Nội dung

246 Winning Results with Google AdWords I’ll assume that you’ve already selected the check box that shows your ads on search partner sites, a first step in expanding distribution beyond Google Search, and that you’ve already thought about which countries are good places to show your ads and selected them accordingly in your campaign settings. Getting the Most Out of the Keywords You Know You may be tempted to think that generating lists of additional keywords and throwing them into your AdWords account is a good way to make more money with your campaign. I’ll cover that, but first, are you getting the best performance out of the keywords that are already there? Deal with Your Lowest-Quality Keywords If some of your keywords have gathered poor quality history, they may be shown in lower ad positions, or they might even fail to meet the “bid requirement” on many queries, and therefore in those cases not show at all. No keyword is technically inactive today under the new regime I’ve called AdWords 2.7, so there is no urgent need to “rescue” any given keyword from “deactivated” status. That means your task here is less urgent, but also harder, because your poor quality keywords still stay active. So overall, the process of improving your account and raising volume is going to be subtler than ever before. In the old days, prior to Quality-Based Bidding, Google never had consistent advice for reactivating keywords deactivated by low CTRs. Yes, there was boilerplate advice for those suffering from “deactivated keywords syndrome,” but it tended to be ineffective. The complexity of the task of cleaning up a low-quality account has grown more complex with the advent of Quality-Based Bidding. You’ll have to make some judgment calls about how well you’ve built your account so far, and consider pausing, moving, or deleting some keywords or groups to improve quality. The usual advice about continuing to test ads, and sending users to appropriate landing pages, certainly applies in spades. First, go back to some of the fundamental advice shared previously. You should have started your account with highly relevant and narrow terms, and gradually built campaigns with broader, less targeted terms as you established a strong account history. If you didn’t do that, and your account is in tatters with poor Quality Scores all over the place, you may have difficulty digging out of this hole, because some aspect of the Quality Score formula includes an account-wide Quality Score that acts as an “overlay”—either helping to “green light” or “red light” new keywords. I recommend you contact your Google rep and ask them whether the account is worth fixing, or whether its poor start makes it unlikely that your actions will fix an intractable situation. Ask them if it makes more sense to delete the current account and start again with a new one. Generally speaking, accounts with histories are better than fresh accounts, even if some of that history is poor. But some might be lost causes. In doing this, you are possibly accomplishing two goals. The first is getting more information out of Google as to the best recommended course. Make it clear to them that you want more clicks and are eager to learn (beyond pabulum boilerplate-type advice) how you might expand your click volume. The second goal, though, is tied to the hope that they’re listening to your CHAPTER 9: Expanding Your Ad Distribution: Opportunities and Pitfalls troubles and willing to take a look at some of the stats associated with your account. If you’re lucky, maybe someone will see that your account-wide Quality Score is an undue hindrance to you, and manually adjust it. This is a gray area. No one really knows how often such manual overrides are done, and how much they actually help in relation to improving the actual signals and stats that go into your Quality Score in an automated way. At one time, there seemed to be no harm to having unsuccessful or ill-chosen keywords in your account. At other times, Google has alluded to “account-wide” calculations that can hinder full delivery of your ads. Today, it seems wise to sweep through accounts periodically, looking for keywords deemed to be poor quality. You can even get a bit more specific than that, now that Google is making available detail that will show Quality on a scale of 1 to 10. You should study the very poor quality keywords even more closely than borderline OK keywords. Don’t delete keywords if they seem relevant to you, but do be ruthless if the meanings of those keywords don’t really quite sync up with the product or service the associated ads are leading the searcher to. For example, in a campaign for a company that facilitates the buying and selling of businesses (a broad-based B2B campaign that is still quite different from similar B2C campaigns), a subtle difference in the searcher’s query (arby’s franchise as opposed to arby’s outlet) seemed to be indicative as to whether the search was most often from a consumer (who would be dissatisfied with the results if they visited the landing page on the query arby’s outlet) or a prospective business buyer, who might search for something like arby’s franchise information, but is certainly somewhat likely to search for arby’s franchise. Using predictive tools and then machine learning, Google’s Quality Score algorithm attempts to first predict, and then confirm, which keywords are irrelevant to your target customers. Tossing unrelated keyphrases into your account doesn’t help searchers, and ultimately, doesn’t help you. In this day and age, you have to tighten things up. Delete poor-performing keywords that are obviously irrelevant (such as arby’s outlet here), to insulate yourself against Quality Score woes down the road. This will have the indirect effect of allowing you to expand your overall click volume, because the offending keywords won’t have a polluting effect on account-wide quality measures. The bottom line? Any keyword that shows up as poor in your account for an extended period of time is a symptom of deeper problems. In some cases you may want to keep them running and take the chance that they won’t be hurting your overall account. But in many cases you should fix the underlying problems (poor relevance, lack of ad testing, insufficient granularity, irrelevant or poor landing pages) or delete those keywords, rather than soldiering on with them. In the example of arby’s outlet, Google’s machine learning can actually measure the dissatisfaction consumers feel when they go to a page thinking it might provide a map or business information, and stumble instead on a niche, B2B site trying to attract franchise investors. The angry consumer hits the back button immediately (or in rare cases conveys their dissatisfaction to Google through a form or email). Such behavior will tend to lower Quality Scores on particular keywords, and also lead Google to begin slapping predictive low Quality Scores on words in new accounts that have a high probability of dissatisfying users in a similar way. So at this point, I am actually offering you a warning. It seems tempting to expand your account willy-nilly. Just give me more, more, more! cries your brain. But if “more” means throwing a pile of keywords into ad groups without regard to searcher intent, the strategy is actually stupid. And Google AdWords, as sure as the day is long, will punish you for reckless kinds of expansion. 247 248 Winning Results with Google AdWords So far, it doesn’t sound like it will be easy to expand your ad distribution, does it? Well, that’s a curious thing. Google’s bias towards tight targeting creates an even bigger paradox than ever before. One way to finesse this is to choose ads that do well in CTR terms, and take a shortterm hit on ROI in some cases. If you build an account with industry-leading CTRs, Google AdWords will begin to love you on the whole, and this might give you leeway down the road to expand your account. Another way to think of it is that you need to think tight and precise, even as you expand. If you want to undertake experiments in looser targeting and experimental expansion, consider keeping those experiments (campaigns and ad groups) separate from the rest of your account, for ease of decision-making going forward. Two-Word Broad Matching Many conservative AdWords advertisers prefer to use phrase match and exact match rather than go broad with their keywords. But this can limit your distribution. To expand your distribution cautiously, choose one popular word and then enter 20–30 two-word broad-match combinations that include that word. This will show your ad to more users but at the same time will allow you considerable control over the types of queries that show your ad. The fact that the second word will typically need to appear in the user’s query will reduce the potential distribution enough to make the ad highly targeted, without ruling out users who type in long, unpredictable queries that include any number of other words. This is not entirely precise, given that Google has made broad match a ground for semantic experimentation; as discussed in Chapter 7, the broad match method is not literally rules-based and may vary somewhat as Google experiments with what it calls expanded broad matching. The additional advantage of the two-word broad match is that it can be fairly specific, so you’ll probably generate healthy CTRs. As such, you can afford to bid less and generate a strong ROI. The same general logic applies to three-word broad matches. Four-word broad matches are rarely worth bothering with. It’s generally too much effort to use long strings of words in your campaign, except perhaps in special situations. For example, you might find that a fair number of users put questions to the search engine, such as how to sell my timeshare. Even here, a three-word broad match—how sell timeshare—would handle many possible combinations, as would the two-word broad match omitting the word how. Expanded Broad Matching: Disable Only if Necessary An additional benefit of broad matching is that it invokes Google’s expanded broad matching feature, which may selectively show your add on similar phrases that include plurals, verb stems, and other close variations without making you do all the work of discovering them. Although at first, many advertisers saw expanded broad matching as worrisome when it was rolled out, Google has been careful to test the technology and to calibrate it conservatively enough that ads are not showing on all kinds of unrelated search queries. A good rule of thumb is, if you are bidding particularly high, you need to be more concerned about the potential of the expanded broad matching technology to show your ad on less relevant searches. The only simple way to disable expanded broad matching is not to use broad matching at all. But if you’ve been overly dependent on phrase and exact matching and are looking for more volume, you should strongly consider broad matching to achieve fuller coverage. CHAPTER 9: Expanding Your Ad Distribution: Opportunities and Pitfalls One-Word Broad Matching + Negative Keywords Many advertisers ignore the potential of the one-word broad match with good reason. It can cast too broad a net, and paradoxically, because uncreative or deep-pocketed advertisers may be drawn to such words, prices can be too high. The other reason that many ignore the potential of one-word broad matches is, it must be said, slavish adherence to conventional wisdom. Google does its part on this one. If you ask staff for advice on your campaign, they’ll often recommend against such “untargeted” keywords. As the imagination of the average advertiser is increasingly captivated by more targeted keyword choices, one-word broad matches may now be a relative bargain. The other important benefit here is that they can provide your campaign with huge additional volume! And for most experienced advertisers, volume is where it’s at. As long as the conversion rates generated by such words are keeping the cost of an order or lead in line with what you’re getting from other parts of your campaign, you’ll be fine. There’s a danger that these words will run into Quality Score problems, because your customers come from a small subset of phrases that include the word in question; so only a few people click on the ad. One effective way of keeping one-word broad matches in a healthy state (higher CTRs) is to enter long lists of negative keywords to ensure that the ad isn’t showing on any popular queries that are irrelevant to your business. You can continue to discover and add negatives to solidify your CTR on the broad-matched term. Another way to guard against single-word broad matches triggering too many irrelevant matches is to use the little-known technique of making them “one-word phrase matches” by enclosing them in quotes. Logically, these should actually be the same as one-word broad matches, but the key difference is that broad match allows Google’s “expanded broad matching” whereas phrase match does not. The single-word phrase match will therefore often provide a more reliable matching experience for single-word keywords. This treatment of one-word keywords is an undocumented aspect of AdWords but is informally confirmed by Google reps. To discover what phrases that include that word are popular of late, you can use the keyword tool, as we saw in Chapter 7 (see Figure 7-2), but with a twist. Select your keyword—let’s say it’s a broad keyword such as nuts. Go to the Match Type drop-down list box and set it to Negative (see Figure 9-1). Depending on what keyword you’ve selected, Google’s smart technology is going to show you a variety of popular matches on that keyword. Some will be words like “organic” that are still in the realm of relevancy. Others may be more like shots in the dark (“bolts”) that could be shown by Google’s matching technology on expanded broad matching, as the matching technology grasps at straws. If you’re broad matching, you can’t do without negatives. Add as many negative keywords as you can in order to exclude those searches that are clearly not relevant. Of course, keep the relevant ones. This means you may need to scan the list of several hundred potential negatives “by hand” (or by eye and brain, anyway). If you over-automate this part, all you’ll do is reduce your click volume to near nothing, by stopping your ad from showing on many relevant queries as well as irrelevant ones. As you can see from Figure 9-1, each potential negative keyword is easy to add to the keyword list by clicking on Add Negative. This is a lot of work, but it allows you to employ broad matching with confidence, which can help you outdo other advertisers in competitive 249 250 Winning Results with Google AdWords FIGURE 9-1 Google is experimenting with the best way of suggesting negative keywords to power users. Their current, coy method is one you should stay on top of. fields and increase volume on a stalled campaign. It can be well worth the few minutes a week you spend doing it. If you’re wondering what is the maximum amount of negative keywords I’ve added to a campaign… it’s 8,500. I’m not proud of that, or anything. Stuff happens! (It turned out that was too many, and we cut it back to 500 in that case.) Beware: some third-party vendors are accessing all of Google’s suggested negatives through the AdWords API and then passing off the list as their own proprietary list. The list is valuable, but the price of access is generally free or close to it if you get the information right from Google. And as I mentioned, if the list of potential negatives is particularly long, you do need a set of eyes and a brain to pick and choose the truly irrelevant words. Advanced Technique: “Go for the Tail” Lately, software and service companies have been moving aggressively into the paid search advertising game, some armed with ambitious business plans and fueled by venture capital. It’s now fairly common to hear such companies advocating the benefits of bidding on 50,000, 100,000, and even a million keywords and phrases. Although this strategy might make sense for large retailers with broad-based catalogs, I’m always surprised when I hear this strategy CHAPTER 9: Expanding Your Ad Distribution: Opportunities and Pitfalls being recommended for small to midsized companies, or companies with a relatively narrow offering. The logic goes something like this. Most advertisers are ignoring the huge numbers of highly specific phrases that are typed by search engine users. By examining server logs for referral phrases coming from the regular (or organic) Google Search results, we can see that in some industries, obscure phrases with only one or two referrals per month might make up 50% or more of overall visits to a website. By entering as many as possible of these phrases into a campaign, the argument goes, the average CPC will come down, and ROI will increase markedly. If the most obvious, frequently searched keywords form the “fat part of the curve” on a search frequency distribution graph, then the large number of infrequently searched terms can be called the tail; hence, to focus on these is called “going for the tail.” I also call this the “keyword dump methodology.” One problem with the tail chase is simply factual. If your “tail words” only make up 5%, rather than 50%, of your commercially viable search queries, then the attempt to find all of them can be overrated. I agree with the premise—indeed, the fact that a high number of unique queries are typed in by search engine users has been fundamental to my approach to AdWords since day one. But I don’t necessarily think it follows that the average advertiser will see a significant improvement in performance by aggressively going for the tail using word-generation software. I believe such overkill can be a distraction from a healthy focus on a variety of determinants of success or failure of a campaign. At the end of the day, it depends on the nature of your business, and the resources you have at hand. The first thing to remember is that by using phrase and broad match, advertisers are reaching much of the tail anyway. That’s the whole purpose of matching options. What proponents of the keyword dump methodology will now say is that the bidding process on AdWords may allow you to reach that tail more inexpensively than if you used phrase match. Perhaps this is true to some small extent, but it is all too often exaggerated. Advertising on exact phrases like find me a good hotel near Houston, because it showed up in your server logs as a search referral, or because some software generated this as one of a million variations, is certainly an option. But you’ll still be competing for position with others in the hotel industry (for example, advertisers using a two-word broad match including the words hotel and Houston), and you’ll still find the CPC expensive. A real drawback to going for the tail so aggressively emerges in your tracking and post-click analytics process. If you decide to track everything by keyword, you’ll be left looking through sheaves of results that show numbers of impressions and clicks in the single digits. Worse still, if you use an automated method that determines how long to keep a phrase running, you could be overanalyzing and turning good phrases off based on random user behavior on phrases with tiny sample sizes. The typical revenue associated with one of these phrases will be zero; every so often, there will be a purchase, possibly a large one, on a highly specific phrase. Who is to say that this highly specific phrase was actually the cause of this purchase? One purchase could lead you to overestimate the value of a certain phrase for months or years to come. That’s why I think it’s safer to think in terms of groups of related words. 251 252 Winning Results with Google AdWords Clearly one of the real drawbacks to using the keyword dump methodology is that the task of interpreting and acting on such fragmented results is too unwieldy for even a hard-working analyst. To this argument, tail-chasers will respond that they facilitate analysis by ensuring that similar keywords are grouped. Some will refer to patent-pending linguistic technology that helps them group words—without mentioning that Google’s own technology in this area is likely to be offered to advertisers within a year or two (and already is, to the extent that the keyword suggestion tool shows related words that you might want to consider); being thorough in the “torso”; and making smart use of matching options. Final consideration: going for the tail too early can hurt account-wide Quality Score. Hardly the low-risk proposition it’s often billed as. So we’ve really come full circle. By grouping keywords, and tracking based on those groups (assuming the software that attempts to automate these “groupings” actually works), we’re back to the methodology I’ve been recommending all along: developing an AdWords strategy that revolves around groups of like keywords. Building on Success: Hypothesize, Extrapolate, and Profit One of the easiest things you can do to increase profitable click volume is to look again at your successes and try to build on them. It’s easier to do this if you’ve made discoveries that are based on testing a particular theory. I hope a couple of brief examples will give you the flavor of this kind of “determine what works, then do more of it” method of experimenting. FourOxen and I are always pinpointing which types of online searchers are likely to be their most profitable customers. We don’t have all the answers, but we have been able to make a couple of interesting discoveries using common sense first, followed by data analysis. Recall from Chapter 6 that FourOxen is in a hotly competitive Internet-related service business. One of the few differentiators we could identify with FourOxen’s service offering amid a sea of competitors was that they pride themselves on superior customer service. (Notwithstanding the fact that my colleagues at Future Now rightly point out that customer service is generally a pretty weak fallback in an otherwise-undifferentiated product strategy.) Since many advanced users view the service as a commodity to be bought as cheaply as possible, it was becoming rapidly evident that targeting these savvier customers—at least when we were paying a lot per click to do so—was a money-losing proposition. I proposed that newbies seem like FourOxen’s best AdWords prospects. These customers would be attracted by advertisements that promised integrity and more hand-holding as opposed to ads promising rock-bottom pricing. To give that theory a solid test, I added more and more phrases that were the types of things that a confused, “unhip,” new person to the particular technology in question might type into Google. I even tried certain very broad phrases indicating an interest in starting up a new web venture, such as new web, and many others besides. Some of the terms, as advanced users might see them, would be considered mistakes or at least very awkward ways of expressing the “correct” idea. No matter. If a newbie typed it in, I wanted to show them this ad. The theory proved correct. The cost per new order for these kinds of keywords was significantly lower than the cost per new order on the rest of the campaign. That meant our work wasn’t done. Since we had strong evidence that the newbie theory was correct, the trick was to go out and find more of them. Keyword discovery in this particular realm—“newbie words”—is ongoing. CHAPTER 9: Expanding Your Ad Distribution: Opportunities and Pitfalls Another area that we tested was various relevant brand names and trade names (“industry words” or “competitor words”). In spite of the ongoing legal controversies over the use of such keywords to trigger relevant ads near search results, we do know that they’re often effective lead generators. Here again, the effectiveness of this group proved itself quite readily, so our job is to continue with keyword discovery as long as we can find new ones of this type. If we find that performance begins to degrade in either of these groups, we’ll take a hard look at recently added words that might be the culprit. As the groups get very large, it makes sense to subdivide them to test further distinctions and microtheories about what works even within this narrow realm. Keep in mind, this doesn’t mean you have to track each and every keyword. Upping the Bid and Movin’ On Up Certainly, it is likely going to cost you more to up your bids as a means of gaining more click volume. Many business owners reject the strategy out of hand. But at a certain point it’s time to take a harder look at your allowable cost per acquisition, and other assumptions. If you’re thinking of moving up into a new ’hood, you might need to change a few things—better haircut, new clothes, finer wines, etc. But seriously: if you can make fundamental changes to your business so that higher CPCs are now palatable to you, the “bid higher” strategy is not to be discounted. Let’s start with the less drastic changes, and then touch on some more advanced considerations. If some of your ad groups are performing at a significantly lower cost per acquisition than others, it doesn’t make sense to keep the bids low on such groups just for the sake of frugality. If your average ad position is, say, 3.3 on one of these successful groups, you might want to find out how much more you need to bid to push it to 2.8, or 2.1, and whether raising the average position creates an unacceptably high cost per acquisition. To keep bids too low means you’re generating too few potentially profitable clicks. But at some juncture, you will have raised bids to the point where the additional clicks cost too much. There is no hard-and-fast rule for how to approach this, and results may fluctuate from month to month. But clearly, leaving one part of your campaign with very low bids if its ROI is particularly strong makes little sense. Generally speaking there shouldn’t be vast disparities in ROI across a campaign. An ad group’s ROI can indeed be too good. An ROI of 400% might simply be an indicator that your volume is too low in that part of the campaign and that you need to bid more to increase clicks. Remember, total profit can go up even if your ROI goes down, if sales volume increases enough. Let’s get into a couple of considerations that will help you move your bids up so that you’re in ad positions 1–3 for your core keywords, even though this currently seems unaffordable. We’ll go into these points in more depth later, in Chapter 11, which covers increasing conversion rates. Launch a Conversion Improvement Program Short of changing jobs or changing your whole business model, the best way to be able to afford the higher CPCs associated with higher ad positions is to raise conversion rates (the ratio of clicks to sales or leads). This process isn’t magic—it’s mostly science. But a good degree of creativity and testing will be required to improve landing pages. Even sophisticated multivariate testing is open to the average advertiser, now that Google offers a free tool called Website Optimizer. Simple A/B testing can be effective, too, especially if you have lower sales volumes. 253 254 Winning Results with Google AdWords Focus on Web Credibility, Online Reputation, and Repeat Business Strict conversion improvement testing on particular landing pages is one thing, but you need to be thinking about related user dynamics and customer interactions as well. Especially when selling higher-ticket items, you need to disclose more about your company and soothe user concerns about security and trust issues. Go beyond your own site to research your online reputation to make sure rumors aren’t creating unseen objections in consumers’ minds. Work on your upsell and repeat sales channels and tactics. The baseline fact is: click costs will continue to rise. Tweaking bids and such will do little to keep you in the race if you don’t also take multiple steps to increase your average revenues per customer. Content-Targeted, or Contextual, Ads: Take a Second Look As I explained in Chapter 2, at one time, online advertising brokers (or “networks”) such as DoubleClick played an important role on the Web, allowing advertisers to place large banner ad buys without having to approach individual publishers and giving publishers access to more advertisers. Today, these first-generation online ad brokers are rapidly being displaced by the second wave: programs like Google AdWords, Yahoo Search Marketing, Quigo, ContextWeb, and several smaller players. (Google has acquired DoubleClick, to put an exclamation point on the story of this trend.) Ads Appearing near Content Having regularly insisted they’d stay “laser-focused on search,” Google surprised some observers in March 2003 by launching an ad network that pays publishers for displaying ads that look very similar to standard Google AdWords ads. If you’re an advertiser, the option to display your ad on these publishers’ sites appears as the Content Network check box in your campaign settings area; you’ll be asked if you want content targeting turned on or off (see Figure 9-2). The content targeting program came under a lot of fire early on due to its spotty quality, but today it has reached a new stage of refinement and many of the old problems are a thing of the past. How does content targeting work? As with many things that Google does, the exact formula is proprietary. One thing’s clear: it doesn’t work the same as the ads appearing next to search results. The key thing to understand is you’re essentially going through a stripped-down media buying process to buy online display ads (often but not always in text format) at various publisher sites around the Web. With content targeting (sometimes called contextual advertising), the keywords in your account still serve a purpose. Google’s semantic matching technology uses these keywords, along with the amount you bid, to decide whether or not your ad is relevant enough to show on a particular page. The semantic matching technology “reads” pages for meaning; it isn’t just pure algorithmic keyword matching. Ads are selected on the fly as the code on the publisher’s page loads the Google AdSense ad creative. These ad creatives (creative is an online ad industry term for the size and shape of an advertising unit) can vary in size, but they take up the same screen CHAPTER 9: FIGURE 9-2 Expanding Your Ad Distribution: Opportunities and Pitfalls Campaign settings with content targeting across the network turned on real estate as the graphical ad banners that were once ubiquitous online. Other units, such as banners, animated banners, and interactive banners that operate using JavaScript code (called Google Gadgets, or generically in the industry, widgets), are now also available through the same Google AdWords platform. Under Create New Ad in the Ads tab of your ad group, Google lists several options, including text ads, image ads (also known as banner ads), video ads, local business ads, and mobile ads. Under image ads, just as an example, Google offers eight banner sizes and allows you to upload files in any of four file formats (see Figure 9-3). Google has now rolled out a new twist that helps you control your placements more precisely: placement targeting. Essentially, you are able to use both keywords and your own specific site choices to plan where your ads show up. In selecting placements, Google offers a menu of sites that you can choose to show up on, and will allow you to enter a maximum bid for each of those placements. Note that in the campaign settings you’ll be asked to choose to show your ads either “on sites from the entire network” or “only on sites I select.” Personally, I’ve grown fond of the wider reach of the “entire network” approach of classic content targeting, since it contains a smart, keyword-matching element. But more precise targeting is the only thing that will work for some companies. 255 256 Winning Results with Google AdWords FIGURE 9-3 Setup screen for Google image ads, a format allowable under content targeting Online content is not restricted to any particular format. So-called contextual ads can be placed near discussion forum content, email messages, articles, or, as the example from the online photo sharing site Flickr (see Figure 9-4) shows, thematically tagged images. As an advertiser, you’ll need to be flexible in how you think about content, because chances are you’ll have the opportunity to show your AdWords ads in a lot of different places in the coming years. The price you pay for any given click isn’t easy to pin down, but your average costs are reported clearly in the main ad group views; more detailed breakdowns are available under the Reporting tab. It’s not an auction in the same sense as the search ad program. You won’t pay more than your maximum bid, but how much less than your bid you wind up paying can be determined by another proprietary Google formula. Essentially, Google has improved the quality of its ad network by firing some publishers, and simply reducing payouts to other publishers whose inventory has tended to convert poorly for advertisers. Google has also begun disclosing more about the types of content your ads appear near. If you’re using Google Conversion Tracker, the “content types” report will show you CTRs, spend, and conversion rates on parked domain pages, error pages, and other offbeat forms of content. Google also shows you whether your ads are showing near news stories on sensitive topics, like CHAPTER 9: FIGURE 9-4 Expanding Your Ad Distribution: Opportunities and Pitfalls Ads by Google appearing near photos on Yahoo-owned Flickr death and tragedy, or adult or suggestive content. You can now opt out of showing ads on any of these subsets of content. Control is at the campaign level. After having experimented extensively with contextual ads, I can confirm that these ads do provide many advertisers significant opportunities to increase the profitable click volume in their AdWords accounts. But the big if here is this: it will likely prove profitable for many of you if you’re able to bid lower on content targeting than you do on your ads that appear near search listings. As discussed previously, Google now offers content bidding, enabled within Campaign Settings. Don’t forget to bid separately on all of your content-targeted placements, and measure and monitor their performance separately. Even more effective, if you have the time, is to mirror your search campaigns with entirely separate content-targeted campaigns. This facilitates separate tracking, but another thing this does is allow you to experiment with different ad creative than you use in search. Because a user’s intent is far different in a contextual ad (often you are interrupting users rather than being searched for), often the psychology is different. To be frank, search is a medium to be treated as kind of sacred ground, because that’s how Google’s users (and algorithms) see it. Content, rightly or wrongly, is more amenable to gimmickry and salesmanship in ad copy. It’s not right for every brand, but you may want to try a variety of styles of text copy. You might also want to 257 258 Winning Results with Google AdWords dip your toe into the world of animated banners and other creative formats, including larger ones. The tragedy of the commons is that this “scorched earth” treatment of online display ad spaces by advertisers means users have developed banner blindness. But for your purposes, you’ll need to consider strategies to break through this blindness. Low CTRs on Content Placements Don’t Impact Quality Score Another thing to keep in mind with content targeting is that CTRs will typically be much lower than with your search ads. Do not panic! AdWords does not count CTRs from content targeting against the performance of your account. So you won’t rank lower on your keywords because of these low CTRs. Google uses only search ads to calculate CTRs for ranking purposes. Special Case: Ads in Gmail and Other Google Properties Did someone say Google isn’t just about search anymore? They now offer Gmail, a fast, innovative, web-based email service that offers virtually unlimited storage. In exchange for a free account, users accept that Google will show ads in the right-hand margin of the interface and sometimes at the top of the page. While some observers fret about privacy issues inherent in customizing ads to the content of email discussions, many users find the ads to be nonintrusive, similar to the familiar AdWords ads near search results. Ads appearing in Gmail are part of the overall content targeting program. Undoubtedly, Google will beef up its reporting and opt-out capabilities in this area, as they’ve done with parked domains and sensitive content. Like the contextual ads discussed previously, ads in Gmail are matched with the content of emails based on a semantic matching technology Google doesn’t disclose. One handy feature of Gmail is the threading of conversations, which makes it easier to refer back to previous emails in a series. I’ve noticed that as a conversation gets longer—as more emails back and forth start to pile up—the matching technology is more precise. Ads seem to get more relevant by the third or fourth email in the exchange. What I’ve said about ads in Gmail also applies to various Google-owned properties, such as YouTube. For the most part, they are simply part of the content targeting program, but there seem to be ongoing mysteries in how Google handles them. How Google handles placement and reporting on your ads showing up on Google-owned properties outside of Google Search seems to change regularly. I expect more disclosure and control to be added in the future. For more on Google’s ever-expanding list of products and services, see Chapter 12. Advanced Uses of Content Targeting: Current Affairs In certain cases, content targeting proves to be more suitable than search targeting. A case study presented by TV cable network Comedy Central at the Search Engine Strategies conferences in 2007 points to an example of an innovative use of content targeting. This media company was trying to promote its television content—for example, The Daily Show with Jon Stewart—in innovative ways. Comedy Central believed that timely current affairs topics would be excellent complements to The Daily Show’s newsy focus. The problem with focusing only on search targeting is that the related keywords might be prohibitively expensive. In addition, such terms might be deemed “irrelevant” by Google’s Quality Score algorithm, making it less likely that CHAPTER 9: Expanding Your Ad Distribution: Opportunities and Pitfalls such ads will be visible next to search results. An exciting workaround was to enable content targeting for such topics. Suddenly, witty ads for The Daily Show appeared on news sites, current affairs forums, and all kinds of hard-to-predict locations around the Web. The exposure was gained for very reasonable CPCs (below 20 cents), and helped The Daily Show build its audience by targeting relevant viewers. If that doesn’t work for you, you can always call Yahoo or Microsoft. Being in second and third place in the search wars, they must try harder. So they might find you more favorable terms for search targeting on current affairs terms. Trademarks as Keywords (“Competitor Words”) There has been ongoing debate about whether it’s legal to use a competitor’s trade name as a keyword in your AdWords account in order to trigger your ad near search results on that keyword. But the status of this tactic is clearer today than it was a couple of years ago. GEICO lost a landmark case, GEICO v. Google, a lawsuit launched to stop competitors’ ads from appearing near search results when users typed queries including the trademarked word GEICO. I see the judgment in this case as a vindication of the principle that AdWords—even when triggered by trademarked words as long as there is no reasonable likelihood of consumer confusion—can foster legitimate forms of comparative advertising similar to those that have long been legal in the United States. Most U.S. trademark experts agree that there is such a thing as fair use, and that whether the use of a trademark in advertising is illegal hinges on whether it causes consumer confusion. Google has, however, lost similar cases in France and is likely to lose them in some other jurisdictions. Eric Goldman, an expert in this field and a member of the law faculty at Santa Clara University, maintains an extensive blog on the subject of marketing and technology law. He is the author of a paper specifically dealing with recent cases like GEICO v. Google and American Blind v. Google. Goldman takes Google’s side, marshaling considerable evidence in the attempt. Formerly the general counsel for Epinions.com, he has also defended WhenU.com, a company that uses pop-up ads to trigger comparative contextual ads. See http://blog.ericgoldman.org for a collection of resources. Several cases since GEICO v. Google have confirmed the general trend; most have been addressed on Goldman’s blog. Remember, you aren’t typically using that competitor’s name in your ad or intending to cause confusion, you’re just using it inside your AdWords account to see to it that your ad appears in the advertising-earmarked area on Google Search should a consumer type in that term. I’ve watched the debate closely and have watched the performance of campaigns that experiment with brand names and company names. For several years, these types of words typically outperformed other keywords, especially in a competitive industry with expensive keywords. Unfortunately, Google has chosen to profit from this situation, and to head off potential controversies, by setting Quality Scores lower on such words. The specifics aren’t disclosed, but the costs of these keywords are on the rise. That’s a shame for enterprising advertisers. In any case, don’t take my word for it. If you’re uncertain about the legality of what you’re doing, seek legal counsel. Also remember that trademark law in North America may be different from laws in Europe and other parts of the world. 259 260 Winning Results with Google AdWords If you try this tactic, you might want to create a separate ad group or campaign for such words (or even two or three for different subtypes) to facilitate tracking. You might use the names of companies, people, and publications, as well as product names, domain names, and so on. Exporting Your Successful AdWords Campaign Some observers have argued that AdWords works like a focus group, giving you valuable insight into your market. Once you’ve tested an AdWords campaign and achieved some success with it, there’s no reason not to put some of that knowledge to work. Google’s Main Competitors What works on Google AdWords is likely to work on at least two other major competing services: Yahoo Search Marketing and Microsoft adCenter. It may or may not work on other such services. The question you’ll need to ask yourself is whether it’s worth the effort. In practice, after Google AdWords and their two largest competitors, it often isn’t worth your time to set up accounts with second-tier services, with the possible exception of Miva, Business.com, and Ask.com. Internationally, you’ll need to research the competitive landscape in different countries. Shopping engines are a separate subject. They also operate on a pay-per-click basis in many cases and are a must for retailers to investigate. Yahoo Search Marketing and Beyond You could take a few different approaches to using the knowledge you’ve gained on AdWords and transferring it to Yahoo Search Marketing, but in recent times the process has become much smoother. With the release of its new search advertising platform, nicknamed Panama, in September 2007, Yahoo took moves towards standardization of formats with Google AdWords. There are some picky differences remaining in allowable title and description lengths, but much of the logic is consistent. The names for the account elements are now identical: Accounts, Campaigns, and Ad Groups. Among other things, this facilitates the migration of an AdWords campaign to a Yahoo campaign. There are still glitches that may occur in transition, however. Google seems to have taken steps to confound the migration process by frequently changing their account export format. You may need to bring in a specialist to clean up a migrated AdWords campaign in the Panama platform. Larger advertisers and agencies may take advantage of Yahoo’s support in migrating campaigns, but this is case by case. A similar situation exists with Microsoft. Another completely new development with Yahoo is that they now rank ads based on a quality index, similar to Google’s Quality Score. However, the formulas remain distinct, and here there may lie optimization opportunities that differ from platform to platform. Like Google, Yahoo offers contextual advertising. Their network and platform are less developed, but we should see improvements coming. CHAPTER 9: Expanding Your Ad Distribution: Opportunities and Pitfalls Industry consolidation seems to mean that we’ll see plenty of good news coming from all of Microsoft, Yahoo, and Google going forward. The process of buying more volume will get easier, and I hold out the hope that the marketplace between buyers and sellers of ads will continue to gain efficiency. ContextWeb is one company that advocates a true “exchange” model (ADSDAQ), where advertisers set “bid” prices and publishers can set their “ask.” Worth a look. Second-Tier Players My clients rarely want to experiment with lesser-known ad providers, and I’m happy to accommodate their indifference. It’s a matter of volume, trust, and focus. In most cases, it just isn’t worth the trouble it might take to investigate the foibles of lesser providers. However, with recurring opportunities to make their cases in conference sessions or on the exhibit floor at search marketing trade shows, it is worth occasionally catching up with companies like Marchex, LookSmart, and Miva. Some of the second-tier players have ripped off so many advertisers with low-quality traffic, it’s hard to assess their credibility. I believe in second chances. Fifth and sixth chances? Maybe not. Google Ad Planner According to Google’s announcement on the Inside AdWords blog, Google Ad Planner is a “research and media planning tool that connects advertisers and publishers.” The idea is, you enter demographic information and sites your audience is associated with, and the Ad Planner tool will return a list of relevant sites your audience is likely to visit, complete with stats like audience size and “fit” with your target. The tool got a big reaction when it was launched. To me, it didn’t seem all that surprising. Then I realized why the process seems so familiar. The logic and visual impact of how you use Google Ad Planner is essentially the same as building a Placement Targeting campaign in Google AdWords. Based on your criteria, the app shows you a list of relevant sites, with data to help you decide whether to place ads on these sites. The key difference is that AdWords Placement Targeting, of course, only provides background audience information on sites that are participants in the Google Content Network (put another way, Google AdSense Publishers). The Ad Planner provides information on pretty much every other significant web property out there, as well. It provides icons and additional information for Google Content Network sites, and leaves that blank for nonparticipating sites. Audience segmentation is a snap with this intuitive tool. To take an example, let’s say I want to get my ad in front of women of a certain age and income. I checked only the snobbiest boxes (the highest incomes, education, and ages 35–64) and this whittled down the available sites in the menu. From there, I’m free to select promising sites and export them to an Excel file if I wish. As you can see in Figure 9-5, from an initially massive universe of U.S.-based websites, I am now down to a smaller universe comprising a total country reach of 2.5%—and still, a long list of sites to consider. From there, you could ask a number of further questions about the sites; questions that are definitively not asked or answered by Google’s tool. Is this site accepting advertising? Is it easy 261 262 Winning Results with Google AdWords FIGURE 9-5 Narrowing down appropriate publisher sites using Google Ad Planner to buy from them? What are the prices, visibility, desirability, user intents on the sites, etc.? As a planning tool, it is a great starting point, but no more. A note on the demographics: these are rough guides to the types of visitors that are seen as coming to the site. They are not accurate counts by any means. It’s merely a question of how a particular site “skews”: Ad Planner, no doubt accurately, reports that potterybarn.com skews female, affluent, etc. But there is nothing to suggest very accurately how much it skews that way, nor is it implied that you could limit advertising impressions to just the selected demographic. All we have learned is: in the event that the site in question indeed turns out to accept your type of advertising, potterybarn.com is probably a pretty decent place to show your banner ad, all else being equal. One current caveat is that the tool only gives the illusion of providing data across many countries. For nearly every country outside the United States, including relatively developed online advertising markets like the United Kingdom and Canada, the demographic information is spotty once you drill down even one level. For these countries, you may be using the tool to find sites related to or similar to a site you specifically enter into the tool, but you’ll be hard pressed to do any comprehensive planning based on demographic segmentation here. CHAPTER 9: Expanding Your Ad Distribution: Opportunities and Pitfalls Some observers believe that Google Ad Planner is shaking up the technology ecosystem much as Google Analytics did. The free tool and open approach to data encroaches on the business models of companies like comScore and Hitwise who typically sell this data for higher prices. But of course, there is not total overlap. This is closer to what Compete.com and Quantcast do: provide background information on website visitation stats, much of it for free. These services use a variety of data sources (including, perhaps, panel-based measurement, installed toolbars or tracking code, user data sold to them by ISPs, etc.) to arrive at educated guesstimates about the audience composition of websites. Before long, Google’s product is likely to outshine these services in terms of accuracy. From a variety of angles, Google simply has the most data about users and websites. Beware of bias: should you trust the seller of ads to be your media planning “advisor”? Well, no. But here, Google isn’t selling but a small portion of the total reach addressed by Google Ad Planner. So it looks like the situation was similar to what happened with Analytics and Conversion Optimizer, among other products. Google saw a hole in marketers’ and advertisers’ arsenals and wished to fill that need, in part to position Google as the ultimate service provider in online advertising. Google became increasingly impatient as various existing providers in the space kept advanced functionality out of the reach of all but a handful of deep-pocketed advertisers. The “cloud computing mentality” pervades Google’s thinking with regard to every new field the company considers breaking into: the analogy will always be with the IT services and bandwidth costs that have plummeted in the past decade, allowing innovative businesses of all sizes to thrive in the information economy. Instead of: “wow, that data is worth a bundle, you should charge a lot for it,” as usual, Google’s thought is: “would providing this tool to businesses reduce friction in the marketplace?” In general, Google wants advertisers and publishers to thrive, and they believe that more transparency and bridging activities will facilitate that. The online media planning and online measurement industries have been moving too slowly in the direction of transparency and ease of use. This is a step in the right direction.1 Offline Marketing The strange thing about search engine marketing is that it’s a bit like “anti-marketing.” You don’t bother people—they find you when they’re ready to look into a subject. Finding customers this way can help teach you a lot about what those customers actually want. Listen to them, and keep giving it to them, or develop new products and services to fill those needs. So under the heading of “expanding your marketing offline,” I’m going to tread a bit cautiously. Seth Godin, in two recent books (Purple Cow and Free Prize Inside), trashes most modern marketing, insisting that the product or service offering itself is what wins over savvy consumers. Marketing is ideally built into the product from day one. The idea that marketing professionals can simply be brought in at will to “put lipstick on pigs” and create demand for a new product is a myth. If your market is growing at a solid pace, or if it’s really taking off, chances are that some underlying forces (the merits of the product, your personal reputation, changing market trends) are at work, not just the fact that you decided to shout louder. 263 264 Winning Results with Google AdWords Telling you how offline marketing works is outside the scope of this book in any case. But one (admittedly radical) approach to the subject of advertising and marketing offline is, do less of it, and plow the proceeds back into innovative products and features that will delight customers and make you stand out in the marketplace. Amazon.com, for example, slashed their television ad budget so they could offer free shipping. Great move. Your company might be better off investing in design software, customer relationship management solutions, or other forms of automation that boost productivity than it would be to invest in ads that create top-line growth of new customers who enter into the same old inefficient relationship with you. You must weigh your priorities. For the less radically inclined, you can certainly transfer some AdWords learning over to your offline campaigns, but you’ll probably need to retest. If certain ad text or sales copy on landing pages works well for you online, it can’t hurt to try it offline. But these things don’t always translate. In a perfect world, the direct mail people would talk over their findings with the online marketers. By sharing insights, marketing projects might proceed more quickly through the early learning phases, with fewer failed experiments at the outset. Time will tell if the integration of direct marketing agencies and online marketing consultancies will lead to a productive synthesis that can help more companies reach their marketing goals more quickly. Above all, marketers need to be open to unorthodox ways of persuading consumers to try their products and services. Infomercials, for example, caught the marketing world by surprise just as AdWords is now doing. One unsolicited endorsement from a leading author, athlete, or celebrity could mean more to your business than a year’s worth of magazine ads. Don’t assume that because large brand advertisers are spending heavily on traditional television, billboard, or radio ads that you ought to do the same. Look before you leap. Many advertisers try to generate traffic from a wide range of sources, making for an unwieldy and chaotic marketing strategy. I’ve tried to argue here that you may be best served by taking simple steps to increase profitable click volume from your existing AdWords campaign before troubling yourself with a host of other online marketing experiments. Endnote 1. A version of this section appeared as my column in July 2008 at Search Engine Land. Repurposed with permission. Thanks to Danny Sullivan. Part IV Winning the AdWords Game: Advanced Issues This page intentionally left blank Chapter 10 Measuring Success: A “What’s Changed” Report T he plain-language name for web analytics these days, “measuring success,” is apt. It’s as simple as it sounds, and yet not as easy as it sounds. Your boss’s or client’s reaction to a campaign depends heavily on defining the terms of campaign success, and then attempting to use tools and techniques that will actually put accurate and actionable reports into your hands. Sometimes it’s relatively straightforward, sometimes not. If you’re the business owner, theoretically it’s a bit easier. You know that the bottom line is what matters the most. But many arbitrary decisions, judgment calls, and unstated strategies come into play in attempting to attribute a value to paid search clicks—so much so that even after this chapter is completed, you’ll need to continue your learning elsewhere on an ongoing basis. Consider this chapter a rather advanced introduction to the field; also, apologies in advance if I don’t make it into a purely technical manual. The ins and outs of getting Google Analytics properly linked to your AdWords account, of using a particularly persnickety goal customization in Omniture, and much more, are matters that give me as much of a headache as anyone. Sometimes I plow through them myself, sometimes we can find a vendor or a customer support person to help. Enough said on that.1 This chapter is divided into roughly three parts. To kick things off, I provide a summary approach to the very basic steps anyone should take to get their house in order to track AdWords clicks “after the click,” condensed in the extreme from the first edition of this book. This, in essence, is the “What Hasn’t Changed” report. From there I go on to talk about some of the eye-opening new developments that have taken hold in our field, making it more likely that any given advertiser—including you—will be implementing basic tracking more seamlessly with less additional cost, as well as exploring more sophisticated add-ons to your routine. The name of this chapter pays homage to (or if you prefer, is shamelessly borrowed from) analytics vendor ClickTracks. One of their nice features has always been a “What’s Changed” report that helps you discover action items more quickly than you would otherwise, poring through the large amount of available data in monthly statistical reports. Likewise, I assume many of you don’t need a rehash of the basics, but would like a bit of 268 Winning Results with Google AdWords a roadmap to what is truly significant and what’s actually different today as compared with the state of the art in 2005. And finally, I briefly outline a long list of foibles, errors, quirks, and self-indulgences that seem to hinder many marketers as much today as they did in 2005. This part looks at what hasn’t changed, but the message is: I hope it will. What to Measure, How to Manage: Skinny Summary Not to belabor the obvious, but there are a few key approaches to measuring success that don’t vary much from year to year. I reported on most of these in the last edition. Here is a concise summary of how smart advertisers are making use of key metrics to improve their bottom-line profitability with paid search. To succeed, you must understand your business objectives, know what key campaign metrics you need to pay attention to, and decide what tracking methods and tools are best for you. There is a plethora of web analytics and tracking systems available in the market today, including the free conversion tracking tools within AdWords, and the more extensive Google Analytics, which I cover in this chapter. Most of the popular analytics tools require you to place some JavaScript tracking codes on your website pages, although there are still some pure weblog analyzers that don’t. How Tracking Works No matter what tracking system you decide to use, they all require two essential bits of information in order to track post-click actions. First, they need to know where the clicks came from, and second, they need to know what action on your site you want to track. The information about where the clicks originated involves appending some tracking codes to your destination landing-page URL, a process known as tagging. For example, if your landing page is http://www.woolscarves.com, the same landing-page URL tagged with tracking codes might look something like this: www.woolscarves.com?source=google&kw=alpaca_scarves&ad=summer The parameters, source, kw, and ad, inform your analytics tool how to associate clicks from your campaigns with visitors on your website. These may be arbitrary parameters you choose, or parameters dictated by your tracking software. Or, the tracking software might even auto-tag your landing URLs by plugging into Google AdWords through the API. It varies. Just choose a method you’re comfortable you understand. If Google Analytics is correctly linked to your AdWords campaign, theoretically you should not need to do anything specific to tag your landing URLs. In the preceding example, we are tracking by network, keyword, and ad, but you can track at other levels as well, depending on your analytics tool. I generally like to start by tracking at the ad and ad group level and then selectively track a handful of specific keywords, typically those that are most popular or most expensive. CHAPTER 10: Measuring Success: A “What’s Changed” Report The second part of tracking post-click activity is to identify the pages on your site that you want to track by placing a snippet of JavaScript code on those pages. This script is provided by your tracking tool and is usually placed on the “thank you” page after a visitor completes a download, completes a form, or orders from your online shopping cart. When a visitor from a paid search campaign gets to one of these pages, the JavaScript communicates that fact to your analytics package, essentially completing the feedback loop so that you can associate a click to an eventual conversion activity. The World Isn’t Perfect, and Neither Is Web Analytics In earlier chapters, I told you that one of the biggest advantages of paid search advertising is that you can track and measure how every part of your campaign is performing, from the price you pay for clicks to the effectiveness of your ad messages and your keywords. Compared with other forms of advertising, this is absolutely true, but that doesn’t mean that it is perfect. Just as Heisenberg famously described uncertainty at the subatomic level, it is also true that the more granularly you examine your PPC campaign metrics, the less certain your data become. You need to avoid the delusion that you can track every sale back to the specific keyword that triggered a specific ad in a specific ad position at a certain time on a certain day of week. Here are some situations, for example, that make tracking imperfect. Let’s say a person shops online for wool scarves on her home computer, clicks on a paid search ad, and finds the perfect website. On her lunch hour the next day, she types the website URL into her browser and buys a scarf. Paid search drove the sale, but it’s impossible to track this sale back to a PPC campaign. The same could be true if she had picked up the phone and ordered. And even if she had ordered from her home computer, it is possible that she searched on a variety of terms, and clicked on several ads, before reaching your offer. Google Analytics and Google Conversion Tracker, discussed later in this chapter, assign credit only to the last ad and the last keyword clicked. In spite of, or perhaps because of, these limitations, it is important to realize that insights gleaned from your tracking data accumulate over time and give you a relative idea of how your campaigns are doing. It is the accumulated trends that give you the actionable data you need. What You Need to Track: Metrics to Consider When in doubt, simplify. Many of my clients do very well with a simple measuring stick such as cost per acquisition (CPA) or cost per order. Even though these may not take into account the dollar amounts spent by the customer, they can be reliable measures with a history in any given line of business. Comparing these metrics among various ad sources, keywords, and ads will give you a sense of your relative performance trends. These metrics are the ones that tie together the performance of your campaigns (clicks, costs) with your business objectives. Here are a few of the metrics that will get you (or any analyst or manager) to the heart of the matter: ■ Rough ROI or ROAS No one is going to be able to prove exactly how well an advertisement paid off, and profit margins can vary from sale to sale and from product to product, making an ROI a challenge to calculate. What many companies do, instead, is 269 270 Winning Results with Google AdWords use a “dollars in divided by dollars out” (that is, revenue divided by ad spend) calculation as a working measure of the return on ad spend (ROAS). As with all metrics like this, it is useful as a relative, rather than absolute, yardstick. ■ Cost per acquisition CPA is a generic metric that could cover any type of conversion activity, such as newsletter signups, new customers, or any other goal you can measure, including offline actions such as phone calls or visitors to your open house. Even if this metric can’t be calculated by your online tracking tools, it is a very easy hand calculation to make by simply taking your ad spend and dividing it by the number of actions registered that you can attribute to your Google ad groups or campaigns. If you spent $420 and got 10 newsletter signups, your CPA is $42. ■ Cost per order Retailers will probably want to keep track of the cost per order related to any particular referral source, keyword, ad, or ad group, since different product lines have different gross profit margins, some high and some very tight. If order values vary wildly, you will want to use this measure in combination with ROAS. ■ Revenue per click Another way to measure ROI, revenue per click, is again a relative measure that can be easily compared to determine whether a bid is too high or too low for a given keyword or ad group. Once you’ve put your tracking systems in place, and you are collecting all this data, don’t be afraid to make changes based on your data. Unfortunately, it’s not always easy to interpret your metrics, and the differences in performance may not be big enough to make big changes, but the simple act of tracking these metrics will give you an idea of when you have ads, ad groups, and keywords that are underperforming or not performing at all. Measuring Success: “What’s Changed” Report In no particular order, here are some of the developments in web analytics and AdWords performance that you should take account of. Analytics as an “Industry” Has Exploded Even in the past three years, the number of individuals who consider web analytics to constitute all or much of their job description has exploded. Jim Sterne’s E-Metrics Summit conferences and related events have grown by leaps and bounds, and membership in the Web Analytics Association has mushroomed. How much this does or should affect you, however, is worth questioning. Much as you may not have the time, resources, or vendor selection smarts to hire a usability consultant or search engine optimization specialist, an analytics focus can only run so deep in many companies. If the effort is not tightly integrated with what you do, then it may be misapplied anyway. So your task, likely, will be to increase your analytical savvy without making a whole career out of it. The name of Avinash Kaushik’s blog, Occam’s Razor, is promising. Kaushik is one of the leading web analytics experts today. Formerly with Intuit and now a consultant to Google CHAPTER 10: Measuring Success: A “What’s Changed” Report and sought-after global speaker, Kaushik backs his assertions with a meaty 443-page book entitled Web Analytics: An Hour a Day (Wiley, 2007). I’ve been delighted and inspired by Kaushik’s work—but Occam’s razor, it is not! The book contains just four pages on measuring “PPC effectiveness,” including a single mention of ROAS (just mentioning that it is a metric that exists), and, well, 439 pages about everything else. We’re left aching for a lot more of that paid search part. Like Jim Sterne, Kaushik does yeoman’s work encouraging large, slow organizations to be crisper and nimbler in their use of web user data. But if your job requires you to be radically crisp and nimble, you’ll want to set much of Kaushik’s excellent book aside as a comprehensive desk reference to be accessed as needed. Us AdWords folks need to really cut to the chase; we need Occam’s Razor, Mach IV edition. Urchin Rules the High Seas Once upon a time, if we are to think of website statistics vendors like ships, a great diversity of crafts roamed the analytical waters. There were big boats, little boats, ugly boats, and pretty ones. Expensive, cheap, and sometimes free ones. Some of the “boats” required you to install a JavaScript tag across your site, or on conversion pages, to record visitor activity. Others used a “captain’s logfile,” analyzing it in retrospect to create more accurate, but sometimes more cumbersome, reporting. The best ones, like SeaTrends, used both methods. The big boats sometimes got into skirmishes, and the odd one went down, but relative peace prevailed. Rather than this competition providing a stimulus for innovation, just as often, it led to confusion among investors and passengers. These folks, collectively known as “customers,” found they were wasting too much time at boat shows debating the technical merits of the various crafts. The act of taking a piece of many different boating companies, and riding in a lot of different kinds of boats, was getting exhausting for the marketplace. The Boat Captains’ Union didn’t much like it, either. While it was a bit of a barrier to entry to outsiders, the effort required to master the controls and quirks of so many different types of boats was more trouble than it was worth. All collectively sighed: “If only there was a Standard. One Big Boat Type that Google Analytics vs. Urchin For ease of navigating this section, which is intended to be a lighthearted look at the difficult problem of web analytics vendor selection, keep the following in mind. Google acquired an analytics software company called Urchin, and made the basic product free, renaming it Google Analytics. You can use Google Analytics in a relatively seamless way to measure results from your Google AdWords campaigns. Today, the offering is far from “basic”—in other words, it’s amazingly powerful for a free product. Still, Google has continued to sell more advanced versions of Urchin, and this product will cost you money. It comes with more features and more dedicated customer support, and therefore competes with full-featured analytics packages such as Omniture, WebTrends, and Coremetrics. The majority of my clients use Google Analytics, and many use more than one analytics package at the same time. 271 272 Winning Results with Google AdWords we could all ride on, invest in, and pilot, and just a few other boats on the side for playboys and hobbyists that we could pretty much ignore.” None thought the day would come anytime soon. Little did the captains of all these ships know that the captain of one unassuming midsized vessel (Urchin was its name) full of engineers—a captain who would later become known as The Urchinator2—was concocting a diabolical plan. The Urchinator observed that one seafaring organization, SeaTrends, had grown by leaps and bounds by making side deals with the port owners (also known as “web hosting providers”). Port owners would recommend SeaTrends’ boats to anyone who would listen. They began acting as distributors, too. “Aha!” thought The Urchinator. “The devil is in the distribution!” And so The Urchinator concocted his diabolical plan. He would sell his midsized company to the granddaddy distributor of them all: Google and their 1,000,000 advertisers, and 20,000,000 attentive owners of businesses vying for better online results. The plan with Google would give away the now-Universal, now-Massive Urchin Boat for free, as a gift, to advertisers! (Like a Trojan horse. The Urchinator, and Googlers, study their classics.) Google would wish to subsidize the giveaway in order to funnel even more advertisers into their system, and to gain access to a goldmine of data. Google would then invest in massive computing power to keep the product free and useful, and would then add features that The Urchinator highly approved of, to make the product even better without charging a dime. And so the deal was done. Google Analytics now rules the high seas. And while some playboys and hobbyists may have their reasons to invest in the other expensive toys and cheaper cool hacks that are available, everyone can agree on the convenience of using one Big Boat Type, and linking it up with an AdWords account. A few holdouts, wary of the Trojan connotation, choose to look this gift horse in the mouth. Seeing his evil grin and diamond-encrusted bling, they vow to “never!” acquiesce to the Google Analytics standard. But most are hypnotized by the grin, and sign right up, linking Google Analytics to their AdWords accounts. Another standard is born. Horses! Boats! Captains! The remainder of this chapter will be mercifully much lighter on the tortured metaphors and bumbling allegories. And no puns, either. I hope you took your allegory medication before reading this section. So sign up for Google Analytics if you’re brave (or is it cowardly?), and follow the help files to ensure proper installation of code on every page of your site. Setting Up Goals in Google Analytics At least a dozen current, high-quality books are available just about Google Analytics and web analytics in general. Analytics offers a wealth of reports about user behavior and can be customized in many ways. But for AdWords purposes, you’ll be looking to tune it so you can read and react to the performance of visitors that is tied back to specific keywords or ads. To lay the foundation for this, you can establish Goals in Analytics. As with Google Conversion Tracker, you’re essentially going to be telling the software what a “conversion” means for your business. Like many computing tasks, this is quite technical and literal. To tell Analytics about your email signup goal page, you’ll be entering the URL of the thank-you page for successful email signups. A slightly more technical variation on this theme will be required for you to establish an e-commerce checkout completion.