Nội dung

300 Winning Results with Google AdWords ■ There was a form that you needed to fill out that didn’t include “District of Columbia” as one of the available states, or Canadian provinces, or in some other way made it impossible for you to “exist.” ■ The credit card security system didn’t like the fact that you were traveling, so your purchase was declined, but there was no alternative means of purchasing. These and countless more issues are the worst kinds of problems because they deter even eager prospects from completing a transaction or forming a relationship with you. If there are serious barriers to people doing business with you, they usually won’t. As Amazon.com chief Jeff Bezos frequently stated on his way to becoming a billionaire: “We’re trying to make it easy for people to buy.” It helps to know what business you’re in, a degree of self-awareness Bezos always brought to his task. Gas stations, I’ve found, generally don’t make it hard for you to pump the stuff into your tank. Extending the analogy to your business, since you’re not Amazon.com, you can’t create “1-click ordering.” You probably don’t have the user’s credit card already on file. You may be constrained in how much free shipping you can offer. You might not be able to create a site that is quite as smooth as Amazon’s. But by making people really want to buy from you (by persuading them mainly through the relevance of the offer to what they searched for, and through the smoothness of the path, with just the right touches of info-candy and brand image to seal the deal), you reduce the need to be 100% perfect in your site architecture, shopping cart, or other elements of the sales process. Given the sorry state of so many websites today, pure plumbing will lead to most of the significant increases in ROI. People will be far less likely to buy from you if your site is hard to use or literally broken. There are probably already people who want to do business with you. If you do nothing else, at least don’t put roadblocks in their way. That means taking users to appropriate landing pages (instead of the wrong ones). It means ensuring that your web hosting is adequate, that your shopping cart works, that your pages load into all major browsers, and so on. Unclog. Renovate. Declutter. Then, improve your overall level of communication where appropriate, if you have additional budget. You don’t have to do it in the middle of the buying process, but somewhere underpinning your marketing (maybe in a lot of places), it helps if you have a “story” to help your customers make sense of why they are buying. Persuade, Convince, Use Psychology (Persuasion and Storytelling) Getting rid of barriers to commerce may not be enough. In a sea of conflicting commercial messages, the one that inspires may be the one that gets the sale. To use a dating analogy, sure, removing major barriers is the first step, since getting a date is nearly impossible if, say, you never get out of the house to meet anyone. Removing the obvious impediment of hermit-hood (with online dating, even that barrier is reduced) might be a first and necessary step to getting a date, but it doesn’t change the fact that at some point, somebody has to like or be inspired by you. You have to convince! You need to make an attractive offer, even if that’s only making sure you have fresh breath when you say, “I know a great coffee shop near here.” CHAPTER 11: Increasing Online Conversion Rates The same goes for your business. The fact that you sell jewelry, and that your shopping cart isn’t broken, is definitely not going to be enough to convince a high percentage of prospects to buy jewelry from you. There are a lot of jewelers. Why should prospective buyers buy a particular product? Why should they buy from you? If your landing page or site as a whole doesn’t provide the answer to that question, then only a small percentage of prospects—those in an enormous hurry, for example—are going to take the plunge and buy. Let’s put it in terms that your ego will hate. But they’re important terms, because it’s what I and every other prospective customer is probably thinking. “So you’re another jewelry store online. I don’t give a @#$@!” There are two primary elements to persuasion online: copywriting and design. Writing good copy is the most obvious of these. Beyond that, web credibility and brand cues are indirect persuaders. Copywriting Great sales copy doesn’t grow on trees. Like anyone else in this business, I’ve tried to mix and match a variety of areas of expertise, grabbing insights wherever possible. If you don’t have the budget to hire an experienced sales copywriter for your site, you’re going to have to develop a little bit of expertise yourself. The most basic requirement (don’t laugh) is that you have copy. I’ve seen far too many sites with basic three-word product names and pictures of products and little else. Some amateur sellers appear to believe the Web is just an order-taking system, a big catalog that will attract plenty of eager buyers no matter what. I’ve come across some mind-bogglers: for example, a successful offline sports apparel business from near Kalamazoo, Michigan, that set up shop online under an entirely different name. They chose a domain name that evoked nothing more than that they were some kind of generic online seller of sports apparel. The site, too, was generic. They had terse product descriptions and little else. No mention was made of their successful bricks-and-mortar presence. What if they had chosen a catchy name like Kalamazoo Sportswear and populated the site with not only full-fledged product descriptions, but an engaging story about the business, including the positive local PR they’d received in newspaper articles? So don’t be shocked when I tell you that the worst kind of copywriting is no copywriting. There are tens of thousands of online businesses out there with virtually no copy on their sites. As a result, they have virtually no online presence. Is a lack of copy also bad for organic search referrals? Don’t even get me started. Believe it or not, some of the advice that is useful for writing small AdWords ads also comes in handy for pages of sales copy that might go on for many paragraphs. It seems to be something of a universal law that in spite of wide variation in industry-specific terminologies, most readers—even prospects of a complicated niche business—get turned off by jargon. Sure, you do have to pay some attention to your prospects’ reading level and degree of expertise to avoid talking down to them. But even for the niche reader, wading through jargon-laden presentations can be tedious. Moreover, copy that is too dry can actually suck the enthusiasm out of a prospect. Not every line of business can be “fun,” but your potential customer shouldn’t approach her relationship with you as if it will be pure drudgery, either. 301 302 Winning Results with Google AdWords Don’t hesitate to tell a bit of a story, provided the story quickly turns to focus on the benefits of your product or service to the customer and, above all, to the offer you’re making and the action you hope the prospect will take. Be clear and direct in your language. Inject emotional appeal and even sex appeal into your copy wherever that’s appropriate. For a software product, you’ll want to talk about ROI (money is emotional) and problem solving (alleviating headaches is very emotional). For a motorcycle jacket, referring to a celebrity that once bought one from you would add sex appeal. Certain adjectives like racy, heavy-duty, or vintage would also add sex appeal, for those who wanted to infer it, or at the very least a sense of status or authenticity. Let me give you an example from my own portfolio. An enterprise software company was experiencing poor conversion rates on their AdWords campaign, even though they were an industry leader in their human-resources-related field. What was needed was a rescue operation on the landing page copy. The rescue required two steps. First we eliminated the landing page list of cold, unemotional bulleted points. Next we tackled the industry jargon. In the end we turned a sterile, confusing landing page into an appealing and informative tool to motivate visitors to take action. Here are a few brief “clips,” if you will, of before-and-after copywriting from that landing page: Before: “performing regular talent inventory gap analysis of your human capital assets” After: “identifying talent gaps in your current workforce” Before: “unparalleled level of domain expertise” After: We eliminated this, along with a variety of other empty boasts, replacing them with concrete information. Before: “largest group of customer references in the industry” After: Here, we asked either that they provide a list of testimonials or delete this boast. At first, the only testimonial that appeared on the site was jargon-laden and lukewarm, which was inconsistent with this claim of customer satisfaction. Before: “...facilitates the end-to-end process of identifying...” After: “facilitates the process of identifying...” (eliminated redundant buzzword) Over time, the longer, more detailed sales presentation is likely to hold more interest than bulleted points would for serious buyers. Moreover, the clearer version should convert better than the initial pass with the jargon-laden long copy. Writing product descriptions that appeal to a target audience in retail is often driven by demographic research or persona research. It might be difficult to prove that one adjective beats another in writing descriptions for Chanel purses, but it’s probably safe to say that experienced fashion writers would do a better job at injecting flair into copy for such products than the average person off the street. In a small business, the owner or owners must absolutely become directly involved in communicating with customers and writing sales copy. If you sell designer purses or complicated home renovations, is it realistic to expect your 21-year-old webmaster CHAPTER 11: Increasing Online Conversion Rates (for example) to feel the necessary intimate connection with the audience? Yet I’ve seen businesspeople delegating the task of writing website copy to just such an uninformed person. It depends on your budget, but at some point, customer profiles need to be researched, and someone’s going to have to put words on some pages. Target, but Don’t Stereotype Overprofiling is a pet peeve of mine, and I tend to rant a bit against persona research when it is misused.9 Marketing is about real customers, not stereotypes; that’s why keyword-based search marketing is so powerful. We can remain a little more neutral in our assumptions and tone while leaning heavily on the keyword search itself to segment users. Think of all the money that’s wasted in media that can’t lean on search keywords. The Holy Grail of the young male is pursued to foolish extents by old-school ad execs and their clients, pitching the product to the “mode” (statistically the single highest-buying age and sex) rather than figuring out how to reach disparate customers across the entire distribution graph. Ever seen a 50-year-old woman driving a “guy” car like an Infiniti G37? Ever seen a young male financial adviser scooting around in a “chick” car like a BMW 1-series, a well-used Miata, or a “vintage” Fiero? Sure you have! In your online marketing, do a gut check to ensure that while targeting appropriate audiences, you’re not using imagery and wording that alienates prospects who fall outside the “mode.” Sure, marketing to “everyone” is a classic rookie mistake. Then again, you can’t possibly be marketing to everyone if you sent people to an appropriate landing page from their initial search for BMW 128i reviews or tax consultant Arizona. By definition, much of what you do with search marketing is already narrow. Is there a need to splinter that market by making silly additional assumptions? (Perhaps that’s why less presumptuous page images, such as the androgynous couple appearing on the Skype offer page discussed later in the chapter, can outperform stereotypical images in landing page tests.) Design Cues: It’s about Communication, Not “Hidden Persuasion” In large part, persuasive design comes back to the improved focus, reduced clutter, standardsbased design, brand cues, and other elements I address in this chapter. As entertaining as I find hypnotism as a spectacle, subtle responses to design aren’t necessarily “hypnotic” or “creepy.” Human cognition and emotion are part of direct response—always have been. Testing can turn up a lot of interesting responses, but your tests will likely stop somewhere short of magic. Many conversion enthusiasts like to experiment to discover emotional responses to certain layouts, colors, shapes, images, and much more. The complex and allegedly subliminal psychology of design has long been studied by a few experts. Especially in an offline environment, for larger companies with a lot of capital investment at stake, like mall owners and store designers, such studies are indispensable.10 Be wary of overestimating the hidden benefits of details such as punctuation, font color, button shape, and imagery. Some of these matters, indeed, could be summed up in a key credo offered by researchers on web credibility: get a site that “looks professionally designed.” Unless you have very high sales volumes, you won’t be able to test “red versus blue,” “triangle vs. oval,” for every 303 304 Winning Results with Google AdWords few pixels on the screen of your landing page. You’ll often be working with a professional who can offer you a holistic page concept, and your test will have to be among two or three versions, each of which sort of hangs together with its own internal logic. And speaking of coherent design logic, sometimes a site wins because it looks “folksy”—or not quite professionally designed. That’s something you have to test. One client I recall had a terrible-looking site. He said it was on purpose, because a “tactile” site was soothing to his customers in that it seemed to belong to a “real person and not a big company.” Then again, he refused to prove it through testing. Although imperfect, it’s often pretty effective to test two completely different versions of an important landing page, each with a distinct design “logic.” One of my clients, FourOxen Corp., tested their main landing page by completely overhauling it, stripping out clutter, changing many visual elements, adding a person’s face, and more. This was A/B tested against the old page; no complex multivariate testing was tried. The new page converted significantly better (most of the time), but FourOxen didn’t come up with that page by studying every variable over a three-year period. The design team put together a new page that would best be described as “completely different” from the old page. Many elements of the new design probably counted as basic professional competence in the field of landing page design; FourOxen was just staying contemporary and appropriate to their industry vertical. Professional competence and emerging standards that are shared among professionals can frequently offer useful shortcuts that allow us to achieve the results we need without starting from square one in the lab. That said, FourOxen has enough volume that they should now test versions of the winning page with more involved multivariate testing, in order to refine and improve conversion rates even more. To sum up, your site designs and landing page tests will be built around an appropriately holistic combination of plumbing and persuasion. No need to take hypnosis courses or to hire the “world’s best copywriter, Dr. Evil.” Unless you’re in some niche direct-response area, you can’t win with “hypnotic writing” alone. The cartoonish image of advertising and marketing as somehow being able to force or hypnotize intelligent consumers into doing things they wouldn’t normally do has persisted since the original advertising critiques came down the pike in the 1950s. But remember what the real leaders were saying in those days. David Ogilvy was telling you to “test the headline,” and make it sell! Test the headline. How tricky is that? So I side with Bob Garfield, a critic of many modern ad campaigns. Garfield insists that many campaigns are so poorly executed that advertising is often not persuasive at all.11 If you can’t get your overt message out there, what value could there possibly be in contemplating subliminal techniques? Testing Protocols: Best Practice; A/B/C; Multivariate Most companies design and redesign their sites and important pages based on a wide range of implicit assumptions. Most do not pay much heed to the art and science of response testing. Extensive user testing experiments (such as focus groups and other laboratory studies) are outside the scope of this chapter, but are recommended for those considering pursuing more CHAPTER 11: Increasing Online Conversion Rates advanced paths to insight about user behavior. Here, I’ll summarize some prominent approaches to testing that are being used successfully by many response-oriented online companies today. To begin with, we know that any data collection process requires you to have an eye for statistical significance and validity issues. Most of us in this field are not professional statisticians, and the accuracy of our efforts may not be 100%. But we can do much better if we just stop making silly, unfounded assumptions, and go out seeking really obvious, provable differences in response to different versions of our pages. “Testing” Method #1: Be a Lot Better from the Start Web professionals of various stripes, and interactive shops that specialize in user-centered design, should get you part of the way along the path towards constructing higher-response landing pages just based on their past experience, conscientious approach to keeping current with user experience trends, and data about response they may have collected in their firms. Any professional approach to site design and landing page design needs to integrate top-level architectural and brand feel concerns with nitty-gritty layout and copywriting issues. Iteration from “ground zero” will take a very long time if you don’t start with something reasonably compelling in the first place. Unfortunately, many design shops and so-called marketing agencies still trade in trends and fads, or are bent on selling you on a particular gizmo or two based on a strong conviction they have about some element of user engagement. A minority of conversion-focused agencies take revenues and testing protocols more seriously. You can either hire them or learn from them (us) to attempt to incorporate smart principles into your page tests. A good place to start is to ask yourself what kind of offer page, and what kind of targeting, you are dealing with. Should it be: ■ ■ ■ ■ ■ A lead-generation page? Will it offer an incentive or white paper? A standalone product page? Should it offer related products? A product category page? A compelling “long copy” information page? How long is long? An introductory page, such as a home page, that neatly segments prospects into the correct category? Depending on which type of page it is, and what other supporting elements are already built into the site, you’ll want to begin with a compelling layout. A web producer or web product manager certainly has enough expertise to provide direction, but a qualified design pro might do a better job of creating the layout for the offer page. There’s nothing wrong with looking for strong examples around the Web as a starting point, as long as the tone and objective-setting are appropriate to your business. Let’s say you settle on the fact that you’re designing a page around a single product but that it is important to increase conversions to the most expensive version of the product. Consider how you will incorporate: ■ A relevant headline (and assume that relevance may trump “salesiness”). ■ A product description (brief but not too brief). 305 306 Winning Results with Google AdWords ■ ■ ■ ■ ■ A brief (but not too brief) statement that differentiates your company. Benefit statements (more, or fewer, and where on the page). Testimonials, if appropriate. Pricing strategy—plan the best psychology, or whether trial offers are preferred. Image or images. Consider whether you need high-quality, high-impact, human, or product-based images. Have a designer consider the “flow.” ■ White space. Is the page too busy? ■ Navigation. Is the page “orphaned”? It shouldn’t be. But nor should the navigational elements be excessive. ■ Call to action, and how it is worded. ■ The shape and look of your action buttons or links. This isn’t a test yet. It’s just one page. Most companies won’t be able to test very well at all, because they’re not even working smart enough to plan that single, first page.12 A/B, or A/B/C, Testing If you want to make some major advances or test key differences in page layouts, but don’t have enough sales or lead volume to reach statistical significance in a hurry, you should still test something: two or three different versions of a key landing page, for example. Only five years ago, A/B testing of landing pages online was still new enough that it blew people’s minds when a test worked. A few entrepreneurial-minded web professionals managed to lead such processes in their organizations rather than sitting back and leaving it to a few experts at larger companies to reap all the benefit. One such professional was Lee Mills. Mills, a marketing consultant who has alternated between independent consulting through his firm Beyond Clicks and in-house marketing roles, has conducted a number of landing page tests for clients seeking improved conversion rates. One such test, for Anonymizer.com, showed surprising results. The first landing page (see Figure 11-7) had fairly brief sales copy, a clear offer, and was attractively laid out. Mills and the client didn’t believe that the conversion rate of 3.2% could be improved upon very much. Indeed, this does seem to be a nice page, and 3.2% was a fine result. Nonetheless, a much longer page was also tried (see Figure 11-8). It included more sales copy, more education about the dangers of spyware and threats to Internet privacy, and more information about the benefits of the product. It even included screen shots. This page did far better than the first attempt—it converted at a rate of 9.6%! In his presentation at a conference in August 2004, Mills said he and his team were surprised by the result because they’d always assumed it was important to minimize scrolling—to keep all the vital information “above the fold.” The result doesn’t surprise me. We often hear nonsense about the fact that people don’t like to read a lot of information—“Keep it simple, stupid”—that sort of thing. Obviously, with a result more than three times better than the short page, this longer CHAPTER 11: FIGURE 11-7 Increasing Online Conversion Rates Good landing page: 3.2% conversion rate page had something going for it. Having extensive sales copy does not necessarily conflict with the need to maintain a singular focus on converting the prospect into a buyer. You can do this, too. What will you be testing for? First, of course, you need to decide on which outcome you’ll count as a conversion: an order, a lead, or even a soft goal such as reaching the beginning of a signup process. You’ll then need to decide how pages will be rotated and identified so that you have a method for measuring which page led to which conversion rates. A handy, but slightly imperfect, way of doing this in the past would have been to use Google AdWords itself. Set up an ad group with two identical ads (which, as you know, should rotate evenly if you have ads set to “rotate”), but send traffic to two different landing pages. If you had AdWords Conversion Tracker installed, you might even be able to read the results right in AdWords. According to some analysts, such as Scott Miller of Vertster, a vendor of multivariate landing page testing solutions, this methodology can lead to skewed results. Long story short, returning visitors are not always shown the same page recipe they were shown on a first visit. They may be seeing two or more versions of the page. It’s a complicated argument and Miller doesn’t prove his point with hard data, but the upshot is, this is a rough and imperfect method to split traffic. 307 308 Winning Results with Google AdWords FIGURE 11-8 Great landing page: 9.6% conversion rate Today, with the available third-party tools, it seems awfully tempting to use tools tailormade for testing and reporting on the outcomes of tests. Yes, you can relatively easily customdesign a split-testing protocol in-house with the right programming and/or the right attention to your analytics stats. But the available solutions make it easier. I’ll discuss these more in the “Multivariate Testing” section. To run an A/B test, think in terms of two or three major theories you’d like to test, and test them all at once. This is far from a statistically perfect way to do it, but remember, you’re trying to get better, not be perfect. It’s an absolute myth to believe that you can make meaningful progress by isolating two page elements and testing those, then two more, and testing those, over time. Variable interactions mean that as you pick winners in some areas, you have changed the playing field for the next test. And running all of these small, sequential tests may take forever, because the impact can be so minimal on some test elements as to be trivial. It’s better, in most cases, to think in terms of big drivers, and layout approaches—almost like a composite sketch of two or three different “page types.” A perfect example is put forward by Avinash Kaushik in his admonitions to marketers to “just start testing.”13 Skype wanted to test an offer page. Two major kinds of pages were tested. CHAPTER 11: Increasing Online Conversion Rates One, a stylish-looking page with a slightly cheeseball image concept: a hopeful-looking male chatting in proximity to an attractive female. The second page tested showed a female with a slightly obscured friend who might be female, talking in a café-like setting. (Subsequently, I’ve noticed Skype testing a similar couple hanging out on a boardwalk. Sometimes, one of the partners wears rollerskates. I’d have to guess that they’re trying to give us cues of “fun” and “freedom,” smart thinking that runs counter to the first instincts of a typical software or telecommunications equipment company.) The key was to determine if the typical cheeseball telecommunications-company sales pitch page would perform better or worse than the understated, but still image-rich, San Francisco-café-chic page. To offer some added perspective, a new page idea entirely, with much more white space, a bold blue-and-white “paint splash” design, and less imagery, was also tested. Reaching statistical significance on a high volume of sales, the verdict came in: the pleasing white-space page (the upcoming Figure 11-10 shows an example) didn’t convert as well as the image-rich pages. And by a significant margin, androgynous freedom-loving friends (see Figure 11-9) beat the earlier-generation cheeseball telco guy-meets-girl trapped in a less evocative white-space layout. The bottom line improvement for Skype, in the form of tangible sales revenue increases, would soon run into six figures. A simple, elegant, and yet reasonably scientific example of an A/B/C test in action. Take note: the specific outcome in the Skype example is not what is important. The process is something that any qualified designer and marketer, working in tandem, can try. Nothing ventured, nothing gained. In Figures 11-9 and 11-10, you can see that Skype is clearly continuing the testing process they began some time ago. Even after ruling out hackneyed telco imagery, they carry on with new tests. Here, they appear to be segmenting tests by nation and language, FIGURE 11-9 This image-rich offer page moved the needle for Skype. This advertiser looks to be testing small refinements at this stage, such as heart balloons and a “no adware” benefit statement. 309 310 Winning Results with Google AdWords FIGURE 11-10 Anecdotally, the “white space plus screen shots and icons” approach wasn’t as successful. But it looks like the advertiser is still keeping it in their testing mix for the time being. and may well be testing smaller variations in an advanced multivariate test, as they should. For example, does the image sell more or less with the green rainbow or heart-shaped balloons? Which version of the buy or try button works best? Does a mention of “no spyware or adware” help or hurt? On Statistical Validity If you run a major e-commerce site or are sending high volumes of traffic to a page, you can test and retest frequently. However, I believe that it’s possible to oversell the notion of rapid experimentation. Many companies generate too little traffic—especially if they have multiple low-volume landing pages—to test in the ways that some experts advise. Several other factors make split-testing more complicated than some would let on. To be sure, understanding basic principles of statistical reliability is helpful, and simple tools like the Vertster Clickthrough Rate Validity Checker (see www.vertster.com/adwords-tool/default.asp) can help you get a feel for this (see Figure 11-11). Using the tool, I told it I had received 28 clicks from ad A and a CTR of 3.0% (which means that I must have had, according to the tool, 933 impressions). For ad B, I told it I got 39 clicks for a CTR of 4.3%, which means that I must have had 906 impressions of that ad. One of the things these tools are good at demonstrating is how you reach a high level of reliability in a split-test sooner when there is a wider gap in CTR performance. In this case, the gap is fairly wide—4.3% to 3.0%. Vertster’s statistical analysis tells me that 80% of the time, the current winner will continue to be the winner in the future. CHAPTER 11: Increasing Online Conversion Rates FIGURE 11-11 Vertster’s Clickthrough Rate Validity Checker (GWO is more powerful) A better tool, though, is not ad-focused but landing-page focused. Google’s Website Optimizer (GWO) tool (discussed in more detail later in the chapter) has a more sophisticated wizard that will offer projections of how long it will take for your A/B landing page tests—or even complex multivariate landing page tests—to reach a high statistical confidence level. Just punch in your particulars, and GWO will offer you some projections. It will update those projections, and statistical confidence measures, on the fly as you run your tests, too. Unfortunately, the reality here is complicated. What if your CTRs or conversion rates don’t diverge as much as this example? Given the wide variety of user motivations and mind-sets as they arrive on your site, how can you know that a conversion rate of 0.70% is really significantly better than 0.63%? It might take you quite a while to find out. This suggests to me that you are often better off gaining a deeper understanding of your marketplace and of web persuasion theories that will ultimately allow you to create new landing pages with vastly improved performance. If you go from 3% to 9%, you won’t have a tough decision as to which page performs better. Unless you have very high volumes, beware the myth that testing landing pages is about making dozens of minute tweaks. It’s about rethinking your communications strategy so that you’re making big leaps forward in performance. Those leaps are the ones that make you feel confident about making permanent changes. 311 312 Winning Results with Google AdWords Beware of Taking “Soft” Events as Gospel in Testing If you happen to be using a lead or other nonrevenue event as your conversion event for the purposes of testing landing pages, you may find that certainty eludes you. When long sales cycles are in play, you’ll need to go through more complex processes of ensuring high lead quality, and assessing whether the best page for leads is also the best for total revenue generation down the road, all things being equal. Again, without a massive data mining operation, you’re unlikely to get right answers to the toughest questions here. Just be aware that you may be fumbling around in a fog if your revenue events are delayed. Beware of leads that are too easy to generate. Many businesses, for all their efforts to improve their measurement protocols, will also need to sift through leads and sales to get a qualitative sense of issues like lead quality and overall business improvement. Not every business is set up to give instant answers as to which page or site design leads to the best long-term response. Response testing is just that: it’s most suitable for direct-response businesses. Let’s move on to discuss more advanced testing of a larger number of page “permutations”: multivariate testing. Multivariate Testing In Chapter 8, in discussing ad testing, I provided an overview of how multivariate testing works. The principles are similar for landing pages as they are for ads. Let’s say you identify four important page elements that you think might impact user behavior and sales conversions: the left navigation area, the headline, the body copy, and the imagery (large photo of a person, say). For the navigation on the left, you feel that additional clutter may be distracting, so you propose two versions: the current one, plus a cleaner version with the same links but not the additional promotional boxes. You want to try two new versions of the headline, and test those against the current headline. Body-copy-wise, you might want to test a relatively simple theory—long vs. short copy, for example—or something substantive, like adding or subtracting a second (redundant) call to action statement, or adding or subtracting a free bonus offer. Finally, you have two versions of the large image that you’d like to test against the current image, and no image. The total number of page permutations, varying these four elements, comes to: 2 × 3 × 2 × 4 = 48 Sound silly to test 48 versions of a page? Well, it isn’t feasible in many cases because of low volume, so you may need to employ simpler 8-, 16-, 24-, or 32-permutation tests. But it definitely isn’t silly. Out of 48 page permutations, it’s not uncommon that one or two versions combine the elements in a way that gets things just right from the user’s perspective. The whole, here, really can turn out to be more than the sum of the parts. With the correct kind of reporting, you get information not only about which is the winning “recipe,” but about which elements contributed the most to better or worse performance. Strategically and mathematically, experts advise that you should continue running the entire test until it reaches statistical significance, even if one element (say, the headline) can be declared a clear winner earlier than the rest. The second-best headline, for example, may wind up being CHAPTER 11: Increasing Online Conversion Rates in the winning overall recipe, when combined with certain other variables on the page, due to a phenomenon known as “variable interactions.” This is like saying: this is only the secondbest sermon, but when delivered in a stone church with a flower garden, by a bald pastor, on a sunny day in April, the second-best sermon is part of an overall winning recipe (for “best overall experience” leading to, say, higher donations) that beats out several recipes that include the “best” sermon, such as the best sermon being delivered by a blond pastor on a sunny day in a wood-and-glass church with a flower garden in August. It’s all about how things hang together. Recently, Google developed a fantastic tool for multivariate testing called Google Website Optimizer. It requires Google Analytics to be installed, and for your techie to install additional code on the landing page. In addition to a certain amount of technical competence, GWO relies on you to have sound thinking and strategy, solid creative inputs (images, copy, and so on), and a structure for planning and evaluating results. Because GWO does such a great job of managing, monitoring, and reporting on the tests, I’ll focus on Google’s product and provide a case study to really hammer home the potential benefit. Which Pages Should You Test? Testing sounds easy, but it’s a fair bit more resource-intensive than it sounds, so consider focusing on one or two key pages to start. The page with the most impact on conversions, especially from paid search campaigns, will be one reasonably close to the end of a transaction. This might be a lead-generation page, a product page, or a category page. On the flip side, though, low-volume pages don’t lend themselves well to testing. Some small businesses can do pretty well by just improving a mediocre home page, even though there will be additional intermediate steps beyond that before the user initiates a purchase. If your home page has by far and away the most traffic, you should probably be testing its independent impact on conversion events. Quick reminder: GWO will measure conversion rates from all traffic sources, including direct referrals, direct visits, and organic traffic—not just paid search traffic. Whereas paid search clicks often go to targeted, internal pages, the rest of your traffic might quite often be coming to the home page. So testing the home page is a good move for a lot of sites. If you’re concerned about messing with success, you can also set GWO to show test pages only to a portion (whatever percentage you like) of visitors, while showing the majority of users the default or current page. As you run a test, the power of the interface becomes evident. Several helpful screens show you how your apparently simple testing ideas are contributing to a complex testing process (for an example, see Figure 11-12). Quick Case Study: Planning and Executing a Multivariate Test with GWO “Big money goes around the world.” Or so says a song title by Rush, a band adored by the clients who own Rex Art, the art store and e-commerce site I’ll be using for this case study. Since the client and I like to trade Rush references, expect me to work a few into the mix here. This is not intended to be an exhaustive study of the GWO product,14 but I’ll show you how a typical small business can benefit from using Google’s tool for multivariate landing page testing. 313 314 Winning Results with Google AdWords FIGURE 11-12 A glimpse at one of my own early GWO tests. It involved four page elements of my e-book offer page, totaling 16 page combinations. The planning phase, bringing background and expertise, or at least hypotheses, into the process, is what separates real testing from random tinkering. Not only does the test improve business results, we hope it will help to educate the client on what makes their customers tick, and what kinds of considerations might inform future tests. Our high-level goal here was to improve response to the home page. In collaboration with the client, Page Zero’s role would include contributing some design elements, headline copy, body copy, and so forth, as well as an accelerated yet sophisticated (we think!) usability analysis. Beware the Sounds of Salesmen I hope this account serves as a realistic but optimistic antidote to the breathless “we increased conversion rates 500%!!” case studies so often seen out there, that don’t tell you that they did so by removing obvious nonperforming keywords from a paid search account or by fixing obvious shopping cart problems. Starting from a low base certainly makes any improvement story sound CHAPTER 11: Increasing Online Conversion Rates more impressive, but I assume most of you are looking to pick the middle-hanging fruit, not that easy-to-reach low-hanging variety. Methodology Basics Although our methodology was based on a wealth of experience and a study of relevant theory, the roadmap we created was practical. The goal was to plunge in quickly to create a properly diverse set of page combinations, while also fixing glaring problems with the home page. ■ After a brief initial cleanup comprised of “must do, can’t test everything” edicts from Page Zero, the existing (slightly modified) home page would serve as the control for the experiment. (The slight modification proved its mettle in short order, in a brief A/B test.) ■ We identified four strong potential influencers on performance, and generated creative alternatives for each page element to be tested. The test would have two values for three of the elements, and three values for the fourth, so 2×2×2×3 = 24 permutations. ■ Our developer used GWO to install the correct code snippets on the page, and to point to the location from which the various creative elements would be served. A visitor to the page is cookied, to maintain the accuracy of the test in case of latent sales or repeat visits. Assumptions Among other things, we assumed: ■ The home page is an important page on this site. A high proportion of the company’s sales still comes from “core” paid search words, and while it’s important to discuss the whole keyword portfolio, economic performance is definitely dependent on making such core words work. The home page served as a good landing page for this core keyword inventory, such as “art supplies,” but it had to improve. ■ Low volume of overall traffic would require a long test period, but related to the preceding point, being relatively high volume, the home page is an excellent candidate for a test. ■ We agreed to live with a variety of extraneous influences on the test, such as seasonality, shifting proportions of paid and unpaid clicks, and so forth. ■ The math of a higher-converting home page made this worth the time, effort, and consulting cost—even for a small business. Significantly higher profit margins carry over for the long term and also make total profit go up faster than expected in many cases, because it becomes economically feasible to buy a higher volume of paid clicks. ■ We used our background in the industry to zero in on what to test, funneled through the lenses of “better plumbing” and “better persuasion.” Thinking deeply about customers and their mindset is one thing we tried to do. The easier work was drilling in on elements of the page we thought clearly sucked. 315 316 Winning Results with Google AdWords Page Elements Chosen After arbitrarily eliminating some page clutter and moving some other things to reorient priorities, we selected four page elements to test. Depending on your site’s traffic volumes, you can’t test everything, unfortunately. ■ Headline If you have any kind of core message, headline, tagline, or other element at the top of an offer page or home page, this seems to be a perennial influencer of results. It’s telling the user what you are, but it’s here that you can also introduce doubts in the user’s mind. You can unwittingly plant negative or misleading thoughts that lead to an aversion to doing business with you. The three headlines tested could be summarized as follows: (a) the original, a long headline alluding to discounts and a free shipping offer; (b) a shorter headline eliminating the free shipping offer and alluding to a wide selection as well as the discounts; (c) a shorter headline alluding to trust and the company’s history selling online since 1995. ■ Body copy This is a company proud of its dedication to its niche and its long tenure online. But we simply wondered about the time-honored long-copy vs. short-copy conundrum. I rewrote the fairly compelling long copy to reduce the number of benefits and information points, while hopefully conveying the same message. At this stage, I feel like the test isn’t quite right because the short copy doesn’t get a fair shot; it seems to unduly change the whole feel of the page by being so short. Body copy, too, falls squarely within the range of persuasion and getting a message sharply across to the target customer, but it’s potentially a navigational issue also. Do people scroll and read? Do they scan? The beauty of multivariate testing is that we don’t need a theory to tell us what is right—we just need the theory to prod us into testing something. The test can prove or disprove the theory that long copy is too long, but it does so more conclusively when the test also takes into account variables in other page elements (a perfect recipe overall). ■ Clutter-muck #1 We thought the whole site was too cluttered. There were four small boxes with different product promotions and other info at the top of the body copy. Small is still annoying if there is too much stuff. This element was tested as a simple present or absent element: yes or no to the promo boxes. It’s mainly about plumbing, in the sense that we’re looking for the statistical rationale to go ahead and declutter what we felt was a cluttered interface. Test, don’t guess. ■ Clutter-muck #2 Yet another clutter element, there was a box for a rotating featured special that pushed other info boxes, such as the shipping offer, farther down the page in the right navigation. Could this element alone contribute to lower conversion rates? We went with the present or absent test here, too, and we did it for the exact same reason as the other test: we thought the home page was too cluttered, but we needed proof. Early Feedback: Bad Plumbing Hurts, and Better Persuasion Doesn’t Help At first, we were getting some pretty mixed results from our testing. This first test ran over many weeks, and the numbers were somewhat conclusive, and sobering. Three of our four testing elements got no better, and sometimes worse, results. But all was not as it seemed. CHAPTER 11: Increasing Online Conversion Rates Read on and you’ll see that persisting to test our hypotheses for a second round was the key to truly translating concepts into action. But for the first go-round, here’s what we found: ■ “Short copy” was the biggest loser. The original converted much better. ■ Eliminating three cluttery product promo boxes above the body text did not hurt conversions, but it didn’t help, either. It seems that “clean design” has its limits. ■ Two new headlines we tried were pretty big losers, also. The longer (original) one was the winner. We believe this was because it contained the shipping offer, which is a real driver to conversions and user behavior throughout the process of filling the cart. At this point the belief that I am smarter than the client is starting to wane. I’m thinking “any escape might help to smooth the unattractive truth.” ■ The decision to remove the “rotating special” promo box in the upper-right margin was a good one. Given what a small part this played in the overall page layout, the modest improvement in conversions here was proof of our hypothesis. This eliminated clutter, but also moved (yet another) mention of the shipping offer up into view. ■ We settled on the winning combination (not exactly the same as the winner of each variable showdown, but close) as our base new home page. At this point, the winning combination was only 33.7% better in terms of conversion rates than the original, with only a 36% chance of beating all combinations. An OK result, but not too conclusive. So far, you’re underwhelmed. In the first test, the winning combination eked out only a slight victory over the original. Many of the contending combinations were so much worse than the original that the ensuing graph of competing combos showed so much “red” it was embarrassing. (Or as Geddy Lee sang in 1984, “I see red, and it hurts my head.”) What We Did for Test #2 Test #2 was much better. In debating what to include in the second test, we had to shelve several test ideas, or simply implement additional changes arbitrarily, because too many combinations hinders the effort to reach statistical significance in a reasonable time frame. We wanted to limit it to 8 or 12 page permutations this time. You don’t want to be testing for months on end. I carried over some nagging concerns about the winners and losers from before. I felt that the “short copy” I’d contributed wasn’t implemented too well. In some page combinations, the short copy came out looking weird. I decided the solution would be to write “medium-length copy” while continuing to attempt to improve the messaging. I also tried to exercise some control over visual layout. Finally, I experimented with signing it more personably—the Rex Art Family—as the family-owned business aspect is touted in other parts of the messaging but for some reason, in the original copy, that signature said “staff” rather than “family.” I also thought the “original” headline still wasn’t optimal. But rather than tinker with it again, I tested a font and color change. Finally, we decided to learn more about the “clutter boxes” above the text. Conversion rates were about the same when we eliminated them. What if we added three more of the pesky 317 318 Winning Results with Google AdWords little critters? My wife looked at this in action before we ran the test and exclaimed “yikes, that’s just wrong. I’d leave the site immediately.” Her concern proved to be more or less warranted. Testing copy variations, headline font, and clutter boxes gave us eight combos this time around, a relatively simple 2×2×2 test. The Consumer Response This time around, the test reached statistical significance ahead of the expected schedule. GWO (rather than the familiar “you have an estimated 29 days to go…”) shouted “Congratulations! Combination #4 is the winner!” a mere ten days into the test. Ah, the payoff. So what exactly happened here? My persistence in trying to improve the sales copy was rewarded. The first time out, we didn’t prove that “short copy worked” but we also didn’t prove the original copy was optimal. As it turns out, my medium-length copy worked well with other page elements. It included a revamp of the H1 level headings and a change of the heading fonts, but also another extensive copy rewrite. This was a big winner, proving that clarity and persuasion are integral to the customer experience. It also proves that iteration is important in copywriting. As superstitious as it sounds, your subsequent drafts always seem to get better when you allow the first version to age on the page (really, in your mind) like a fine wine. The new, crisper font and color for the headline grabbed an early lead, but convergence set in and it didn’t make any difference in the end. Bear in mind, different wordings for the headlines had been clear losers in test #1. People were reading and noticing what was being said, the first time around. And what about the clutter boxes? Having three boxes above the text wasn’t a negative, but six boxes was a slight conversion killer. A no-brainer? Maybe so. But now we know. I doubt we’ll be going to nine boxes. As we concluded the test, the winning combination was converting 106% better than the winning combo from the first test, and the second- and third-place combos, 63% and 51% better. After two tests, we were clearly doubling conversion rates from the original. The changes pursued here stop well short of a full site redesign, multichannel branding campaign, or business plan changes, which could have a more dramatic effect. But a doubling of conversions from mostly organic referral traffic, and broad paid terms like “art supply” arriving at the home page, is no small victory. Despite the increasing accessibility of complex testing to the average business, the multivariate nature of the interactions makes this process more complex than one-dimensional, haphazard hackers and tweakers often let on. As for practitioners of the “old ways” of testing response—to close out with another lyric from Rush’s Neil Peart—can they “face the knowledge that the truth is not the truth”? Can you say “obsolete”? Will This Testing Potentially Harm Organic Rankings? For a small company to so ruthlessly test its home page, especially when it’s responsible for more than 50% of the company’s overall traffic due to favorable organic rankings, seems reckless. CHAPTER 11: Increasing Online Conversion Rates Jamie Roche has advocated such testing,15 and GWO spokespeople have stated that such tests won’t harm search rankings. But where’s the proof? Is it a question of appetite for risk? A question of whether the conversion rate improvement offsets any disruption to rankings that leads to a decline in organic traffic? Or are rankings left largely unaffected? Are there cases where organic search rankings might actually improve, as in the decluttering effort that leads to potentially quicker page load times and more user time spent on site, which are quality signals that might increase organic and paid search standing? I’m afraid this elephant in the room is so complicated that I’ll have to mix that metaphor in a future publication, perhaps titled The Complicated Elephant on the Home Page. But clearly there is enough doubt here to suggest that no one should just plunge into arbitrarily testing important pages, without understanding the risks and rewards. GWO Does A/B Tests Too Good news! If you want to run a simpler, A/B test, GWO is a handy interface to run one of these as well. It has specific setup options for A/B testing. In using GWO for this, you’ll benefit from the quality of the reporting interface, and fantastic features like showing the test page to a smaller proportion of incoming visitors, to minimize the potential negative impact of testing on your business results. More Advanced Testing Services GWO isn’t geared to handle every complex application. Some leading third-party testing software/services are available and well respected for allowing large businesses to conduct advanced multivariate or A/B testing. These include Vertster, Optimost, Test&Target by Omniture, and SiteSpect. SiteTuners.com is a well-known leader in the service end of this business, marrying advanced mathematical analysis and consulting with proprietary software. A Few Things to Consider Testing In retail, it’s customary to spend time testing the actual content of your offer. What is your customer getting, for what price? You can always test to see if highlighting the following elements improves your conversion rates: ■ ■ ■ ■ ■ Free shipping (and how that benefit is conveyed) Free gift or add-on Time-limited discount Bulk discount Price increase or decrease The last element is one that many sellers overlook. Sellers of software and information often underprice their material. Unless you test the total revenue potential of different price points, you may be leaving money on the table. You might also be projecting a discount image that could hurt you long term. On the other hand, your goal might actually be to increase your customer base quickly, if you calculate that lifetime value is potentially high. In that case, you might consider lowering prices if you can see a significant volume increase at that lower price point. 319 320 Winning Results with Google AdWords If They Don’t Buy, Get Them to Do Something Many businesses sell products and services with a price point that is just so high, the conversion rate is too low to gain any measurable feedback for many months. If your business is like this, you need to create proxy metrics. If a potential customer of Cubicles, Etc., requests a fabric sample, for example, this generates zero revenue at first. But if fabric sample requests tend to convert into sales at a consistent rate, then the conversion rate from clicks to fabric sample requests becomes a valuable metric, one that occurs with more frequency than sales. When you’re measuring conversion rates, then, keep in mind that in looking at final sales, you might not be measuring the most appropriate or helpful conversion rate. By helpful, I mean the conversion rate that will help you analyze data and quickly adjust your campaign to respond to it. An early client of mine, Brookbend Outdoor Furniture, considered distribution of a color catalog to be an important metric even in the absence of an immediate sale. Similarly, another small company, Bruce Baird’s California Golf Schools, considers a brochure request as an important metric, along with an online information request (a lead, if you will). Many service businesses capture leads or other “weak” expressions of interest given the infrequency of, and delay in completing, sales. The goal is that you organize your sales process such that you’re measuring something, so that you can take the feedback on that conversion rate and improve from there, for example by adjusting bids, deleting ad groups, or turning off content targeting. Earlier in the chapter, I warned about taking “weak” or “soft” conversions as gospel in testing. True, they don’t equate 100% to revenue. But many businesses feel strongly that customers converting to these soft actions are consistent in their behavior down the sales cycle, so they’re willing to highlight the soft conversions in their response tests on the site. In fact, without using these, they wouldn’t be testing at all. If you are one of those businesses that has a long sales cycle, don’t put off testing; instead, build a system that allows you to test some kind of a response. You might be amazed at what you find. Without real testing, you’ll be relegated to looking woefully at web analytics stats like high bounce rates. Seeing that a lot of users “puke” when they arrive at a page is not only uninspiring, it’s relatively uninformative. What Are Typical Conversion Rates? Just a few of the examples in this chapter indicate the wide variety in conversion rates. As I discussed earlier in this chapter, two Anonymizer.com landing pages tested by Lee Mills were widely disparate at 3.2% and 9.6%, respectively. My advice is, don’t try to guess what others in your industry are “converting at.” These numbers can be artificial anyway. If your AdWords account is restricted to only very highly targeted words, you might convert at a very high rate. But that might mean your volume is too low. The more words you add in an attempt to increase volume, the greater the chances that your average conversion rate will fall. But if these new words were quite inexpensive, they’d perform well from an ROI standpoint regardless of the actual conversion rate number. If you read reports on typical industry conversion rates, take them with a grain of salt. Conversions are highly dependent on what type of listing users clicked on, what type of search they did, and so on. Conversions also might vary from product to product. CHAPTER 11: Increasing Online Conversion Rates For companies generating certain types of leads, conversion rates could be higher or lower depending on how much effort is required of users filling out a contact form or survey. On its core (most targeted) keywords, an insurance-related campaign I worked on saw 20% of the visitors who clicked on an ad beginning to fill out a survey that would ultimately generate a lead for the sales force to follow up on. Only about half that number—10%—completed the survey, thus counting as a “lead.” That might have indicated problems with the usability and smoothness of the survey process, but some of the drop-off could be attributed to a normal filtering process, where inappropriate prospects dropped off as they discovered they did not qualify. There is always something to improve in any online sales process, but by and large, this result was satisfactory. What wasn’t so easy to take was the escalating prices on clicks for those core keywords! Once generated, only about 10% of the leads turned into sales. Because 10% of clicks had turned into leads, only about 1% (10% of 10%) of the clicks on these core words ultimately turned into sales, meaning that 100 clicks were required for each sale. So in this case, at a hypothetical cost of $5 per click, the cost per sale was $500. Depending on how you look at it, this could indicate a problem with the sales force, but it also could have been attributed to the quality of the leads. Or, it might simply have been normal for the industry. The horrible secret (though it should come as no surprise to anyone with empathy for consumers) is that products and services that aren’t in demand generate conversion rates close to zero. Confused visitors clicking on misleading ads convert at rates near zero. And inappropriate, poorly thought-out landing pages convert at rates near zero. No matter whether your industry benchmark is 0.5%, 1%, or 10%, you’re clearly getting nowhere and likely losing money quickly when your conversion rate is zero. I wish I could say that never happened. In reality it does, because in a competitive marketplace there are often too many sellers of goods and services, and not enough buyers. Retail Landing Page Design: Focus vs. Selection One of the key dilemmas in landing page design is whether to be highly focused on a particular product or to also provide related product suggestions or similar products in a category, to evoke selection or to trigger a Goldilocks effect. Category Page vs. Single-Product Page If you run a site with a catalog involving a lot of individual products, what converts better, a category page listing a variety of products under a given heading or a product-specific page? For starters, that depends on the ad. For example, a specific brand name attached to a wristwatch wouldn’t take potential customers to a huge selection of all watches. At the same time, they probably wouldn’t take them to the page for a single watch unless their selection was small or that watch was a particularly hot item. They usually wouldn’t get more specific than a category page for, say, a variety of Timex watches, simply because most watch makers have hundreds of individual models and it would usually be impractical to build a campaign with several thousand ads all going to separate landing pages. 321 322 Winning Results with Google AdWords Most of the experts I’ve talked to lean towards category pages, but none of them rely exclusively on them. A category page, as long as it looks inviting (such as the nicely designed “coffee, tea, & espresso” page at KlinQ.com in Figure 11-13), offers an interesting happy medium between the home page and a single product page. Keep in mind that if everyone already knew that they wanted a “Beacon Hill Sugar Bowl with Spoon,” there would be no need for search! Consumers, at least those who are shopping for gifts or housewares to upgrade their current lifestyle, want to browse various designs, brands, and sizes. A category page is often a great place to start. This can be enhanced with “featured suggestions” to give the user a sense of the breadth of the site’s selection. The thing about rules, such as “minimize clicks at all costs” or “reduce the number of items on the page,” is that you need to have enough flexibility to violate them for a good reason. By all accounts, there is absolutely nothing wrong with showing the shopper images of several related products in a category. If a brand is strong enough, or if the site makes a lasting enough impression on users, they may return later to buy one or more products, or they may decide to buy something they weren’t considering when they first performed a search. FIGURE 11-13 A nice category page with custom photography and cosmopolitan flair. The text probably should be tested further. CHAPTER 11: Increasing Online Conversion Rates Your site may not have an obvious category that relates to every popular keyword that you’re bidding on, but you can still take the user to a dynamically generated “site search results URL.” The Pier 1 Imports site has no category page for “director’s chairs,” for example, but users are still taken to a page full of director’s chairs by way of the URL for a site search. (Depending on your situation, you may want to work with your site developer to hard-code such results pages and make them into shorter URLs, or indeed turn them into category pages.) Misty Locke of Range Online Media, who works on Pier 1’s campaign, has stated that pages like this selection of director’s chairs in Figure 11-14 do quite well in terms of conversions from click to sale. Locke has also found that successful results came as a result of testing the search advertising copy. She found that an ad with flair, including the phrase “lights, camera, action,” produced higher conversion rates on director’s chairs. She believes this may be partly as a result of Pier 1’s strong brand and its middle-class target audience, who already know basically what type of product they want. Advanced shopping cart functionality (or even more advanced personalization technology at leading destination e-commerce sites like Amazon) will do a good job of suggesting related items FIGURE 11-14 While singular in focus, this page offers the shopper an easily navigated selection of director’s chairs. 323 324 Winning Results with Google AdWords Goldilocks and the Three Soup Ladles Consider a “good, better, best” attitude when working on related product layouts, to encourage users to opt for the “middle” or “top end” product depending on their psychological inclination. Moving more buyers to the middle category (“Goldilocks approach”) can be achieved by putting the high-end product slightly out of reach, moving the midpriced item up in price, and positioning the lower-end product as too basic, or somewhat deficient. Need anyone remind you that even a soup ladle can be “aspirational”? that the searcher might also want to look at. While that may appear to violate my rule of thumb “don’t suggest other things until they’ve become a customer,” some retail environments are more amenable to users browsing among various items. Indeed, some sales of lower-priced items may have virtually zero profit margins, so it’s incumbent on retailers to make potentially well-heeled customers aware of a couple of related higher-priced items as well. Ensure Keywords Are on the Landing Page We’ve discussed scent. Let’s reinforce the point, perhaps ad nauseam. (If you’re puking, skip ahead. There is no Back button in a book.) In search, consistency is key. You already know that ads often receive a higher CTR when you include the searched-for keywords in your ad title and/or ad copy. This applies to landing pages as well. Presenting case studies on his clients Anonymizer.com and St. Bernard Software, Lee Mills has stressed on several occasions that taking care to add core keywords to the landing page will almost always raise conversion rates. (It’s pretty cool that Lee made this point before Google ever introduced Quality-Based Bidding, and before folks started “harping on that stuff.”) This is why many companies will write several dedicated landing pages to improve conversion rates. For example, if you’re running ads on car insurance as well as personal watercraft insurance, you’d probably get better response if you had a separate landing page for each including those keywords on the page, rather than taking all searchers to the same generic vehicle insurance page. One of my former clients, a financial institution, wanted to focus heavily on home mortgage refinancing, but during the AdWords campaign, their IT department was slow in creating a tailored landing page for that product line. Instead, those who clicked on the ad for mortgage refinancing were taken to an application form tailored to new homeowners! Needless to say, conversion rates were poor, as many visitors to this page immediately left. When a more targeted landing page was finally developed, conversion rates doubled. Should be obvious, but until recently, the “get me bottom line results, but I have no control over the website” crowd had the upper hand for some reason. Today, the “scent-savvy” community is gaining the upper hand. CHAPTER 11: Increasing Online Conversion Rates Web Credibility The state of mind of many new visitors to your site can probably be summed up in one word: incredulous. Who are you? Why should they believe you? Will you deliver on your promises? Will you rip them off? When I go through the swift process of buying a book on Amazon.com, I’m 99.9% certain that the information I see is reliable and that my order will be fulfilled to my satisfaction. That degree of certainty drops significantly if a site is less familiar to me. Unless you are a major brand or are dealing with a repeat customer, you have a fair bit of work to do to ensure that the visitor is “in your camp,” taking your word for what you claim on the site. I’m convinced that if everyone took the various principles of web credibility seriously, starting with a deep understanding of the fact that online interaction can still feel distant and unfeeling to many, most of the other pieces of the puzzle of creating better conversion rates would fall into place relatively easily. Do yourself a favor and pick up a copy of B. J. Fogg’s book, Persuasive Technology, and focus on Chapter 7, “Credibility and the World Wide Web.” B. J. Fogg and Stanford Research on Web Credibility Web surfers face an ongoing challenge: to “determine what information is credible.” Web designers, to connect businesses and organizations with potential clients, must “create highly credible sites.”16 Do we actually know what increases or decreases web users’ perception that the information they find on a site is credible? Thanks to studies carried out by the Stanford Persuasive Technology Lab, we actually have a very good idea. Between 1999 and 2002, Fogg’s team conducted largescale studies as well as some smaller pilot studies involving over 6,000 participants in total. The 1999 study involved 1,409 participants, and the 2002 study involved 1,649. The average age in the first study was 32.6 years; in the second, 35.5. More than half the participants were Finnish; most of the rest were American. Gender was about 45% female and 55% male in both studies. Participants were typically middle-income college graduates with extensive Internet experience.17 By and large, the findings do not contradict the investigators’ hunches about what factors increase and decrease credibility scores. According to the Stanford studies, the following are some key factors that increase credibility: ■ ■ ■ ■ ■ ■ The site gives the organization’s physical address. The site gives a contact phone number. The site looks professionally designed. The site lets you search for past content. The site is linked to by a site you think is believable. The site has articles containing citations and references. 325 326 Winning Results with Google AdWords ■ ■ ■ ■ The site lists authors’ credentials for each article. The site provides a quick response to customer service questions. The site was recommended by a friend. The site represents an organization you respect or a news organization. Want to blow your credibility? The following seem to be no-no’s: ■ ■ ■ ■ ■ ■ ■ ■ ■ ■ The site is sometimes unexpectedly unavailable. The site makes it hard to distinguish ads from content. The site is rarely updated with new content. The site is difficult to navigate. The site has a typographical error. The site’s domain name does not match the company name. The site links to a site you think is not credible. The site has a link that doesn’t work. The site automatically pops up new windows with ads. The site takes a long time to download. The people—at least those in this study—have spoken, and the preceding are the factors they say most influence their perceptions. In the real world, some of these mistakes can be even more damaging than the study might indicate. And some of the “must-haves,” like providing contact information, are becoming more important all the time. Bear in mind that Fogg and his researchers believe that credibility is made up of two dimensions: perceived trustworthiness (unbiased, fair, honest) and perceived expertise (knowledgeable, experienced, intelligent).18 For a handy summary of web credibility guidelines, visit http://credibility.stanford.edu/ guidelines/index.html. Don’t Neglect Site Search Recall that an element of web credibility is the ability to search for archived material. On an e-commerce site, people are looking for products, not articles. It’s vital that there be a search box to help consumers who are having trouble navigating through your categories to find the item they’re looking for. Site search has become increasingly sophisticated. Successful e-commerce site developers know which options are available and whether they are built into the shopping cart package you’re using, or whether a customized version of a third-party solution might be required.