused (but are credited to those particular ads), they are now tied with, or beating, the “winners”! The only recourse I have in those quirky cases is to unpause those ads and keep running the test. Rule #6: Inject Some Flair Keep in mind that copywriting flair and writing catchphrases that cement brand awareness can be overrated in a realm where targeting and testing reigns. You might require more panache if your job is to write longer copy for the sales page on the website. Here, though, I am referring solely to your AdWords ads. For example, earlier I mentioned the example ad text “elegant martini misters for the discerning martini-meister.” In many contexts, this ad would be too silly, but my test identified this ad as the winner, likely due to the whimsical tone. There are many more examples, however, of attempts to inject flair, whimsy, or style into an AdWords ad that simply fall flat. First and foremost, search users are trying to find their way around, not get a chuckle. In an ad for a website-building product aimed at novices, I tried the ad title “No Web Geek? No Problem.” This didn’t work; the ad got low CTRs. However, the factual ad text “no programming required” did work. Flair-driven advertising is still more the exception than the rule in AdWords copywriting, but you need to incorporate it as part of your repertoire of techniques depending on the target audience. CHAPTER 8: Writing Winning Ads Some Ideas for Testing Ads The most basic form of testing Google ads is through a split-test or A/B test. The AdWords interface is set up so you can test the performance of ads “head to head.” For ideas about what kinds of variables you’ll want to consider testing, see the section “Ideas for What Variables to Test,” a little later in the chapter. You’ll typically (though not necessarily, if you use Google’s tracking products) be using special tracking codes to distinguish one ad from another. The subject of tracking is covered in considerable detail in Chapter 10. In the broader marketing world, split-testing is an advertising methodology that has been perfected over the years by direct marketers, often at considerable expense. Sending two offers randomly distributed over two segments of a mailing list is one way of testing response. The marketer varies a key variable, such as envelope color or the copy in the introductory paragraph, and tests response rates. It might take many years to learn which combination of elements is optimal. How Split-Testing Works Since 2002, Google AdWords has made split-testing easy and cost effective. In any given ad group, you simply use the Create New Ad feature (see Figure 8-2), and presto, you’re testing two or more ads against one another. The primary metric you’ll be looking at is CTR, but if you’re tracking by ad (using tracking URLs or Google’s conversion tracking products), you can also track conversion rates and ROI on each ad. The most detailed reports are available by running custom reports under the Reports tab in the AdWords interface, but conversions are also noted at the campaign, ad group, and ad variations levels. In some cases, at these latter three levels, you may not find all the data you need; you can add it to the display by clicking Customize Columns. When you’re running multiple ads in a single campaign, Google automatically rotates ads as evenly as possible. Therefore, if you run four ads simultaneously, each ad should be shown approximately 25% of the time. If your numbers do not show that to be the case, it might be the result of looking at a date range during which not all ads were running for the entire period. To make a reasonably accurate comparison, you’ll want to wait until all ads have been running long enough, and then choose a date range during which all ads were running. Unfortunately, you cannot adjust Google’s automatic rotation so that some ads appear with more (or less) frequency. There is an automatic “favor the ads with the highest CTR” (the Optimize setting) feature in the Edit Campaign Settings interface (see Figure 8-3), but for active testing I recommend this be switched to Rotate, or “show ads more evenly.” The results of split-testing can be educational. Even small changes in the ad copy can cause CTRs to vary significantly. But, as discussed above, the best practice to follow at first is to vary the ads significantly rather than testing minute differences. Test smaller differences at the refinement phase. Ideas for What Variables to Test The first thing to consider when planning a test is the number of variables to test. Initially, it may seem that the more variables you introduce, the better your testing. While it’s true that you could introduce an almost infinite number of modifications, keep in mind that the more variables 233 234 Winning Results with Google AdWords FIGURE 8-2 Click Create New Ad, and presto, you’ve initiated a split-test. you have, the more difficult it is to achieve conclusive results. Minor variables will eventually test differently over long periods of time. Therefore, the best method is to test with four or five possible objectives (brand, ROI, CTR, clarity, credibility) for your ad copy firmly in mind, in whatever proportion feels right. FIGURE 8-3 Select Rotate, or AdWords will decide for you which ad is the “best performer.” CHAPTER 8: Writing Winning Ads However you choose to approach your testing, you’ll probably find that your first couple of tests weed out the really poor performers from the better performers. At this stage you can either make drastic changes to the poorest ads and put them back in the mix, or just concentrate on finetuning the better ads in the bunch. However, there is a limit to the amount of fine-tuning you can do. Eventually, the law of diminishing returns will kick in, and it will be time to stop the tweaking and use the ads that have provided the most profitable results. Remember that other factors such as shifting tastes, randomness, timing, and so on, can affect the performance of your ads. Testing on Calls to Action and Offers One of the easiest things to test is the effectiveness of a simple call to action. In one ad, simply describe your product. Then create an identical ad, but include a call to action such as “Buy now!” “Download Free Trial,” “Try 30 days free!” or some other variation. Generally speaking, your call to action should connect with an action that’s easily available from the landing page that the ad takes the prospect to. When testing multiple calls to action, be sure that each one takes the user to a different landing page. This is the only way you can accurately determine the effectiveness of the different calls to action. Closely related to calls to action are offers, which typically combine a call to action with a time limit or discount. Even without a call to action, an offer implies a call to action (purchase before a certain date to receive a discount, for example) and typically creates a sense of urgency. This real-life example should give you an idea: Cool Birthday Gifts For anyone! An amazing selection of quality products. Delivered. Uncommonlygifted.com Cool Birthday Gifts For anyone. An amazing selection. 15% discount for a limited time! Uncommonlygifted.com These two ads were tested over a relatively high number of clicks—in excess of 2,000 for each ad. Searchers clicked on the second ad more than twice as often (1.9% to 0.9%), allowing Uncommonly Gifted to bid lower on popular gift-related terms (average CPC was reduced from 17 cents to 12 cents) while maintaining visibility with a high ad placement on the page. The benefit continued after the click, as this ad generated a higher conversion rate to sales. One large retailer I work with was swamped with Christmas orders by simply resorting to offering “free shipping until Dec. 15” with no minimum order or restrictions. Such is consumer psychology: such offers may be worth less to people than their reactions warrant, in terms of vastly increased conversion rates. And why is free shipping, which reduces the price to a consumer by, say, 10%, more attractive to that consumer than cutting the price by 10%? Perhaps because shipping feels like an unknown, and it turns into a known, thus reducing uncertainty. Also, people are poor at making quick calculations. Whatever the psychology, these are the types of offers you should be testing. 235 236 Winning Results with Google AdWords Testing on Syntax Variations Minor syntax variations, like plural instead of singular, can have a significant impact on the response your ads elicit. Since the ad space is so limited, you’ll be tempted to use abbreviations and symbols where possible. Remember that Google policy limits the use of symbols to their true meaning. Whatever you use, try testing both—the abbreviation and the full word or phrase, or the symbol and the word. For example, try using and in one version of an ad, and an ampersand (&) in another. See if it makes a difference. But doing this should generally only make a difference if the saved space allows you to write a significantly better ad overall. Some small differences in wordings actually have a deeper meaning, so I don’t always consider these trivial. “Ships same day” might actually ring more true than “same day shipping,” so the consumer sees the former as more credible. Or, it might be the other way around—the second variant sounds more natural, and faster. I’ll leave it to you to test for yourself. Differentiation of Ad Copy from Other Ads on the Page Many of the ads that appear for a given search will have very similar titles and copy (see Figure 8-4). If you find that to be the case, try setting your ad apart by using different copy or a totally different approach. The traditional approach to differentiation is to communicate your FIGURE 8-4 If all the ad titles look similar, try making yours different. CHAPTER 8: Writing Winning Ads company’s unique selling proposition or points of difference and focus on them; I’m referring to something a bit narrower here. As many advertisers will begin thinking alike about what creates an effective ad, you could try creative, unusual ad titles that might boost your CTR relative to the others. If seven ads say “pumpkin seeds” and yours says “premium pumpkin seeds” or “grow giant pumpkins,” you might stand out just a bit. Sell a Solution, Not a Product or Service, and Test It If you were selling knee braces, you could have “Knee Braces” in your ad title, of course. But why not test an ad with “Knee Problems?” as the title and “knee braces” in the body copy? In general, ads that remind the consumer of their fears, problems, or concerns—“scary” ads—can be effective if the product or service being offered in the ad resolves the prospect’s perceived problem or fear. This approach is particularly effective in certain industries such as computer network security, tax law, and insurance. Differentiation, Period In a related vein to both making an ad different and selling a solution, depending on your audience, you’ll need to be addressing them with differentiation rather than a bland introduction. It’s fair to say, for example, that someone seeking liposuction or facial surgery has done a decent amount of research already. Saying that you offer these services isn’t going to be much of a differentiator, obviously. Why should they choose you and not the next plastic surgeon on the page? It’s tough to explain in a short ad, of course, and in a world populated by tens of thousands of similarly qualified professionals, being told to “come up with a unique selling proposition” for your website and ad copy may feel like an insult to the intelligence. Sure, but if you want anyone to click and convert, it’ll help to have a hook, whether that be testimonials, famous clients, your upscale facility, your blue-chip medical training, a new piece of equipment or technique, a white paper battling myths, a free gift, financing, or something else. Some of your clients may be attracted by an enticing description of the end result, much like they are by exercise-equipment ads that tout “a flatter stomach in 30 days.” How about “a flatter stomach in zero days”? The long and short of it is, for increasingly savvy consumers, closely matching their query and saying you “have that” may not be enough in 2010 and beyond. We’re getting a long way from the old Sears Roebuck or Eaton’s catalogs circa 1910. Consumers have choices and have done research, and in this age of e-commerce, they know they can easily buy things. So for example, the headline “Is Your Family Safe?” might galvanize more buying interest than “Carbon Monoxide Sensor.” However, it might attract people looking for locks, alarm systems, and smoke detectors as well. The only way to know if it produces the desired results (higher ROI) is to test it. Flair vs. Flat As I’ve already pointed out, plain ads often work well in Google AdWords, but depending on your business, light humor and flair can help connect with the audience. For one client (KlinQ.com, a seller of designer housewares), I tested a fairly plain ad for teakettles against one for “whistling teakettles” that concluded with the hokey phrase “just whistle.” The latter won out, much as 237 238 Winning Results with Google AdWords the “discerning martini-meister” ad had performed well for another one of their products. This surprised me because such flair does not usually make a significant difference. The target audience in this case is obviously looking for the psychological lift that comes from purchasing “fun” kitchen items; they are seeking style, not merely basic need satisfaction or the best price. In the realm of style, identifying a product with a celebrity might also be worth testing. Drawing attention to someone who endorses a product (a baseball player who wears an Indian Motorcycle jacket, an actor who wears a certain type of jewelry in a movie, a celebrity who has learned to drive a go-kart at a training facility) can provide context that reminds searchers of hot trends and also hints that the celebrity’s glamour might rub off on them. If you use this technique, the content on your website needs to be factual in such cases to avoid potential liability issues. The third-party endorsement falls under the general rubric of credibility. Rather than seeing this as merely a dimension to test, I believe that advertisers should always be thinking about credibility of this nature. “Voted best free content by PC World” or “USA Today recommends” are excellent endorsements that would stand out from other, similar ads on the page. However, it must be clear that the endorsement is real, as proven by the content on your landing page, or Google Editorial may reject it. Brand Impact and Story-Telling Consumers will look at the display URL before they decide to click. They’ll also look for cues in the ad to decide if you’re full of hype. So does this mean that you’re facing an uphill battle if you’re not a big company with a recognizable URL? Not really. To be sure, in some industries consumers will outclick a “familiar” ad by a factor of 4 to 1 over an ad from a company they’ve never heard of. But you can overcome the familiarity factor by capitalizing on your unique position in the marketplace. Using familiar phrases, features, and benefits associated with your company, or emphasizing your long history, are things that should help you stand out from a crowd of wannabes. But if you don’t have these, you need to invent something. You need to begin telling your own story and writing your own history. That doesn’t mean misleading people! But your buyers often respond well to imagery and mythology. They want you to create this. “Diamonds are forever” isn’t just a cheap slogan. Marketing campaigns by de Beers repositioned diamonds completely in the marketplace. They went from being just another precious gemstone, to the standard for engagement rings, with a suggested “two months’ salary” dollar value attached. Seth Godin reminds us that in the wider world of marketing, you don’t get much time to tell a story. But if told consistently, encountering just a tiny sliver of the story later on can remind people of what you’re all about. Godin shows an image of two-and-a-half letters of the Starbucks logo, and he’s right: by viewing just a couple of letters and a single pointed shape from the logo, I immediately began to feel myself walking into the familiar Starbucks store layout. The other image he shows is a tiny drawing of the tip of an elephant’s trunk. Just by glancing at a couple of circles and lines, the mind begins to evoke the majesty and beauty of this hulking beast.12 So you might consider your AdWords ad the “tip of the elephant’s trunk.” It can work in your favor whether you’re a large company or a small one. But not if you haven’t made the effort to CHAPTER 8: Writing Winning Ads invent a mythology to help your customers convince themselves about the experience of buying from you. Some categories lend themselves particularly well to smaller players. A maker of fresh pastry shipped overnight—let’s call her “Little Old Pastry Chef”—may do better than a large company whose brand name has already been associated with lower-priced products selling in supermarkets. Sometimes consumers are captivated by “small.” Supermarket products aren’t remarkable, but custom-made items are. If the smaller retailer seems dedicated to a niche like fresh pastry, consumers will be open to buying from it. Small companies with focused ads can beat large companies who run ho-hum, me-too ads. In some cases, it’s the big guys, not the little guys, who look out of place. Unless you’re really sure the recipient would like it, for example, you’re probably not going to order a box of Dunkin’ Donuts as a birthday gift. The increasing consumer appetite for customization, and the growing adoption of local search, also gives smaller firms a potential advantage. Such trends at least don’t appear to put the niche player at a significant disadvantage, as long as they’re content to stay small. Larger retailers can also capitalize on such trends if they understand the psychology of their customers. A great example is Zingerman’s (www.zingermans.com), which began as a deli in 1982 and today is one of the most successful online food retailers in the world. The remarkable thing is how few of their competitors are currently taking advantage of the wide-open keyword inventory on keyphrases like online deli, cheeses, and specialty vinegar. Zingerman’s today advertises on some deli-related searches; they didn’t seem to be doing so in 2005 when I first wrote about them here. Their competitors would be wise to follow suit: the market and word-of-mouth potential surely makes it worth taking a shot with some low-cost search marketing. As people search more nowadays with local intent, a sense of physical place can be the biggest advantage going for the smaller retailer. A Canadian boutique called The Added Touch evolved into a mail-order (and now, online) sales leader. The company’s brand was built largely around the Oakville, Ontario location of the original physical store, and customers are well aware of this. If Martha Stewart found herself in the Toronto area, you can imagine her making a beeline for tony Oakville (wouldn’t that spice things up). Many online sellers today have a great story to tell, but they aren’t telling it. Why not test ad copy—and website copy—that refers heavily to your physical location and humble origins against copy that emphasizes only product features and benefits? Testing the Display URL Speaking of the display URL, recall that this is prominently displayed at the bottom of every ad. Users may look at this for some cue as to the identity of the company they’re dealing with. Some URLs, particularly short, readable ones, may inspire a more positive response than others. If your company name is two words, a typical thing to test is whether an all-lowercase URL gets a higher CTR than one with a capital letter at the beginning of each word—dolphinzone.com versus DolphinZone.com. You shouldn’t uncover major differences in user response, but it’s worth a look. Because Google may allow you to include a subdirectory name in your display URL, you may find that a keyword-rich subdirectory name helps. For example, if you worked for Apple, you might test www.apple.com versus www.apple.com/ipod-nano. For some time now, Google 239 240 Winning Results with Google AdWords hasn’t required that the display URL containing a keyword-rich subdirectory name take a user to a page that matches that exact URL, as long as the root URL is accurate. Incorporate this trick into your testing. Some companies set up dedicated URLs to capitalize on their brand name while focusing the user’s attention on a hot new product. For example, on a U.S.-based query for “Blackberry 8820,” Research in Motion is sending me to a site called Blackberry8800Series.com. Cool. Microsites like this often offer tighter testing opportunities for the marketing team, and the combination of the beloved brand with the specific product name makes for an interesting experiment. Granted, many users will opt to buy that device in an offline store or through their wireless carrier, so tying these revenues back to the AdWords campaign isn’t easy. Beware of the microsite strategy if it drops your trusted brand out of the picture. In general, a well-respected brand should strive to garner the benefit of that brand in the display URL. A financial conglomerate I worked with tried (generic examples given for confidentiality reasons) www.FinCon.com against a more generic site with the URL www.home-finance-savings.com. A second test pitted a longer but more targeted URL against the brand-name URL: www .FinConHomeFinance.com against the original www.FinCon.com. Run your own test if you like, but if I had to guess, I’d bet your generic microsite will see significantly worse performance than the one that incorporates your brand name into the URL—including the display URL. It’s Not What You Say, but How You Say It As we’ve all discovered at one time or another, your choice of words does matter. There’s a big difference between happy and ecstatic, even though both describe the same basic emotion. Therefore, as you test your ads, try using different words and phrases with similar meanings. If you’re looking for a catalog of suggestions for word variations to try, see Richard Bayan’s book More Words That Sell (McGraw-Hill, 2003). Tracking Results The best way to track the performance of your ads is to tag each ad with a unique tracking URL, usually using parameters dictated by your web analytics software, and a naming system that you find helpful. For example, you might refer to campaign 2, ad group 23, with the numerical code “0223,” and to denote each successive ad you test in that group, you might use a letter of the alphabet. So a typical tracking URL for the first ad might be http://www.qvack-qvack.com/shirts .asp?source=adwords&ad=0223a. The next ad in that same group would have the same URL, but end in the letter b, and so on. To make sure you also get keyword-level data (without excluding the ad-level performance data), you can enable autotagging in AdWords and use a special parameter that will dynamically append whatever keyword triggered the ad, right on the end of your own tracking string. To do so: 1. Go to Account Preferences. 2. Select Auto-Tagging. 3. Add the variable, kw={keyword}, to your ad destination URL like this: http://www.qvack-qvack.com/shirts.asp?source=adwords&ad=0223a&kw={keyword} CHAPTER 8: Writing Winning Ads Statistical Significance in Testing Advertisers often wonder if there ever comes a point at which you can be confident that one ad will outperform another over the long term. Some statistical experts will say that as few as seven occurrences of the desired outcome (clicks or sales, in this case) provide enough information to be confident. While such claims are based on raw math, they assume that no other influences are at work in the experiment. In the real world, I have found it necessary to run tests longer than some experts suggest. Unevenness in ad delivery across different sources, different user behavior at different times of day, changing ad positions, comparison behavior by users seeing different competitors’ ads at different times on different keywords, and numerous other factors make it difficult to trust the orthodox statistician’s approach. Without a better way to describe it (I’m no math PhD), allow me to suggest that patterns can be “wonky” over the short term, but become very reliable over a longer period of time. That does not, of course, explain why previously solid patterns begin to reverse themselves so that black becomes white and up becomes down. But often you can chalk that up to changing market conditions, shifting consumer demands, and strategic behavior by competitors. Since AdWords clicks are generally pretty inexpensive, I suggest that my clients run ad tests for a relatively long period to ensure the results are accurate. “Relatively long” plays out differently for different ad campaigns. Practical experience has shown that ads may even out in performance, even though probability theory would have given them a 95% chance of continuing their gap in performance. It is beyond the scope of this book to fully explain why this is so, except to say that conditions change rapidly in this environment, and there is more randomness and volatility in small samples than there might be if you were flipping coins. Clicks and buying interest might flow in from one region all at once, for example, skewing results towards one particular ad, but only temporarily. Probability theory doesn’t account for consumer enthusiasm, which might flood into this process and then vanish as quickly as it came. So, as long as a campaign is within the normal range of attracting a steady number of impressions and clicks, I make sure not to jump the gun on interpreting split-tests or multivariate tests. Instead of thinking in terms of numbers of clicks or sales, I often think in terms of a representative time period, such as two days or a week. My rule of thumb is to give tests time to play out fully. If I’m looking at CTRs primarily, I’ll look for 100 or even 1,000 or more clicks on each ad, if I can get away with holding the test that long. As shown in the example for UncommonlyGifted.com, we generated over 2,000 clicks for an ad containing a time-limited discount offer to compare it with an ad that did not contain the offer. Our findings form the foundation for a long-term campaign strategy. We are surely not going to plan such a strategy based on some statistician’s views (not directly suited to this complex medium) that seven clicks are enough to compare ad performance. After generating over 2,000 clicks on each ad, at 17 cents per click, we’re rock solid in our belief that the ad with the offer works better. The $680 spent on that test is an inexpensive piece of market research, especially considering that they are making sales while conducting the test. Looking at conversions and ROI is even more fraught with uncertainty, because the frequencies of sales can be low and big-dollar sales can be “spiky” and skew results. You can wing it and make 241 242 Winning Results with Google AdWords sure you don’t make any decisions based on fewer than 20–30 sales, or use any of the generic tools you can easily search online that will give you statistical confidence levels for picking long-term winners based on your inputs. These models incorporate the degree of separation between the two results (if you have ads with conversion rates of 2% and 8%, versus ads with conversion rates of 2% and 3.2%, the latter pair will require considerably more data to reach the same confidence level) as part of the calculation. After some practice, assuming you have decent sales volumes, you get better at picking out the winning ads after an appropriate period of testing. Moving from Ad Content to Campaign So far in this book, I’ve addressed the short history of paid search and how it fits in with other forms of marketing. I’ve shown you the basics of how the AdWords interface works and introduced you to some intermediate-level campaign management concepts that should help you succeed over the long haul. And you’ve now gotten a taste of how easy it is to tweak your ad copy to improve performance. It’s the world’s fastest and cheapest market research! Speaking of testing and market research, a lot of companies get a bit gun-shy when it comes to actually launching their AdWords campaigns, because of the difficulty in forecasting results. Not uncommonly, paid search will have an advocate within a company, but this advocate may lack the support of senior management. Whether your company employs two people or two thousand, it’s worth taking an in-depth look at how this kind of marketing campaign fits into your overall corporate strategy. Some corporate cultures still resist innovative marketing methods because they feel they’re too unpredictable. If you glossed over Chapter 6, you may want to revisit it now. There, I explored ways of convincing the boss (and yourself) that AdWords is a relatively low-risk, if unpredictable, marketing opportunity. I also explained how to shape your campaign from the beginning by identifying key campaign goals, carefully assessing your sales process from clickthrough to revenue generation, and asking where your business model fits in the context of other successful online business models. Let’s move onto the challenge of squeezing out more volume from a campaign that is already working well. Endnotes 1. Nassim Nicholas Taleb, The Black Swan: The Impact of the Highly Improbable (Random House, 2007—no the publisher info is not an April Fool’s joke). Taleb’s work illustrates so well for the layperson how apparently simple math problems reach incredible complexity with the addition of a greater number of highly uncertain variables. He gives the example of predicting the location of a billiard ball as it bounces around a real-world (not theoretical, two-dimensional) pool table. AdWords is at least as complex as that. Educated readers may wish to arm themselves with Taleb’s wisdom in this and his previous volume, Fooled by Randomness: The Hidden Role of Chance in Life and in the Markets (Random House, 2005). Among other things he offers a critique of over-interpretation of data and an exaggerated sense of causality that is wired into our brains—or what he refers to as the CHAPTER 8: Writing Winning Ads Baconian “narrative fallacy.” Empirical skepticism is a mentally tiring habit, according to Taleb, but it seems like one we’ll need to adopt if we’re to avoid making major errors in judgment in important situations. Perhaps more to the point, in a climate of big potential upside, you should be open to possibility and bullish about the future, in spite of the many layers of uncertainty that face the would-be “planner.” 2. You need to be particularly aware of the “spiky” nature of episodes of “bad” clicks, which can cause you to have the mother of all bad days. Whether it’s daily budgeting, an alert-based bid management system, or a high degree of rapport with your AdWords rep in cases of click fraud, avoiding or reversing catastrophic events can be (as in many situations in Extremistan) financially more important to you than frequently squeezing out a tiny bit of extra performance from a “tweak,” especially if you overinterpret and assign too much causality to such “tweaks.” A lesser version of the catastrophic day of nonconverting clicks is the Post-Christmas Problem: many retail advertisers fail to lower bids or pause campaigns in the four weeks following December 25, during which, quite predictably, in many sectors, conversion rates drop to their lowest points of the year. This is exacerbated by the fact that many advertisers have (often belatedly) raised bids to capture sales volume during a period of unusually strong conversion rates. Giving back all your holiday profits based on predictable events isn’t random or improbable, though: it’s just stupid. 3. Matt Ragas explores this phenomenon nicely in his book The Power of Cult Branding: How 9 Magnetic Brands Turned Customers into Loyal Followers (and Yours Can, Too!) (Crown Business, 2002). 4. For a recent critique that makes this point, see Bob Garfield’s And Now a Few Words from Me: Advertising’s Leading Critic Lays Down the Law, Once and For All (McGraw-Hill, 2004). 5. Jay Conrad Levinson, Guerrilla Marketing: Secrets for Making Big Profits from Your Small Business (Houghton Mifflin, 1998). 6. OK, maybe not that. To quote Seth Godin, the marketing guru who is so influential he has his own action figure (as well as being cited too often in this book), “I solved my headache problem twenty years ago.” 7. When I clicked their ad, HengDongChem.com’s site was down. More recently, they seem to have disappeared from the page. Another weak player weeded out by the AdWords environment. 8. Date of query: December 30, 2007. 243 244 Winning Results with Google AdWords 9. After the dust had settled on June 3, 2008, Obama became the Democratic nominee for President, although it took Hillary Clinton some time to concede. Shortly thereafter, final numbers on each candidate’s Google ad spend came in, showing Obama’s $2.08 million spend in 2008 far outstripping Clinton’s. The strategic nature of the Obama campaign’s ad buy is further emphasized by the fact that $1.7 million of it was deployed in February 2008 alone. See Kate Kaye, “Clinton Spent Far Less Online Than Obama,” ClickZ, June 16, 2008; Kate Kaye, “Obama Spent Most of $3 Million on Google,” ClickZ, May 29, 2008. 10. Appropriate thanks are due here to Mona Elesseily, account director on this client account, with whom I collaborated on the copywriting and testing efforts. 11. On the ins and outs of different types of multivariate testing, the counsel and writings of Scott Miller of Vertster and Tom Leung of Google have been very helpful. 12. Seth Godin, All Marketers Are Liars (Portfolio, 2005), 70. For the counterpoint, which criticizes large companies and their advertising agencies for being propagandists, see Laura Penny, Your Call Is Important to Us: The Truth About Bullshit (Crown, 2005). Penny is an enemy of phoniness, especially phoniness in the pursuit of profit, but never makes it clear what authenticity would look like. Chapter 9 Expanding Your Ad Distribution: Opportunities and Pitfalls P robably the most common challenge I face in helping out “mature” AdWords accounts— either those that my company has been overseeing for some time, or new clients who come to us having reached an impasse in their efforts—is insufficient click volume. The problem is easy to describe: the advertiser loves the results so far and just wants more of them! “Great. We’re at $28 a lead, comparing favorably with the $74 per lead generated by offline advertising. Now we need more leads!” might be a typical directive. This can lead to some interesting judgment calls. Do we increase the average cost per lead of the whole campaign in order to generate additional volume (by, say, increasing bids to improve ad position)? Or do we hold firm on the cost per lead and search harder for ways to increase targeted clickthroughs at low cost? (The latter, obviously, is the bigger challenge. It’s pretty easy just to go in and increase all your bids.) Conventional wisdom suggests that when you’ve found the low-hanging fruit of inexpensive customer acquisition methods, you’re forced to pay more for incremental customers. As discussed previously, there are tactical and philosophical considerations that determine whether a company wishes to pursue market share or profitability at a given juncture. Marketers typically use terms like “aggressive” as shorthand for pursuing more leads or customers by raising average CPC and ad position. Those who can’t afford to raise their cost per new customer too high, and want to squeeze the maximum ROI out of every click (or not incur that click at all), might refer to themselves as pure direct marketers, conservative or cautious, slow growth oriented, or ROI focused. Even if you leave bids where they are, you can push your ads out in front of more prospects by using AdWords. This chapter offers some suggestions for the most likely avenues to accomplish that, so you can implement an expansion plan with a minimum of fuss. This chapter also covers a couple of advanced topics in ad distribution: local targeting, new types of content targeting, and new developments with content targeting (such as new reporting capabilities that might give you more confidence in bidding on content).