Nội dung

CHAPTER 6: Big-Picture Planning and Making the Case to the Boss ad-supported online content. Large media companies, in particular, are beginning to do this. Get the feeling that this might be a controversial practice, because Google is also a large media company directly competing with some of those large media companies wishing to find cheap ad space on Google? That’s the feeling I get. Because driving “paid traffic to ad-supported content” is a respectable cousin to the less-respected “thin content” strategy of click arbitrageurs (recall the discussion of click arbitrage in Chapter 5), it is a trickier area because of the arbitrariness of Google’s approach to quality-based bidding. Information sales in general have been tougher to achieve through AdWords lately because of the risk of falling into “false positive” territory, whereby a legitimate offer is being painted with the same broad brush as business models that Google now says typically incur low scores in automated and human assessments of website quality. Google pointed to some types of “e-books” and “get rich schemes” in a recent overview of business models that tend to get “slapped” by low Quality Scores.9 For the legitimate information publishers among us, let’s hope Google lets that enforcement pendulum swing back a bit. Assess Your Sales Process Knowing that you want to reach customers through paid search marketing, unfortunately, does not automatically translate into a successful AdWords campaign that makes money for you like clockwork. Attaining decent conversion rates on targeted clicks depends on developing a solid game plan. Most businesses today need to maximize their revenue per customer, especially as the cost of gaining a new customer in this channel rises. If you use a shotgun approach to your ad copy or landing pages, you’re less likely to succeed than if you have clear, concise goals. First determine what type of customer relationship you want to establish. Equally important, think about how those relationships can be turned into additional revenues in the future. If your competitors are calibrating their bidding strategy to the lifetime value of a customer, and you’re only looking at the immediate benefit, you may be underestimating the value of each click. You could wind up underbidding and letting competitors increase market share at your expense. What’s Your Goal: Retail Sales, Leads, Registrations, Buzz, Subscriptions? Even where you have multiple goals for each customer relationship, you should make every effort to isolate a main objective for clarity of the sales process, as well as for the sake of benchmarking how well you’re doing. From there you can expand to the other details and prioritize them in a sensible manner. For example, if you know that making an immediate return of 100% or more on the ad spend through retail sales of items such as LEGO toys is your goal, that’s enormously helpful to understanding what can and can’t be accomplished with your AdWords campaign. A further stipulation—that dump trucks and space stations are higher-margin items and thus worth bidding 165 166 Winning Results with Google AdWords more on—might be helpful here as long you don’t waver from a tight focus on gaining new retail customers. Finally, in this hypothetical scenario, you might want to remind yourself not to bid any more than 20–25 cents on the lowest-margin items (such as that long-forgotten “How to Remodel Your Kitchen with LEGO” book by Bob Vila). Remember, these are initial and primary goals for your sales process. You literally cannot function if you don’t pick one primary campaign metric and one primary purpose for any given landing page, and use the performance of clicks driven to that page as a yardstick of campaign performance. That does not preclude you from selling other things to these customers or other customers, or from convincing them to become a member of your frequent buyers’ club, or entering your contest, or... all of the other plans you have for your business. But you can’t attempt to do all of this at once. You should engage in, at least, a formal process of narrowing down your main general goals; following that, you’ll want to select campaign metrics like cost per sale or ROI and stick with them. In the informal narrowing process for the initiative, then, you might think “emphasize converting first-time visitors to buyers of high-margin products; failing that, get them to buy something else.” Consider other goals as secondary. From this type of thinking, you’ll probably gain valuable cues as to whether your website is suitable to attaining your goals. If pages are cluttered with other things you’re asking users to do, they’re likely to go away confused or perform actions that are more like browsing than entering into some kind of relationship with your company.10 At this point it’s worth noting that strong brands don’t always have to follow the rules to the letter. You can get away with some clutter and disorganization because users already accept you as a vendor; they’re going to poke around your wide selection or other offers more often than they would with a niche vendor, where they’ll be looking for credibility cues. A famous case is the tale of one particular Amazon.com test that showed an initiative to “radically declutter” many pages on the site resulted in lower sales, so the company quickly reverted to the full-featured pages users have come to expect. We’ll return to these themes in Chapter 11, which discusses increasing conversion rates. Lessons from a Seafood Company’s Campaign One of my former clients, a global conglomerate that owns a popular supermarket frozen fish line, decided to test the waters for online sales of fresh lobsters in fall 2003. At the time, they were concerned that conversion rates might be low because consumers, trained by years of seeing their displays of low-cost frozen products in supermarkets, might have trouble perceiving their products in higher-priced “fresh delivered” categories. In spite of assurances of special packaging to ensure freshness, this fear turned out to be well-founded. Conversion rates were poor. This deep-pocketed company was being outbid on AdWords by owners of small boutique fresh lobster shops. By all indications, those boutique shops were converting more sales because the more tactile and “small feel” of their operations and brands reassured customers that they were getting fresh goods right out of the ocean, worth paying extra for. Since the campaign was based in the United States, some buyers possibly liked the idea of supporting a smaller local business as opposed to a Japanese-owned conglomerate. CHAPTER 6: Big-Picture Planning and Making the Case to the Boss Despite this shortcoming, as I saw it, the fish conglomerate’s fresh seafood campaign objective was pretty clear: sell specialty fish products, especially lobsters and a couple of other rare, high-margin items. If they had also tried to promote their frozen lines, build their brand, get consumers to join an email list, or enter a contest, I would have had a harder time helping them out. From an AdWords standpoint, focusing on converting visitors to buyers of identifiable products would at least translate into a campaign that could be measured and improved. As I’ll discuss later in the chapter, I had to debate this with the client. Sure, I think opt-in newsletters are neat, but do they sell fish? Do they sell enough fish to make up for the cost of driving people to the site? It’s unlikely, especially given customers’ increasing propensity to ignore email. While I thought the fish company’s primary campaign goal was worthwhile and achievable, I ran into a snag with the retail expert who’d been brought in to provide a second opinion. I suggested a streamlined shopping cart that got the customer to place the order before filling out the shipping and billing info. Experience has shown that with complex shopping carts and preregistration procedures, shopping cart abandonment rates are very high, thus preventing the initial transaction from occurring in too many cases. Clearly, someone on the verge of shelling out $100 for fresh mussels to be shipped overnight in special packaging (something that does require quite a leap of faith) would be a potentially lucrative customer to have. Therefore, to my way of thinking, you should do everything you can to avoid turning them off just before the successful completion of their first-ever transaction, especially since clicks aren’t free and it takes as many as 50 to generate a customer on the relevant keywords. As too often happens, though, the combination of a third-party web development firm that apparently didn’t care about usability or revenues unless they were paid extra to do so, and a retail expert who seemed to be living in the 1950s and believed that a good way to develop loyalty was to bombard potentially valuable customers with to-do items instead of turning them into customers quickly, the shopping cart abandonment problem wasn’t solved. This shopping cart’s outmoded seven steps to checkout were killing conversion rates. With each additional step to checkout, more potential customers fall by the wayside. This is simple, unwavering math. To make a sale online, reduce the number of steps wherever possible. Amazon didn’t patent “one-click” ordering for nothing. What’s Going on Behind the Scenes? In the case of the seafood merchant, something very simple was made into something complicated. This led to more questions than answers. Should we rehire the developers to fix this issue? What about a focus group? Are there any studies we can look at? Probably, but in some cases, a focus group of one or two people can tell you what’s glaringly wrong. Why does the retail expert assure us that the extra step doesn’t matter when it clearly does? Budget allowing, empirical proof can be gleaned from user testing of the sort that is done by usability labs (Nielsen Norman Group are pioneers, but now these kinds of labs are widespread). But even without that, proper “funnel analysis” using a well-customized installation of Google Analytics, Omniture, or other analytics package would provide backup for early decisions about how to rebuild the cart. 167 168 Winning Results with Google AdWords With less budget, but some common sense, all that needed to be done here would have been to look at other checkout processes around the Web (some of them belonging to quite dinky-looking, but very successful, smaller businesses), and engage a web development firm that would implement a better checkout process as if usability issues and cart abandonment rates were their business, not just the client’s problem. In this case, “check out without registering” should have been an obvious and easy option to implement by a qualified programmer. Some web development shops are no better than rote implementers of third-party solutions on a “white label” basis, and don’t even understand the technology well enough to go out and find a different, more appropriate canned solution, let alone having the technical chops to write custom applications. Rather than getting caught up in outdated retailer-centric moralizing about what a typical customer should or should not do, today’s online marketers must attack problems pragmatically and test responses iteratively. In the earlier example, we needed to focus on lowering cart abandonment rates so that the initial transaction could take place and the fish would get to the valued customer’s door. Regrettably, since it wasn’t clear that that was the primary goal, it was easy to forget about this and to gradually focus more and more on other, less quantifiable priorities. The debates about the seafood site didn’t end there. Another goal was to ask people to sign up for a free newsletter. This would contain free recipes and offers, and would either help the conglomerate to build their brand or lead to an online sale (the one this consumer didn’t complete in the first instance because he or she wanted something free instead). This goal created new problems. How much is a newsletter subscriber worth? How much should we be bidding on keywords in order to attract the types of people who might not want to buy today but, based on feel-good communications including tasty recipe ideas, over time might become good customers? Which keywords should we be highlighting to attract those kinds of customers, as opposed to customers who might be expected to buy lobster today? Who do we contact to install the tracking software on the appropriate pages associated with successful newsletter sign-ups? Nobody knew right away, and answers were slow to come, since this was a secondary goal on its way to being a tertiary goal. It just sort of hung there as another thing to look at. This opens up a much broader topic—the current cluttered state of permission marketing through email—but for the present purpose, marketers do need to clarify their customer relationship strategies. Most know in general they “need a newsletter” (because that’s the conventional wisdom), but how aggressively should they pursue the effort of building up that subscriber base, and how much (if anything) should they spend on it? Many of us assume that emailing customers is a good thing, but is it? The exercise of thinking this through might lead some marketers to conclude that building this “permission asset” may not be worth the cost today. Take a hard look at the newsletter idea, which many marketers today just haphazardly fold into their thinking based on best practices that are five years out of date. Perhaps the act of asking customers if they’d like to join something like a newsletter should be reserved for the most enthusiastic subset of customers, or not undertaken at all. Customer Relationship Strategies For marketers looking to maintain relationships with a loyal customer base, it will pay to keep thinking about their needs in a cluttered world. Consider less intrusive formats than email, such as RSS-based news feeds, blogs, or a nonintrusive discussion group or social media platform. CHAPTER 6: Big-Picture Planning and Making the Case to the Boss If you’re a real chatterbox, you can even Twitter (Obama does, why can’t you?). I know that many of the current social media fads will die out, but some will thrive, so don’t be afraid to experiment. (My acid test seems to be: if “real people” are finding success in leveraging such strategies for business advantage, I take it seriously; I do not put much stock in the exhortations of insider technorati types such as Robert Scoble or Steve Rubel, in the so-called “echo chamber” of the “blogosphere.”11) Because other marketers have been abusing permission and creating a “tragedy of the online commons” situation over the past five years, you need to think ahead if you want to form lasting relationships with your customers. Don’t assume that email is a must-have channel, in particular if your efforts to build out that channel are killing or confusing your online sales process.12 Consider this: there isn’t any law against using direct mail to send good customers offers. A postcard from a sunglasses retailer won’t offend anyone. Believe it or not, phone calls from the suit salesman at Harry Rosen or the service department at Downtown Acura don’t offend me (while the upsell calls from Rogers Cable do). Somehow, businesses struggled through and maintained customer relationships in the days before email. Many of the retailer websites I visit force me to use my email address at checkout, and then they send me special offers. I’m not sure I like it. More to the point, the way many of them are doing it today might not be legal next year, or the year after that. Spam filters are also getting much more sophisticated. I’m not privy to how the spam filter works in Gmail, for example, but I notice that whereas permission-based messages I actually read seem to make it into my inbox, those I don’t look at get labeled eventually as spam, especially if the vendor emails me every couple of days. Tiger Direct permission emails seem to get through, but a volley of emails from an apparel vendor I recently dealt with, Everlast, were stuck in my Gmail spam folder along with the various Nigerian bank scam emails and the offers from “v*agra” vendors and purveyors of Russian brides. As they should have been. I might want to hear about a monthly special, and I might tolerate email every two weeks. Email me every day and we’ve got a problem. Clean up your AdWords campaign strategy by streamlining your goals as they pertain specifically to the campaign. You’re paying for every click, so you need to develop a goal that revolves around a specific success metric, whether that be cost per order, cost per lead, or cost per action. Cost per Acquisition, Cost per Order: Two Brief Case Examples The forthcoming example may be quirky, but we learn more from driving on rocky roads than smooth pavement anyway. Let’s turn to an example of a fairly messy AdWords campaign experiment that nonetheless yielded a clear cost per acquisition (I also like to call it cost per action) metric within 60 days. Then we’ll look at a case where tracking cost per order added a useful dimension to the analysis. Brian’s Buzz (Generating Newsletter Subscriptions) At the time I worked with this client, Brian’s Buzz (also called Windows Secrets) was a newsletter produced by a well-known technology author and consultant, Brian Livingston. PC World reviewed it and declared it good enough to be on their “best free stuff” list (one of the tags we tried out in the ad copy). But of course nothing in life is truly free. If you’re paying for clicks, it had better not be! 169 170 Winning Results with Google AdWords Brian’s business model was too complicated for my tastes, but I am grateful for the wealth of ideas he brought to the campaign, some of which caused us to identify key flaws in the AdWords platform itself. The business model was a mix of free newsletter subscriptions, paid subscriptions, volunteer “thank-you” tips from free subscribers, and “other income.” Going into the AdWords campaign, the goal was to increase subscriptions to the free product and then to track voluntary donations. The hope was that the cost of the clicks would be recouped within a month or so (requests are made at the end of each newsletter, every two weeks), with any additional revenues being gravy. This would require a 100% ROI within 30 days—a timetable I felt was unduly restrictive, especially considering how well the AdWords campaign went. Certain keywords attracted high volumes of clicks, and conversion rates from clickthroughs to newsletter sign-ups were consistent if not stellar (13%–15%). After two months, we came away from the exercise with an important piece of knowledge: what it costs to generate an individual subscriber. (This is often called cost per acquisition, or CPA, but at times I call it cost per action, the measurable action being a newsletter sign-up.) The cost per subscriber of about 60 cents seemed well below industry norms for such a filtered group. More importantly, the advertiser, Brian Livingston, has full control over delivery and can assess many factors in the process of attracting these subscribers. By contrast, many lead-generation services can be mysterious as to how they generate leads or subscribers for you, and the quality can be uneven at best. The absolute number (60 cents) on its own isn’t all that telling until you can determine the long-term value of a subscriber. The ROI after 60 days, focusing solely on the voluntary thank-you payments given by some subscribers (fewer than 3% chipped in, thus identifying a hole in the model that allowed 97% of readers to free-ride on the generosity of others), was only 40%. That is to say, from about $2,000 spent on clicks, only $800 was raised from these donations. But over a one- or two-year period, if we could identify, say, an average of $3 in revenues per subscriber (through advertising, book sales, paid subscription conversions, and so forth), the ROI would probably look excellent—4× or better. Some clients are shy about sharing their full revenue picture with me, preferring conservative disclosure to ensure we’re careful about bids and budgets. Later on in discussions with Brian, I satisfied myself that his various revenue streams, which indeed included premium subscriptions, books, and speaking, would have made the average subscriber worth quite a bit more than he was letting on. Focusing on the thank-you payments only, and tying bids to the ROI on those alone, was an attempt to create a campaign that was self-funding from the beginning. But longer term, each new subscriber was clearly worth more than 60 cents—$3.00 or more is my educated guess based on further discussions with the client. Even in the short term, the raw “60 cents per new subscriber” cost-per-acquisition metric is useful because it allows Brian to directly compare the effectiveness of the AdWords campaign with other lead-generation sources: Yahoo Search Marketing, subscriber generation services, banner ad campaigns, and so on. One subscription generator service that we were tracking in tandem with this AdWords pay-per-click campaign charged only 30 cents per subscriber, but as long as I was watching the tracking data, these subscribers subsequently contributed $0 in revenues to Brian’s Buzz. CHAPTER 6: Big-Picture Planning and Making the Case to the Boss Either these were just a bunch of recycled, infrequently checked Hotmail addresses being sold off to many other marketers, or these were the world’s least responsive readers. When you track your results carefully, you can prove or disprove a lot of claims, as we did here. The people clicking on the AdWords ads were reasonably responsive over the short term. At least we were able to identify a pulse! Brian’s campaign was a qualified success. Although ROI goals were not achieved as quickly as he had hoped, the campaign did work well on several levels. It could not have generated useful data or moved closer to achieving its goals, though, had the primary goal not been identified from the start: generating as many well-qualified free newsletter subscribers as possible for the lowest feasible cost per subscriber. This goal coincided well with the ad copy and keyword selection, and the custom-tailored landing page (see Figure 6-1). By being focused—by avoiding the trap of conflating a number of objectives—key metrics were established that should allow for sensible adjustments and progress towards solid profitability. FIGURE 6-1 Brian Livingston’s landing page asks you to sign up for a newsletter. Nothing complicated here, because the goal was well-established from the start. 171 172 Winning Results with Google AdWords FourOxen Corp.: Pinning Down an Allowable Cost per Order FourOxen Corp. (not their real name) sells a commodity technology service. It’s an ongoing challenge for this company to achieve decent ROI from a paid search campaign because many of the keywords are expensive. On their main keywords, large companies like Microsoft and Yahoo are bidding very high, presumably because this product is merely a loss leader in a whole suite of services these conglomerates hope to offer once they hook a new customer. By avoiding bidding wars and by tracking results, FourOxen has been able to show steady improvement in their campaign performance. In spite of FourOxen’s being a fairly recognizable brand name in their own right, and with plenty of brainpower resources at hand (over 200 employees), at first they were doing little if any meaningful post-click tracking. Because all company computing functions are centralized, I was told we could only get access to full stats reports on the 1st and 15th of every month. This makes it difficult to feel the impact of campaign adjustments. (Almost through accretion—or was it attrition?—more recently, I was able to convince FourOxen to allow me on-demand access to conversion data through Google Analytics and Google Conversion Tracker, greatly reducing the headaches associated with data analysis and campaign adjustment.) The main bits of data that were sent to me every two weeks were the numbers of orders corresponding to specific ads or keywords in the Google AdWords and Yahoo Search Marketing campaigns, and the revenues associated with those orders. By plugging in the total cost for all clicks generated by those same ads or keywords, the spreadsheet summary I received would spit out a dollar figure for whether that part of the campaign (represented by a unique tracking URL that I created for each ad and/or important keyword in the AdWords interface) was in a “profit” or “loss” position vis-à-vis revenues associated with a keyword or ad group. This reporting seemed problematic to me, because the more popular keywords generate a lot of new business at a slight loss. It makes these keywords look like culprits in an otherwise healthy campaign, because they generate a high volume of “money-losing” orders. If their click volume were lowered, these keywords might generate a smaller aggregate loss, and that would (possibly wrongly) be read as “improvement.” Clearly, there is a point at which you may be losing too much money on each sale, but unless you know what that point is, the profit/loss methodology might cause you to abandon keywords that are excellent for acquiring new customers who will become profitable repeat customers within a year. Therefore, I suggested that we also add cost per order into the reporting to give us a consistent feel for what it was costing to acquire a new customer, instead of looking only at a “loss” number that would escalate as you put more money into the campaign. For a long time, FourOxen’s management remained vague as to the allowable cost per order. At around $15, I knew that the company was making money on the average first order, so clearly, that would be a safe figure. But we haven’t been able to get some high-volume keywords much below $30—territory I still feel is profitable long term and worth mining given the high volume of new customers available through this route. Given the vertical integration of this particular tech space, I suspect a well-run company in this industry might have an allowable cost per new customer acquisition of up to $50–$60. Market share, not just pure profitability, is always a consideration in any cut-throat industry. CHAPTER 6: Big-Picture Planning and Making the Case to the Boss Something fascinating happened about a year after I finally had access to their on-demand analytics. The CEO and Marketing Director of FourOxen talked with me about a radical new business model that would potentially make a lot more sense for AdWords promotion. This was based in part on recent acquisitions that refocused part of the company’s resources on high-ticket “premium” items. While the “commodity” campaign runs on autopilot at low volumes, doing slightly better than break-even, we’re now mapping out new ground. It’s music to my ears. So often, companies conceive of “improving their campaigns” inside tight little mental boxes that hinder growth. Sometimes, new product lines, new prices, and new business models are the only things that allow them to regain some swagger in the AdWords auction. While FourOxen’s campaign objectives were not initially clear, they have become more so over time. By monitoring both the profit/loss metric and the cost-per-order metric, we gained a balanced perspective on whether any parts of the campaign were performing unacceptably. Confident that AdWords wasn’t a sinkhole, we eventually dropped the confusing profit/loss metric entirely. (It was confusing because their various prices and subscription models make it pure guesswork to assign specific dollar figures to many of the sales events that occur. Some are mere low-dollar renewals by existing customers who nonetheless used Google ads as a navigational tool; others are high-ticket subscriptions from new customers who stick around for years.) It was decided that the CPO number provided a decent “read” of campaign effectiveness. In the past three months, the aggregate CPO for the “commodity” account has carried on at an unimpressive, but still acceptable, $40. But the really exciting part is the transition towards higher-margin products. Difficulties in Forecasting Many marketing managers charged with the task of “costing out” a Google AdWords campaign may be facing an uphill battle because the cost really is difficult to predict in advance. It isn’t impossible to come up with estimates, but they may be far rougher than you might wish due to the unpredictability of the relevant variables. The aggregate cost of a campaign will be hard to pin down due to the instability of a number of factors. Forecasting Cost per Click and Click Volume When adding new keywords or changing bids, AdWords offers a Traffic Estimator tool that shows you how many clicks that bid level is likely to generate for you in a day, in what average ad position (see Figure 6-2). Multiplying this by 30 gives you the estimated cost for a month for a given keyword or ad group. By now it’s fairly common knowledge that you can’t put much stock in these estimates. There are several reasons for this. First, AdWords tries to predict how many searches the general public will perform on a given phrase, but this tends to be volatile. Second, advertisers may enter and leave the auction. Third, the very fact that it’s an auction means that prices are hard to predict, as advertisers may change their bids frequently. Fourth, your ad position is difficult to predict because it depends on how well you optimize your ads (and your competitors, theirs) for Quality Scores, which, along with your bid, affects ad position. Since ad position will affect 173 174 Winning Results with Google AdWords FIGURE 6-2 AdWords offers a tool to help you with spending projections, but real-world campaigns are volatile and often miss projections. Luckily, you can also adapt your budget day-to-day in the (same) real world. visibility, and thus the volume of your clicks, this adds yet another element of unpredictability, which Google’s Traffic Estimator tool is hard pressed to deal with. Finally, the Estimator likely does an even worse job of predicting patterns on partner sites and the content network. Geolocation parameters can add yet more uncertainty. Click volume is difficult to pin down because the early stages of a campaign should involve not merely deployment of a predetermined keyword arsenal, but also ongoing keyword discovery. You need some initial data to build on, but it’s often difficult to tell how your efforts will progress week to week. A more accurate method than using the Estimator tool is to look at what actually happened (in other words, build a small test campaign, and see what happened in real time in a given week). Unfortunately, the first couple of weeks of your campaign (for reasons of editorial review, partner and content network syndication, or other undisclosed technical issues with ad delivery) will not be a particularly good bellwether. New campaigns seem to ramp up over time before stabilizing at a more predictable ad spend pattern for any given day of the week. CHAPTER 6: Big-Picture Planning and Making the Case to the Boss For these reasons, then, costs can be difficult to predict. If solid budget figures are absolutely required, you will have to budget conservatively at first, while expecting a sharp increase in your spend (and hopefully, associated revenues) as the campaign progresses. Forecasting Clickthrough Rates and Conversion Rates I’m more comfortable predicting CTRs than conversion rates. Developing a series of ads that will pull pretty well is not particularly difficult, so it’s fine to go ahead and estimate that you will achieve 1%, 2%, or 4% CTRs, depending on how confident you are and on how ultratargeted your keywords are. Then again, even if you do hit very close to the CTR you’d been expecting, this does very little to tell you how much you’ll spend. A strange thing with CTRs is that you can have very high ones and very low ones mixed together in the same account. Content targeting might garner a paltry 0.1% CTR, for example, or it might even be ten times worse than that! Again, the rate doesn’t tell you much about the bottom-line impact. Recall that with content targeting, as opposed to the search component of a Google campaign, you aren’t penalized for very low CTRs. The game of predicting how many visitors will turn into customers after they click (conversion rate) is also subject to such serious miscalculation that you probably shouldn’t forecast. You can get rich quite easily on the back of an envelope. When I began selling my own information online, I had no idea how many copies I would sell, or what it would cost me to generate sales over and above word of mouth. I just hoped someone would buy! In reality, the patterns kept shifting. I vastly exceeded expectations early on, but after a few years, the dynamics changed, with competition, rising click costs, and the buyer’s need for constant novelty in format and content. Today I suppose my ebooks’ conversion rates on AdWords are between 1% and 3% depending on the day. But that could easily be raised to 6% by constraining the keyword list to only the most highly relevant (and expensive) words, or lowered to 0.1% by going really broad with keywords (and paying less for them). A raw conversion rate number doesn’t tell you much. It also doesn’t tell you how many “walk-ins” your initial advertising might generate down the road. Although users are cookied by most common tracking services, and tracking can accurately attribute conversions from AdWords campaigns 30, 60, or 90 days down the road, people do switch computers, clear their cookies, and so forth. Insofar as tracking is imperfect, a recorded 5% conversion rate might be a real conversion rate of 10% or higher based on “nonattributed” sales from users navigating directly to your site in any given week. One thing is clear: once you establish your approximate conversion rate, you want it to get progressively higher. You want to know about factors that create conversions when customers are waffling. But pulling a number out of thin air in advance, as if you’re entitled to this revenue, never made it happen. An Alternative to Forecasting: A $2,000 “Testing Budget” Simply try this: put aside at least $2,000 and 2–4 weeks for a trial run on your AdWords campaign. This will help you generate benchmarks for CTRs, conversion rates, ad position, CPC, and so forth, which will be helpful in further forecasting. 175 176 Winning Results with Google AdWords If that’s too much, and you’re a very small company, you can still generate 500 clicks at 40 cents per click for only $200. This will often be enough to give you a sense of where things are headed. Preferably, too, you’ll make more than $200 back from the effort! If $2,000 is too small and will make your colleagues laugh at you, forget we ever had this conversation, and up it to $10,000 or $40,000 as need be. A division of a financial institution I worked with in 2004 had a 90-day “testing” budget of about $300,000! A few companies “test” using budgets even higher than that. Based on the learning from that financial firm’s successful test, the budget for the same period in 2005 was set at $1 million. The same logic applies regardless of company size. In running your tests, try to start with very targeted terms that are more likely to lead to a sale. It’s easier to build on a modest success—a few initial sales conversions—than to know which way to turn if you strike out completely. As I discussed in Chapter 5, another good reason to target quite narrowly at first is that initial Quality Scores matter a great deal in the process of gaining early traction towards establishing a data track record in AdWords. More relevant words, higher CTRs, and so forth are always important, but particularly important in the early going. Your account history now matters, so you need to establish a strong one from the outset. That means don’t test ridiculous stuff in an account you’re planning to use for any serious business purpose. If you’ve got an old weathered boot of a failed account lying around somewhere, use that for strange tests. As I’ve made clear, I prefer testing to forecasting. In many ways, the first six months of a paid search campaign is like a lab experiment gathering valuable data. Building on that initial effort, campaign management should require less ongoing effort each month, even as its budget increases. Endnotes 1. Janis Mara, “Search-Style Ads Lift Brand Awareness, Study Says,” ClickZ.com, July 15, 2004. 2. Interactive Advertising Bureau, “IAB and comScore Release New Research on the Effectiveness of Online Local, Directory and Classifieds Advertising,” March 13, 2006, archived at iab.net. 3. For a brief summary of this study, released in May 2006, see http://yhoo.client.shareholder .com/press/releasedetail.cfm?releaseid=196082. The PDF of the full study is also freely available and should be findable with a brief search effort. 4. See also his recent tour de force, Meatball Sundae: Is Your Marketing Out of Sync? (Portfolio, 2007). 5. He also wrote “Andrew Goodman is good.” You can’t believe how much flak I took from the “search engine optimization establishment” for being the lucky recipient of that snippet of praise. CHAPTER 6: Big-Picture Planning and Making the Case to the Boss 6. Seth Godin, The Big Red Fez: How to Make Any Web Site Better (Free Press, 2002). 7. I received some friendly guffaws mixed with some howls of protest for this section as it appeared in the first edition. I’m leaving it in for both entertainment and shock value. Of course some people make money as affiliates advertising on AdWords. Many others waste gobs of time and money in a losing effort. Ever heard of the expression “the house always wins”? 8. Two excellent manuals covering lead generation and related tactics are: Michael Stelzner, Writing White Papers: How to Capture Readers and Keep Them Engaged (WhitePaperSource Publishing, 2006); and Brian Carroll, Lead Generation for the Complex Sale: Boost the Quality and Quantity of Leads and Increase Your ROI (McGraw-Hill, 2006). 9. For my overview, see “AdWords Quality Score: Can Your Business Model Be Banned?”, SearchEngineLand.com, September 25, 2007. 10. Helpful suggestions of this type are now becoming second nature to panelists at interactive “website clinic” sessions held during industry conferences such as SES and SMX. A site clinic panelist might ask an audience member to explain the purpose of several competing goals on a home page, and in general help the business (as manifested in its online communications) focus. For one popular panelist’s written work in this area, see Larry Bailin, Mommy, Where Do Customers Come From? How to Market to a New World of Connected Customers (Larstan Publishing, 2007). 11. But see Rubel’s mea culpa, “The Web 2.0 World is Skunk Drunk on Its Own Kool-Aid,” Micropersuasion.com, October 29, 2007. 12. Yes, I managed to get through this paragraph without mentioning Facebook, but that doesn’t mean you shouldn’t investigate it. 177 This page intentionally left blank Part III Intermediate-Level Strategies This page intentionally left blank Chapter 7 Keyword Selection and Bidding: Tapping into Powerful AdWords Features K eyword selection and bidding are core features of AdWords. A clear understanding of how each feature works will help you use them effectively. At the most basic level, your keyword list is a list of words and phrases that you expect your potential customers to use in Google searches. Bidding is the process by which you indicate how much you’re willing to spend each time one of your ads is clicked. Your bidding strategy is one of the determinants of how prominently your ad is displayed when a user enters one of your keywords. We’ve already covered some of these ideas, but they’re such important building blocks in your targeting strategy, they’re worth probing in more depth. How Matching Options Work Before delving into the theory and mechanics of keyword lists, let’s look at matching options and how they work. Matching options provide the tools to fine-tune your keyword list and ensure that it focuses on your target audience. Without a clear understanding of how matching options work, you may be wasting a significant portion of your advertising budget by casting your net either too wide or not wide enough. Exact, Broad, and Phrase Matching Many campaigns I’ve reviewed were not working well simply because the advertiser wasn’t aware of the fact that the default setting of AdWords is “broad match.” I’ve also seen plenty of campaigns hobbled by an advertiser determined to eliminate all nonproductive clicks by forcing ads to display on exact matches only. As you can see in Figure 7-1, matching options are activated by the use of special punctuation in the keyword list. Notice that this advertiser is using two matching options within the same ad group: exact match and broad match. The exact match for the single-word query 182 Winning Results with Google AdWords FIGURE 7-1 A short keyword list using both broad and exact match seeds is in brackets, and the broad match is listed without punctuation. Because this advertiser set aside this ad group specifically to test the performance of a popular single word along with the plural of that word, there is no need for a phrase match in this case. (A single word within quotes actually acts the same as if it were not in quotes—since there is no word order to worry about with a single-word query, phrase match would be treated the same as broad match.) Note that the broad match for the query seeds generates about five times as many clicks as the exact match. This illustrates how broad matching helps you cast a wider net without doing a lot of work. So how does this work exactly, and what is the benefit to you? We’ll get to how it works in a minute. As for the benefit, it is significant, especially in terms of ease of use and improved campaign performance. Recall that in the early days of Overture advertising, everything was an exact match, which meant that advertisers’ ads would not appear unless the keyword list contained the exact word or phrase typed in by the searcher. So, if you wanted to get your ad in front of the user typing mustang gt 2002 ford—not an unimaginable query, just a very rare one—your keyword list would have to include that exact phrase. Since anticipating every conceivable phrase a user might enter is impossible, you ended up with an enormous keyword list and still missed some potential customers. CHAPTER 7: Keyword Selection and Bidding: Tapping into Powerful AdWords Features Google’s introduction of matching options has made it easier for advertisers to generate a higher volume of clicks without being forced to generate huge lists with every imaginable keyword combination in them. As discussed earlier, there might be some benefit to pursuing “The Long Tail” of unusual keyword combinations, but this benefit can be overstated. If you are pouring 50% of your time and analytical energies into 2% of your total click volume, you’ve got a problem. You might find substantial differences in your cost per click, and probably in the degree of targeting, among the three major syntax forms available on AdWords: ■ Exact The entry typed in by the user must exactly match a word or phrase appearing in your keyword list. Enclose your keyword or phrase in brackets to force exact matching. ■ Broad Keywords and phrases entered without brackets or quotes will be interpreted as broad matching. Broad matching is designed to trigger your ad whenever all the words in your broad match keyword string appear anywhere in the user’s search query. ■ Phrase When you want a particular phrase to trigger your ad every time it appears, surround it with quotation marks. This is different from exact matching in that additional words can appear either before or after this phrase in the user’s query. The use of negative keywords, which can also be considered a matching option, is covered in Chapter 9. Exact Matching Exact matches are useful for controlling the number of clicks generated by common words and phrases. For example, if you distribute electric drills, saws, and the like, you might include the phrase power tool in your keyword list. However, left on its own (broad matched), the phrase would display your ad any time both words appeared, in any order, in a user’s search query. The result would be a large number of uninterested users seeing but not clicking your ad, lowering your CTR. The other alternative is equally bad—curious, but uninterested users clicking your ad with no intention of purchasing, thereby increasing your cost and reducing your ROI. Exact matching enables you to eliminate both of these problems because your ad will display only when the user enters the words you’ve selected, in the order you’ve indicated. To create an exact matching keyword or phrase, enclose it in brackets—[power tool]. Keep in mind that exact matching, unlike phrase matching, does not trigger your ad if the phrase is part of a larger query. Therefore, [power tool] would not trigger your ad if the user entered cordless power tool. Exact matches and phrase matches are excluded from expanded matches, which include plurals and variations. Therefore, using the previous example, the [power tool] entry in your keyword list would not trigger your ad if the user typed power tools rather than power tool. Broad Matching The broad matching option is the least targeted and, depending on your competition, may be the hardest to make work, so you generally need to bid lower on broad matches. The other side of the 183 184 Winning Results with Google AdWords coin is that broad matching offers the widest reach for the least amount of effort. For example, if you use the word tennis in your keyword list, your ad is going to show every time someone types a query like ticket prices for tennis tournaments or history of tennis. If you add other words but still use no quotes or brackets in your keyword list (for example, tennis discount store), you may capture a broader range of slight variations than you would get if you used an exact match like [tennis discount store]. Let’s say someone types the phrase tennis store discount or where do I find a store with tennis discount gear? A broad match will still show your ad on these two searches, whereas an exact or phrase match would not. While broad matching is not usually a good idea for common single keywords, it is great for specific terminology and names. The number of queries containing the word tennis is just so high, it’s unlikely your clickthrough rate, or your sales conversion rates, will be sustainable if you use such an untargeted broad match. On the other hand, if you’re in a niche industry selling Egyptian papyrus as a gift, your search volume on a word like papyrus, even used as a broad match, will be much lower, and you should be able to keep your campaign running without dropping below the minimum clickthrough rate enforced by Google, or blowing through your budget too quickly. For one word that actually means something really specific, like aromatherapy or Caligula, you might find again that the term is targeted enough that you can get away with broad matching. Those rare advertisers who have been able to keep a one-word, popular, broad-matched term such as tennis alive (achieving high enough Quality Scores to make them economically feasible) have often improved their overall performance markedly. Think of the following scenario. While most advertisers can’t keep their ad running on a particular term, you write an ad that’s just a little more relevant. Then, you filter out a wide range of irrelevant words using negative keywords (see Chapter 9 for more on negative keywords). Between these two tactics, you manage to eke out a consistent CTR of 0.7%, and wind up paying an average of 16 cents per click on a high volume of clicks. Such a term can generate a volume so much higher than the rest of the phrases in your account that it can have a decisive impact on whether a campaign has a positive or negative ROI. And since one of the main problems with online advertising is too small a reach, many advertisers love it when they can connect with a wider range of targeted prospects while keeping the cost down. Phrase Matching Phrase matching is my personal favorite among matching options. When you want a certain phrase (two or more words in a specific word order) to trigger your ad, all you have to do is enclose the phrase in quotes. For example, surrounding the phrase landscape architects with quotes (“landscape architects”) triggers your ad whenever a user types those words, in that order, regardless of what else the user includes in the query. So, adding the phrase “landscape architects” to your keyword list means your ad will display when users type queries such as these: seattle landscape architects landscape architects society east coast landscape architects who travel south for the winter CHAPTER 7: Keyword Selection and Bidding: Tapping into Powerful AdWords Features Many advertisers will discover substantial cost savings with better use of matching options even if they do nothing else to improve their campaigns. With some experimentation you will probably discover the combination of exact, broad, and phrase matching that works best for you. CPCs on Different Matching Options Many advertisers look at the differences in cost per click (CPC) and ROI that seem to occur with different matching options. It can be an interesting guide to user intent. Users who type three-word queries that include the words in the phrase match “storm insurance”, for example, might be more sophisticated on average than those who simply type the two-word query. Your bids might need to adjust upwards to reflect that, if these are better customers for you. Some advertisers seem to think that exact matching is “safer,” or “more targeted,” but it isn’t inherently so. Based on the assumption that exact matching is more targeted, one common question is this: are prices always higher on exact matches? If so, why? The answer is that it depends. There is nothing inherently “good” about exact match from a campaign economics standpoint. But sometimes exact matches—especially exact matches of three or more words—will perform relatively well from a CTR standpoint, perhaps because competitors using other broad and phrase matches are content with reaching a wider audience with an ad that is less relevant, on average, to any given searcher. If you really know the psychology of your user, you might find that you pull very high CTRs on long exact matches like [research gay travel] or [apply for a student loan]. Even if that exact match works great for you, you might not need to bid as high due to the CTR benefit you’re receiving. There’s a hole in this theory, unfortunately. Google assesses keyword Quality Score as if the keyword were an exact match—that is, in such a way as to discount match type in Quality Score calculations. So the reason you’ll generally see higher prices (not lower) on exact matches is because they’re more tightly targeted to user buying intent. Broad matches are not penalized for their (fully expected) lower CTR’s, but their conversion rates are typically lower, so they attract lower bids to achieve equivalent ROI to the same keywords in exact match form. In many cases, you’ll be a more effective advertiser if you cast a slightly wider net with a three-word broad match like how vc funding as opposed to a long exact match like [how to get vc funding]. It’s just too much work to find every long exact match phrase when you can be nearly as targeted with multiple-word broad matches. Keyword Research One of the most important tasks faced by an advertiser is the generation of the core list of keywords used in a campaign. While there are many keywords that are obvious, there are probably just as many that aren’t. Fortunately, a number of resources are readily available to help you in your quest for the most effective keyword list. The Google AdWords Keyword Tool While a large number of third-party keyword-generating software packages have sprung up, the first tool you should explore is only a mouse click or two away from your AdWords campaign. The Google AdWords keyword tool is doing an increasingly better job of providing lists of suggested 185 186 Winning Results with Google AdWords keywords to augment the lists that you may build yourself at the outset of your campaign. The best way to access the keyword tool is at the Ad Group level under the Keywords tab. Click on “keyword tool.” Or, find the link under Tools, under the Campaign Management tab. With the latest version of the keyword tool, you will see an interesting menu of options. For starters, you can enter a few specific phrases to get further information on those and related keywords. In Figure 7-2, you can see a list of keywords related to the phrase “disaster recovery.” The monthly search frequency is included for each keyword. Note that the frequency listed will be geared to the geotargeting of your campaign, so you’ll see larger numbers of searches in the stats if you’re targeting many countries or all of the United States, whereas you’ll see smaller numbers of searches if your campaign targets a small region. Don’t overlook the rightmost column heading, Match Type. This offers a drop-down list that quickly switches all the keywords in the list to your preferred matching option. To avoid the typical inefficiency of the average campaign that just uses the default—broad match—try FIGURE 7-2 Google AdWords offers an increasingly robust keyword research tool, with hard data based on real monthly Google search frequencies. CHAPTER 7: Keyword Selection and Bidding: Tapping into Powerful AdWords Features switching it to phrase match. This comes in handy if you actually select any of the keywords to add to the basket of keywords you’ll be importing directly into your ad group. The interface makes it easy to do this in an instant. No cutting and pasting is required. Just click on the Add Phrase link. On the far right of the screen, the keywords are added to the list, and you’re offered several options, including, of course, adding them to an ad group. The keyword tool will alert you to the many queries that might resemble your researched keywords but might be quite irrelevant to your line of business. Using the keyword tool can be a real wake-up call, reminding you of the fact that very few people typing keywords that include your word or phrase might be searching for what you offer, and that lack of targeting could be the explanation for why you are finding it difficult to keep your CTR high. Unlike a third-party vendor, Google can’t explicitly discuss keyword strategy. They’re too close to the situation, with issues about their own revenues, data overload and complexity, and user satisfaction with search results to consider. So this tool has its pros and cons. The worst part is that it’s rather poorly documented and won’t really go the “extra mile” for you in terms of discovery. But maybe it’s all too easy to complain. The keyword tool has gotten a lot better, and many advertisers still have no idea that it’s better than third-party tools that are more highly touted (and cost money). Say, wouldn’t it be cool if the tool would analyze the content of your home page and related pages, and come back with a list of suggested keywords, grouped into concepts? Well, the AdWords keyword tool offers this! This is the Site Related Keywords option (one of the two main tabs at the top of the keyword tool main screen). This looks at the text on your website, or any site of your choice, and then attempts to suggest lists of related keywords, broken into themes. This requires some legwork on your part. It will give you ideas, but you’ll need to incorporate those ideas into your ad group structure in a way that makes sense. But wait, there’s more! Wouldn’t it be nice if you could also cut and paste chunks of text from news articles, trade magazines, marketing materials, and various places on the Web, and plug them into a keyword tool that will come up with a bunch of related words, grouped by concepts, again? Well, the AdWords tool does this too. This is a box you can enter text into, positioned just below the main Site Related Keywords section. Perhaps it’s not the lack of Google documentation that is really the issue with the keyword tool. It’s a lack of motivation! Even as everything is handed to advertisers on a silver platter, most won’t get past default options in the AdWords interface; many will assume third-party overlays do a better job. Yep, what’s nagging at me is probably that Google is so darn low-key about their tools, they won’t be on your back pushing you or instructing you to use them. I’ve met one of the architects of this tool, Ariel Bardin, and he and his team sound quite humble to me. Ariel probably likes the tool a lot. But he probably won’t be coming to your office to play show-and-tell. So the next steps are up to you. Keyword Research Tools and Tips As good as the Google AdWords keyword tool is, it may not be enough for your needs. In that case you might want to consider one of the many third-party solutions that are available. In addition, you can pick up ideas simply by reading the news and watching TV. 187 188 Winning Results with Google AdWords Use Third-Party Tools: WordTracker Every few years, about as often as I write an edition of this book, I revisit the major third-party keyword tools, to see if there’s anything about them that has improved significantly. WordTracker (www.wordtracker.com) is the best-known keyword research tool on the market. What can be confusing is to sort out whether the various add-on bells and whistles actually add anything to the core functionality of the tool, which is finding you better keywords based on what people are actually typing. These products may be helpful, but they’re far from ideal. The flawed core data collection method—search query data from Metacrawler and Dogpile, two metasearch engines with relatively low volumes—is at least comparable to getting data straight from Google, but not quite as good. Several of the add-ons are red herrings that will have you wasting your time with ponderous planning rather than just building keyword lists directly and quickly within AdWords. And a couple of the tools, such as the very sensible lateral keyword research to give you ideas about what other words and names are common for your general industry niche, seem like they’d be easier to use if you could buy them a la carte. Instead, what you get is a bloated-up package of fairly decent, but cumbersome, tools, all at a high cost. Nonetheless, it’s worth trying WordTracker to see what it does, and a free trial will keep you busy for an hour or so. KeywordDiscovery, by Trellian Visiting the KeywordDiscovery.com website, you’re treated to a who’s who of search marketing experts, including my good friends Jill Whalen, Andy Beal, and Greg Jarboe, all singing the praises of this product in the testimonials. At the risk of not being invited to a cool Trellian party at their Australian headquarters (have a cold one for me there, Jill, Andy, and Greg!), I have to wonder if the product lives up to the hype. “Beginning at $69.95 a month” is, again, a fairly high price to pay for keyword research you could probably do somewhere else. The highlight appears to be the ability to garner keyword search frequencies from several databases, including a historical one. Does it add value beyond the Google tool? I still need convincing. Taking It to the Next Level: Competitive Intelligence One way to advance your keyword research, as well as your overall level of knowledge of your industry vertical, is to use tools that “spy” on what your competitors are up to. Certainly there may be some ethical dilemmas at stake here depending on how public the information is—you should sort these out before you proceed. AdGooroo is one among several moderate-to-low-cost offerings that offers support to paid search advertisers in the form of a suite of “power tools.” One example of a report that can be generated using AdGooroo is a list of every keyword your competitors are bidding on. (How is this achieved? Possibly, by generating massive numbers of unauthorized automated search queries, and “screen scraping” to determine which ads are showing up on given keyphrases.) In practice, this seems to be less useful than you might think. I tried an apparel-related query for this competitive intelligence report, for example, and got a list of tens of thousands of terms being bid on by Victoria’s Secret. Because the list isn’t categorized helpfully, and includes tons of keywords that are unrelated to the products my client sells, it might actually do more harm CHAPTER 7: Keyword Selection and Bidding: Tapping into Powerful AdWords Features than good. Dumping large numbers of irrelevant keywords into an account is likely to hurt the account’s overall Quality Score. Overall, the usability of AdGooroo is seriously hurting. If forced to rate the product I’d give it no better than 5 out of 10. I’ve had better luck with Keyword Spy, which offers uncanny (and more usable) competitive intelligence on the keywords your competitors are advertising on, their ad copy, their organic search performance, and much else besides. It offers levels between “free trial” and reasonably priced versions priced well under $100 per month. It’s particularly cost-effective if you want to track multiple competitors, or if you are working on multiple projects at once. Today, competitive intelligence has achieved a higher degree of significance in paid search strategy. Though not all the tools are reliable, and some guerrilla strategies are just plain mean or counterproductive, for most of us, it’s tempting (and useful) to peek under the curtain. Spyfu is another popular and well-liked tool in this area. There are several others worth checking out. For deeper-pocketed advertisers, Hitwise Search Intelligence offers more-reliable third-party data about user search patterns. A recent favorable review by keyword research expert Christine Churchill noted a price tag of $50,000–60,000. Depending on your needs, you may be able to scale back to purchase a less full-featured version for $10,000–20,000. Depending on your overall marketing budget, it could be worth every penny. Read the News As you get closer to filling out your keyword list, you’ll need to search for more insights and lateral-thinking opportunities. I find that a news article on a particular industry will often contain insights into key problems facing consumers and companies in that sector. Let’s say it’s credit problems. Where there is a sense of urgency, you’ll likely stumble across new and unusual keywords that are commonly typed by your prospects. In the credit area, for example, let’s say the article is about credit bureaus and FICO scores. As you read quotes from consumers and industry players, you may find additional real-life ways of describing consumer credit problems that your more formal research has overlooked. The nice thing is you can use a source like Google News, Topix.net, or Yahoo Finance to look up archived news on a given topic or stock symbol. Watch TV No, I don’t mean you should sit around watching Leave It to Beaver reruns. (Though, if you’re in the advertising business, try a personal favorite of mine—Mad Men.) If you come across a television ad for a company in your industry (if you’re one of those increasingly rare folks who doesn’t just avoid them via Tivo or flipping around), you might find certain buzzwords or promotions being mentioned that appear nowhere in your AdWords account. Picking up on these kinds of catchphrases is a perfectly legitimate way that you can leverage someone else’s hefty TV advertising budget. Your competitor may be drawing attention to an issue and causing a spike in Internet searches for information related to that issue. The Bottom Line: Use Your Own Campaign So far, the abstract game of figuring out what people are going to search for in your industry area has a fairly unsatisfactory feel, doesn’t it? It’s because we’re trying to generate projections based 189 190 Winning Results with Google AdWords on user search behavior, which is in fact valuable data that search engine don’t publish for just anyone to see. We’re fumbling around and guessing. Keyword research tools can be helpful for brainstorming, but if you know your business well and build your keyword list first based on that knowledge, chances are, most of what the tools tell you will elicit a “tell me something I don’t already know!” response from you. Perhaps the most accurate and useful tools are available within the actual paid search keyword advertising interfaces themselves. The most useful projection and keyword research tools for our present purposes is the one inside Google AdWords, but you can only take this so far. You might be forgetting that the most accurate source of keyword search frequency should be your historical, real-life campaign data! If you’ve built out a reasonably comprehensive account, and run the budget high enough for full delivery of your ads, you’ll have a genuine sense of how many searches are being done on your keywords in a given time period. That’s why it’s so important to test and react to real-world data, rather than messing endlessly with projections. Let’s walk through the kind of exceptional keyword data that are available to you after you’ve actually been running an AdWords campaign. In Figure 7-3 you can see some fairly standard FIGURE 7-3 Ray Allen knew roughly how many people searched for the term [seed packets] in May 2004. CHAPTER 7: Keyword Selection and Bidding: Tapping into Powerful AdWords Features reporting for a typical AdWords campaign over the month of May 2004. By looking at the number of impressions of his ad on any given phrase, the advertiser, Ray Allen, now knows (assuming he kept his AdWords campaign at full delivery throughout the month) that 5,512 people in the United States searched for the phrase seed packets or some query including that phrase, during May. The number is slightly lower than the previous month’s 6,587. That’s not a projection, it’s real, and it makes sense intuitively. People were more into planning for their gardens in April than they were in May. These patterns may vary from month to month, but it’s a much clearer benchmark than the information provided by a third-party tool that has no way of knowing what real users are doing on Google. Moreover, this advertiser knows that 5.7% of the people typing that term into Google clicked on his ad, down slightly from the 6.7% in April. Pretty good. You’ll never learn that from any “research” tool. The total cost of running that advertising to learn this? About $50. And that $50 was spent on advertising that also generated new customers. You can get as specific in your analysis as you like once you have your campaign running. In this example, 143 people searched for seed packets on May 6, 2004. And 9.0% of them clicked on Ray’s ad. Pretty tough to predict that without actually running a campaign. Keywords You’re Already Using For many of my clients, especially those in retail, generating an exhaustive keyword list is a task that looks deceptively easy, because most of this material is already on their websites. Articles and sales materials may contain much of the needed jargon that would go into a successful AdWords campaign. Product pages might contain product descriptions, product names, product attributes, or brand names. Sensible categorization still eludes many companies, unfortunately. If your categories themselves aren’t intuitive, then your paid search campaign will have a hard time working around that. Paid search experts can often suggest work-arounds, like custom-generated (more granular than exists on the default website) category pages. We can also take a basic keyword list, beef it up to be more robust and comprehensive, and then work on additional expansion. It’s not quite as simple as it looks. If you want to be really exhaustive, review not only your website and your competitors’ websites (including any keyword or description meta tags you might be able to see in your browser by viewing the HTML source of the competitors’ pages), but also your website statistics (server logs or web analytics such as Google Analytics) to see what kinds of search phrases are leading people to your site. Examples of Unsold Keyword Inventory Discovering new and untapped keywords can be like the proverbial ocean voyage to the new world. If you get there first, who knows what riches might await? The “new world” is getting more populated, however. To stretch the analogy a bit, there might not be much free land left, but there is still undervalued land. As you build your keyword list, use both deductive and inductive logic to expand that list based on what you know of the demographic profile of your target customers. In other words, develop a correlation you might like to test—whether you can sell skateboards to people who type Tony Hawk, for example. If it hits (pulls a good CTR at high volumes with low cost, and site visitors begin performing desired actions on your site, such as buying something or signing up for your newsletter), you know you’re onto something. The Tony Hawk example is cheating a bit, 191