Nội dung

CHAPTER 3: First Principles for Reaching Customers Through AdWords Google no doubt has high hopes to grow this back to 50% or more, as the definition of “content” expands into a variety of media that might be amenable to booking placements through an auction platform like AdWords. This program places your AdWords ads on pages of publishers’ websites, ranging from large publishers like CNET, the New York Times, and About.com to smaller content sites published by small independent publishers of high-quality content—and yes, even on blogs and junky MySpace pages. Due to the improving quality of the content network, and its continued growth, it offers an opportunity for advertisers in terms of both quality exposure and additional reach. Many advertisers put considerable effort into researching and building their accounts, and into ongoing bidding strategy and analysis of results, so the added reach is always a good way to make that effort worthwhile. You should be aware, however, that content targeting is quite different from search-based advertising. It should be treated more like banner advertising, even though the ad displays are triggered by the keywords in advertisers’ Google AdWords accounts. Speaking of banners, the choices available in the content program are increasingly impressive (if bewildering). The content-targeting program started with plain text ads, but now allows a variety of sizes and shapes of banners, animated banners, video pre-roll advertising, and more. You may hear the term AdSense used interchangeably with “content targeting” (or a term that others have used, “contextual advertising”). AdSense is the name of the interface that publishers use to place the Google AdWords ads on their sites to receive revenues from Google (ultimately from you, the advertiser) when users click the ads. Figures 3-16 and 3-17 provide examples of the different ad formats used on content sites. Pricing for content targeting is based on proprietary semantic matching technology developed in-house at Google that actually determines which ads to show on the fly as a page loads. The key criteria are how closely the meaning of the content on a page matches the keywords you’re bidding on in any given ad group in your AdWords account. We can also presume that your reach is heavily influenced by your maximum bid. Bid high enough on content, and your ad will show up on far more pages—although relevance will suffer. CTRs for content targeting are typically much lower than they are on search ads; however, these CTRs are not factored into the CTR that determines your ad rank score for the purposes of ranking you on the page. Don’t worry too much about these low CTRs regardless of how bad it makes your stats look. In spite of the lack of negative consequences attached to these low content-targeting CTRs, some advertisers will see cause for worry when they attempt to interpret their stats for periods when content targeting is turned on. In statistical summaries for given ads, periods of content-targeting usage will frequently drag down the aggregate CTR number. Thus the strong performance of an account may not be immediately evident without scrutinizing the data more closely. Also, turning content targeting on and off can make comparing the CTR performance of ads difficult. Newer ads that were showing during periods of heavy content-targeting use are difficult to compare head-to-head over, say, a month-long period, when pitted against ads that were showing with content targeting switched off (or simply left on for a shorter duration). Until Google improves this reporting, you can be easily misled about ad performance unless the ads you’re comparing have been running with the same settings applied to all. Keep this in mind when testing ads. 111 112 Winning Results with Google AdWords FIGURE 3-16 A typical AdSense publisher, HowStuffWorks.com, displaying text-based ads in the left margin. These ads are served by Google AdWords. Don’t mistakenly stop an ad that may be doing well but appears to be a slow performer due to content targeting. Ads near content perform differently than search ads, because user behavior and expectations are usually different when they’re casually reading articles rather than actively searching. Thus the economic worth of content ads to advertisers may be lower than what we see from ads placed near Google Search results. Since the inception of content targeting, Google has maintained that conversion rates on content ads are comparable to those on search ads, even if CTRs may be lower, so the value should be about the same. In April 2004 Google introduced something called “enhanced smart pricing” for content targeting. Many advertisers had asked if they could bid separately on the content-targeted ads or even create separate ad copy for content targeting. Although this smart pricing stopped short of those demands, it used a formula to adjust click prices based on their expected value to advertisers. This expected value is based on information Google may have about the probabilities that certain types of pages (say, a page containing reviews of digital cameras, as opposed to a feature-length CHAPTER 3: FIGURE 3-17 First Principles for Reaching Customers Through AdWords Google AdWords ads for golf-related products show up in a text box in the middle of this article on the About.com Guide to Golf. article about the history of photography) have of converting to a sale for the advertiser. Google says it uses “all possible pieces of information” to determine the expected value. Following this advance, Google later did release something called content bidding. This is absolutely vital. In the Edit Campaign Settings interface (Figure 3-14), if you don’t disable content targeting entirely, you’ll at least want to enable content bidding, by clicking the check box to “Set a Separate Bid for Content Network Impressions.” You then have the choice of adding separate content bids to your account’s ad groups now or later. Content bidding only takes place at the ad group level. An example would be a keyword group full of terms like “forex trading.” These are valuable terms when found through a Google Search, so assume I bid $3.00 on most of them individually, and leave a default bid on of $2.00 for the ad group for any other keywords I don’t bid on specifically. I know that my ad does perform somewhat decently in the content-targeting program, but the ROI is sharply lower. I don’t want to give up the sales volume; I just want to bid 70 cents on content clicks, to even out the outcomes. So I do, using content bidding. 113 114 Winning Results with Google AdWords Content targeting is a different animal from search targeting. If you’re unsure, opt out of it for the time being by leaving the Content Network option unchecked at the campaign level. As you become more experienced, you may decide to try experimenting with it, since it can significantly expand the reach of your existing campaign. The power of large networks like Google’s is that they are certainly far easier to enable and test than is possible under the traditional media-buying methods of negotiating ad buys with individual websites or traditional ad brokers. Google actually now has multiple “flavors” of content targeting. Two options are most prominent in the interface: keyword-based content targeting (what I often call classic content targeting, because it was a key product innovation at Google) and a newer program, site-based placement targeting (formerly called site targeting). Placement targeting is really a separate program in itself. To keep us moving here, I’ll discuss placement targeting and other Google network initiatives in more depth in Chapter 9. Country and Language Many of you will be focusing most of your efforts on the original and largest AdWords market, the United States, in English exclusively. Unfortunately, running campaigns to attract viewers who are using Google set to display other languages is not an automated process. For each language, you would have to run a separate campaign, choose different keywords, and write the separate ads. By and large, you’ll find that displaying ads to all countries is a money-losing proposition. Your mileage may vary, but not all English-speaking markets are equally responsive from an economic standpoint. More importantly, of course, your company might only ship its products or perform its services in the United States, or the United States and Canada. Unless you’re prepared to do business in other countries and you know your product is marketable in them, you might want to take a cautious approach and go with the United States only, or United States plus Canada. For those who want to branch out a bit further, a typical approach seems to be to add the UK (one of the largest AdWords markets), and perhaps Australia and New Zealand, to the mix. For business-to-business and professionals as well as midsized to large companies (especially those with a strong international base), it may make sense to run ads in English in a variety of target countries in the hopes of influencing decision makers in those markets. As a general rule, though, such efforts can be a waste of money, and my instinct (honed by client anecdotes from the past) is to be cautious. Chapter 4 Setting Up Ad Groups F rom what I’ve observed, at least half of all new AdWords advertisers make the same set of predictable tactical mistakes. To help you avoid these, let me review some of the most common errors. There seem to be a few common patterns here. Most revolve around a couple of tendencies: first, the desire to create an enormous list of keywords at the beginning rather than a smaller “beginner set” of keywords that fit logically into groups; and second, an interrelated belief that with the right amount of effort in the planning (prelaunch) phase, the campaign can explode out of the starting gate, generating huge numbers of customers right away. Small problem with the “explode out of the gate” mentality: Google has more than 500,000 advertisers. Lots of them already exploded out of the starting gate, and you’ll be competing with them. You’ll need to ease into this process at first and then build on your early discoveries. This process rewards smart “guerrilla” advertisers who can learn from feedback, not just those with a bigger marketing bazooka. There are some historical reasons why many paid search advertisers seem bent on doing things in a certain way (the way that I consider to be “wrong” for AdWords). Advertisers who had experience with Overture became accustomed to the idea of large numbers of keywords. One reason for this was that Overture didn’t offer broad matching options in the past; so unless your keyword or phrase matched the user’s query exactly, you didn’t show up. There is nothing strictly wrong with using every possible word combination of hundreds of words, culminating in a file of five thousand or more keywords. But the reason for doing it was initially because you couldn’t capture enough search volume without wild card–type matching options. Those who overdo it on the keyword generation front today are banking heavily on the value of infrequently searched keywords, sometimes to the exclusion of balanced priorities, campaign organization, and thoroughness with more important keywords. Keyword research tools come in various shapes and sizes. There are some “gray market” aids that will even help you determine what competitors are bidding on. To come up with extra 116 Winning Results with Google AdWords keywords to add to a well-functioning account, to test their effectiveness, is a great idea. But don’t dump them all in at once. Another historical reason for large keyword files was that Overture’s early interface was a first-generation utility with limitations in the usability department. The cumbersome process of dumping large files of keywords into the account without any really convenient or intuitive way of then managing or editing them seemed worthwhile to early Overture advertisers, who felt like they were getting in on the ground floor of something exciting. It certainly delighted the makers of third-party account management software. I never much cared for it. When Google AdWords came along, it gave advertisers better tools for keeping everything straight—most of all, an intuitive way of grouping keywords. In any case, the result of all that history is that an orthodoxy sprang up whereby marketers felt they could impress one another by sending each other gigantic Excel files of keywords. Let’s take some time to explore ad groups, then, which I consider to be the core of Google AdWords. Why Grouping Keywords Makes So Much Sense When my colleagues and I use software to track what users are doing after they click through to a client’s website, we don’t overanalyze the performance of individual keywords, especially those that generate low volumes of searches. Because the infrequently searched words can’t give you statistically significant feedback on their own, we often prefer to track no finer than the “ad group and specific ads within those groups” level, because, if the groups are designed logically, tracking the results by group actually provides highly actionable and meaningful data. Sometimes we track everything right down to the return on investment on individual keywords and phrases, but this is not always necessary or even beneficial. Keeping the data well organized seems to oversimplify things, but you have to “apparently oversimplify” AdWords accounts, because your ability to correctly influence events is actually tied to a lot more complexity than you are likely to be able to handle. Machines can do some of it, but you need to free up as much of your time as possible for “softer” analytical work that explores the full range of potential responses to the data you’re seeing. Think about the analogy of an American football playbook with 500 or 1,000 plays in it, grouped according to different types and situations. The quarterback and the coaching staff need to master and memorize these plays so that they can deploy them correctly at the right times. These plays are difficult to digest even for many quarterbacks—hence the tiny crib notes you see written on many quarterbacks’ wrist guards. With the play clock ticking, it wouldn’t help that quarterback at all to receive a giant Excel file of new plays, or an even larger file of past and probable outcomes for 10,000 other plays. Not only must coach and quarterback choose among a relatively small universe of courses of action in calling the next play, but once the quarterback steps up to the line of scrimmage, he must have the ability to call an “audible” (a new play based on the defensive formation he sees). The number of possible audibles is typically tiny—there might only be two or three alternative plays to choose from. I don’t think the analogy is so far-fetched. To reduce confusion, reduce the number of potential decisions you need to make. Then make those decisions with full consideration and as often as you can feasibly make them to improve your performance. CHAPTER 4: Setting Up Ad Groups Granted, AdWords isn’t a football and you don’t have to physically throw it while avoiding human tacklers, so as you get more advanced, you’ll want to explore ways of automating decisions where this makes sense. For now, thinking about doing it all manually will help you understand the underlying principles. Ad groups give us that manageability we’re looking for. I tend to believe that each group of keywords expresses an idea of something a user is searching for. That might be a big idea or a very narrowly conceived idea. The idea could require only one keyphrase to express (let’s say the exact match for goat cheese), or it could require 250 phrases covering a long list of low-volume but highly targeted industry jargon words. So, when someone asks me how many keywords is a lot, I usually avoid that question because I believe campaigns need to be thought of in terms of ad groups. I sometimes think in terms of this analogy: putting just a few of the most obvious keywords in a few groups is OK at first, because you’ll find the process of expanding to more words within those groups quite natural. They’ll almost multiply like bacteria (icky, but that’s kind of how it works). Actually, you’ll be using your own brain and keyword suggestion tools, but the basic idea is that ad groups often start off small and grow larger over time. This can be an intuitive process, because you’ll also give names to those groups within your account; so, you’ll be able to glance at them quickly and say something like, “I see the ‘Last Minute Travel’ group is generating a higher than usual number of clicks today,” or, “The ‘San Jose Sharks apparel’ group is generating a low CTR lately; better figure out why.” For my money, that’s better than poring over huge files of keyword-specific data, because the intuitiveness of groups with sensible names allows you to read and react steadily to changing conditions. If you structure your data analysis task so that it’s more daunting than that, you might find yourself putting it off for weeks and months, and that’ll cost you. Think of this as a kind of sorting or filing. The database-driven nature of the AdWords application is actually not too far different from the idea of a directory, with multiple levels in a logical progression. As librarians and search technology experts sometimes say, categorized directories (think of Yahoo or the Open Directory, or anything with categories and subcategories) possess an ontology. In other words, a professional categorization team needs to create a tree that breaks the world down into different levels of meaning. Your account won’t be that comprehensive, but I hope the analogy helps you to understand that your job in creating a little “meaning tree” for your account will help you to do a better job of sorting out search users who see your ad after they’ve expressed meanings of their own by typing a query into Google Search. This structure will also make the campaign easier to make sense of down the road. Account Campaign Ad Group Ad Group Campaign Ad Group Ad Group 117 118 Winning Results with Google AdWords Ad groups express a thing (the “soup bowl group,” for example) or an idea (“agricultural pesticides litigation” and 40 other ways to say that). Your advertising copy (or multiple ads) are tied to the keywords in that group. Different groups, different ads. Sure, you could use the same ad all the time, but it’s best to write different ones, as I’ll show later. Basically, whatever ad (or ads) you enter for, say, Ad Group #3 (or the “Tile Flooring Group”) will show up whenever a user’s query matches one of the phrases in that group, assuming your campaign is active. That ad won’t show up for your other ad groups unless you specifically create the same ad in those groups, as well. The AdWords interface allows you to control exactly which searchers are seeing which ads. Once you’ve got a few phrases that all express something related to an idea or thing, you’re on your way with your first ad group. It should be easy to set up several groups in no time as long as you aren’t fussing with huge keyword lists. You can edit everything later as much as you like. Not only will you write separate ads tailored for each group, you’ll notice that you’ll be bidding separately on each group. All the words and phrases in an ad group are tied to a global maximum bid. That makes it convenient to change the bid for the whole group, although there is also an optional feature called powerposting that allows you to set individual bids on keywords or phrases (more about that in Chapter 6). This advice, then, ties into advice given later in the book about how to write winning ad copy. There should be less mystique about how to write successful ads once you understand that your ads’ performance will improve almost automatically by dint of the fact that you’ve written a variety of tailored ads that closely match or reflect the ideas or exact phrases in each ad group. The question won’t be only “which ad works the best” across the board, but also, in many cases, “which ads work the best with which groups of keywords.” You’ll want multiple ad groups for two key reasons, then. First, ad groups offer the convenience of tying your maximum bid (the highest you’re willing to pay for a click) to all the keyphrases in a group, to save you the trouble of bidding individually on every keyword. Most of us use a mix of keyword-specific bidding and groupwide bidding. Figures 4-1 and 4-2 show two key views inside the Google AdWords interface: the summary view within a campaign showing a list of ad groups, and a fairly typical example of an ad group. The ad group shown in Figure 4-2 has a maximum bid of 80 cents that applies to all the phrases in that group, and as you can see, the 2 phrases in the group resemble one another. (Of course, 2 is an unusually small number of phrases to put in a group. It could just as easily be 5, 20, or 50, but this suffices for illustration purposes.) A single ad applies to this group of phrases, although this advertiser had previously tested multiple ads with this group to see which one performed the best. He has also made his ad timely, telling readers that the site contains specific information about planting tips for the month of June (not a common month in which to plant), which likely conveys freshness and expertise. This may be part of the explanation for the robust 10% clickthrough rate on this ad. In this reporting summary, various performance data, including CTR, are broken down by keyphrase. Note that this advertiser is using the classic approach to bidding, using the global bid for the group so that all of these keywords have the same maximum bid. Many advertisers now make finer adjustments, adding specific bids to keywords within groups, which is often necessary to adjust bids to market demand. Still, there is a certain tidiness to the classic way of doing it. CHAPTER 4: FIGURE 4-1 Setting Up Ad Groups A list of various ad groups within this advertiser’s “Campaign #7” A second, and not unimportant, reason that organizing around ad groups is helpful is to ensure that each group of keyphrases linked to any given idea is linked to an ad (or multiple ads) that closely targets users searching for whatever that idea or thing might be. The closeness of the match to users’ interests, and those users’ feelings of being catered to (basically, extreme relevancy in search), seems to improve campaign performance. If Google is giving us the ability to micro-target users with an offer that might really appeal to them based on what they’re typing into the search engine, should we run a generic campaign that acts more like the traditional run-of-site banner ads? No! Groups remind you to target your ads more tightly to the user’s query. As I’ll explain in more detail in Chapter 8, within an ad group you can run multiple ads at the same time. (Some call this split-testing or A/B testing.) So even within a tightly focused area, you can still experiment with different ways of catching searchers’ attention to find out what works best, and the independent impact of variations in ad title and ad copy will be measured accurately. 119 120 Winning Results with Google AdWords FIGURE 4-2 A summary of AdWords campaign data for a week in the life of “Ad Group #5 Wildflower Seeds” Google’s Strange Advice on Ad Group Size Oddly, Google staff do (verbally) tend to give you strong advice to limit the number of keywords per ad group. Why? The unspoken reason is that very low-volume keywords muck up the system and give Google data overload they can do without, given that it costs you nothing to add 100,000 of them to your account. Google won’t make much more revenue out of all that keyword inventory, as long as their various matching options are causing ads to show up on a lot of queries at good average CPCs as advertisers bid things up. So Google, from a purely selfish standpoint, isn’t going to applaud huge keyword lists. The overt reason, according to comments made by various Googlers, is that the same keywords are being interpreted by the content targeting algorithm that attempts to match the overall semantic meaning of your ad group with the overall meaning of text on a web page. More than 10–15 keywords starts to dilute the effectiveness of that matching, supposedly. That leads me to wonder whether the tail isn’t wagging the dog. Should our ad group sizes be dependent on the foibles of a content program that came out after the search ads program, layered on top of that program in an idiosyncratic way by Google product managers? So on this front, you probably should march to the CHAPTER 4: Setting Up Ad Groups beat of your own drummer, and figure out other ways of maximizing your content performance, such as setting up separate campaigns or using placement targeting (discussed in Chapter 9). Ad group sizes shouldn’t be unwieldy, but limiting a group to 10–15 words is unrealistic in many cases. While there is ultimately no single rule of thumb for how many keywords in a group is unwieldy, you will soon get a feel for this from your actual campaign data. Certainly, over 100 keywords is probably too high. If the bottom 75% of your keywords combined doesn’t get a single click in a month, that might be a sign you are overdoing it! But that is not the only way to have too many keywords in a group: the other way is to have too many keywords with different meanings, with different user intents, jumbled together. Again, then, rather than merely counting, just do what makes sense from the perspective of users seeing relevant ads, and from the perspective of ease of reporting and interpretation long term. In some cases it might make sense to place only one or two keywords in an ad group, if these are particularly high volume. This might sharpen your focus in reporting on, and testing ads for, your main drivers. Getting Very Granular with Groups How fine-grained you want to make your group structure is up to you. A large retailer with 20 ad groups in a textiles-related campaign might do better if they subdivide those into 80 morespecific groups. When called upon to improve my clients’ campaign performance, one tactic I’ll try is to take a medium-grained ad group and break it into more fine-grained groups. Let’s say, for example, that the various words for fabric have all been dumped into one group. There are six main ways (let’s say) to express the concept of fabric. For some reason, people typing in those different terms might respond differently to different offers. Working with different ads for each, and tracking their performance separately, might lead to better performance. It’s not as if the advertiser was lazy before—given their various other products, having a “fabric” group was reasonable. Just not 100% optimal. Let’s say you can use six different keywords that mean something more or less the same as fabric, and you want to build commerce-friendly phrases around each (cloth, pattern, material, and so on). Make sure you use six different ad groups, each one revolving around a different way of saying fabric, and then build related phrases onto each. It also helps if your ad title contains that keyword. Your clickthrough rates are usually higher if there is an exact match between a word in your title and the phrase the user has typed in; even very close synonyms don’t seem to do as well. You’ll probably want to write ad titles that are different for each group, for example: Fabric for Less Wholesale Cloth Looking for Patterns? Buy Unusual Material Using a single ad title for all of the diverse keywords in the campaign usually lowers performance (CTR), and this can cost you money. The same goes for the written copy that goes with the title. It should be tailored as much as possible to the keywords in a logically sensible ad group. 121 122 Winning Results with Google AdWords Advanced Tip: Couldn’t You “Set” the Level of Granularity Somehow? Wouldn’t it be cool if some kind of tool were available to help break down big campaigns with big ad groups into finer-grained ones? Obviously, the tool wouldn’t be able to rewrite all your ads, and it might get things a bit wrong, but it would be cool. My hunch is that Google support staff already have beta versions of such tools available now to help them advise advertisers on account improvements. In the future, such a feature might even find its way into the AdWords interface. Depending on your objectives, each of these ads might actually take the user to a different part of your website, or a different landing page (the target URL for the ad). When deciding on the target URL to enter with each ad, consider the user experience. Is the searcher’s experience going to be intuitive and seamless? Does the “buy unusual material” ad take her to an appropriate page on your site, or just the home page? If you want, you can even test both to see which performs better. As a rule of thumb, this process is always about improving your targeting. Secondarily, it’s a matter of usability and sensible navigation. Think carefully about targeting at each step of the process. Sales conversion rates generally go up when users get the exact information they were looking for right away rather than having to hunt for it. Currently on Google.com, on advertising as well as regular search results, the search engine user’s search words are being highlighted in bold, so this may also lead to higher CTRs (bold text is eye-grabbing) if you focus on making sure your ad titles and copy contain relevant keyphrases. Organize, Organize, Organize A fastidiously organized account leads to faster AdWords success and prevents headaches later. Let’s review four reasons for being careful about how you organize your campaigns, groups, keywords, and ads. The kind of organization I’m referring to here is basically what we discussed in the previous section: the idea of carefully piecing together a meaning tree within your account, with sensible labels on everything. Pretend you’re the corporate librarian and assume that your job is to set things up so that the average person could understand where to find everything. Multiple Persons Managing the Account At the beginning, it might be just you managing the account, but that’s rarely the case over the long haul. In many companies, a succession of people will be involved at one point or another. Even if it’s just you, you’ll find that things go much more easily if you organize carefully in the beginning. The “later-on you” might have real trouble figuring out where the “old you” put various keywords, or why certain bidding strategies were employed. Haste makes waste. CHAPTER 4: Setting Up Ad Groups Post-Click Tracking Depending on your goals, you will usually track user behavior after they click on your ads. Most typically, you’ll want to trace a specific type of conversion event (did they convert to a paying customer, a lead, or a newsletter subscriber?) back to the assessment of how well different ad groups and ads performed. Analytics can be a breeze when you’ve set up ad groups based on a logical structure of meaning, whether it be product line or different variations of similar words. By contrast, data can seem meaningless and random if you’ve built the account hastily, piling disparate phrases into various ad groups rather than organizing them thematically. If you set things up carefully, strong performance in a particular ad group is easy to interpret, and you can build on that knowledge. Set up tracking URLs (destination URLs with custom tracking codes) to represent each ad group, or preferably every ad and ad group. (Again, tracking URLs aren’t really required if you use certain tracking solutions, especially Google Conversion Tracker.) I recommend figuring this out right at the beginning, because the setup of tracking URLs is busywork that can eat up the better part of a day for larger accounts. You don’t want to have to do it twice. If you do nothing about tracking at the beginning, with a logical campaign structure, you can, at least, come back later confident that it will be easy to add different tracking URLs to represent each different ad in all of your groups. Tracking URLs are not hard to enter and don’t require any complex math or programming skills, just a numbering system (often one you invent yourself) that will help you keep score later. We’ll come back to this. Bottom-Line Performance (Ads Match Keywords) All the major search engine advertising representatives will tell you this: CTRs go up when your ad title matches exactly with the keywords typed in by the user. For the time being, there is no disputing this, although I think it’s a little too pat. If that were the only secret to good copywriting, everyone would do it, and everyone would have the same ad titles on any given search query. Zzzzz. Taking the general principle to heart is the important thing. The better you organize your ad groups and the keywords in them, the easier it is for you to write a variety of different ads to achieve granularity. In other words, by writing differently worded and differently titled ads for each group, you’ll wind up with ad copy that is closer in meaning to the keywords you’re targeting. This almost invariably creates a higher CTR across your campaign. There’s a very good reason to shoot for a higher CTR: it’s the predominant factor in the Quality Score algorithm that rewards advertisers with higher rank on the page for ads that are more relevant to users. In some cases, ads may not be shown at all if your Quality Score multiplied by your bid fails to reach what Google terms a “bid requirement.” I’ll cover this fully in Chapter 5. By doing everything right from the standpoint of organization and granularity, your campaign will become easier to run and make you more money. This frees up your time to work on other things while your competitors are killing themselves managing their pay-per-click accounts. They might also be goaded into bidding too high, then wind up shutting down their accounts in a panic, because they don’t understand one of the big secrets behind your high ad positions: proper campaign structure. 123 124 Winning Results with Google AdWords Avoiding the Horrors of Overlap I’m often asked if you can put the same keywords and phrases into different ad groups or even different campaigns. The answer is yes, but the result isn’t exactly as you might expect. If you sell eight different products that are all relevant to the same keywords, guess what, you can’t have eight of your ads (or even two of them) showing up on the same page of search results. Google calls this double serving, and it’s against the AdWords terms of service. As you can imagine, many advertisers would take advantage of this to crowd out competitors with multiple ads, and this goes against what the user is expecting to see in that space—a choice of different vendors. So, in the case of overlap (the same keywords in different ad groups or different campaigns), the AdWords system will choose to show a single ad from your account corresponding to only one of those keywords. How is that choice made? As with many such questions, Google is evasive on the point, preferring to emphasize the relevancy aspect over any messy talk about simply choosing the one where you’ve bid the highest. Strictly speaking, the ad chosen is the one from the group where that keyword has the highest Quality Score. So why do I speak of “horror”? After all, nothing breaks if you overlap. Your ad is shown. No problem. Usually it isn’t a problem. If you happen to get forgetful and create some limited overlap, nothing terrible will happen. But with significant overlap across different groups and campaigns, an account becomes virtually impossible to comprehend and manage, especially when multiple stakeholders come on the scene. You (or others) will find it difficult to interpret tracking data, to know which ads are going to show up, and to get a handle on how much you are paying for the keywords. If you’re caught in an account that already looks like this, pause some campaigns and groups until you’ve built a cleaner campaign. The Tale of the Impatient Client Overlap in the early going can be particularly troublesome, because highly targeted ad groups compete with clumsily built, more general ad groups. In Google’s new Quality Score regime (again, see Chapter 5), new accounts may be held to rather harsh scrutiny, so getting high CTRs out of the gate is a must. My team built a very focused, granular campaign for a B2B client. These campaigns are typically low volume with very high CPCs. They have a number of other unique qualities. Impatient with the volume, the client built their own general ad groups around our more specific ones, not bothering to check out where keywords were being duplicated. They enabled those, and promptly began contributing bad Quality Score karma to the account. With the very low CTRs that are inevitable when marketing to a niche audience using very general keywords, they made it harder for the account as a whole to generate the positive user feedback signals Google is looking for. Don’t “go scattergun” first, and refine later. In the new era of AdWords, especially for tricky accounts, go narrow first, and build out once your Quality Scores are proven to be OK. Magic Candles: Accounts That Need Repeated Deletions A typical example of what can happen with rampant overlap is the advertiser who decides she’s paying too much for some of her keywords and wants to lower the bids on them by up to 60%. Her trusty AdWords consultant goes in and lowers those bids significantly, hoping the total daily CHAPTER 4: Setting Up Ad Groups spend will be slashed from, say, $800 to $400. It seems OK for a few hours, so the consultant takes off for the long weekend and the client thanks him for his help. Lo and behold, like those gag birthday candles that never blow out, the same keywords “pop up” in another campaign and start working to display ads for approximately the same cost as before. The following week, it’s discovered that some higher-volume keywords are hiding in no fewer than six campaigns, and they must be ferreted out and deleted. Yikes! The same thing can happen even when similar keywords are in different ad groups or campaigns, if they are using broad matching. If you shut off college loans because it isn’t converting well, but leave school loans on somewhere else, school loans may “pop up” unexpectedly—unexpectedly because previously it had been dormant due to college loans siphoning off most of the impressions. This is a bit complicated; I discuss Google’s “expanded broad matching” in Chapter 9. The only easy takeaway from this description of the magic candles syndrome is that one round of bid adjustment is never the last word on your account performance. Your campaigns will need regular monitoring regardless, but require particularly close monitoring after you make a round of bid changes whose intended effect is to lower bids, or after you pause or delete certain keywords entirely with the expectation that this will improve campaign economics. Various bad things happen with too much overlap. Avoid it, and build a clean campaign in which the vast majority of your ad groups have a clear purpose, with keyword lists that do not also appear in other ad groups. You should also consider focusing on Google’s phrase matching option as opposed to broad matching. See Chapter 9 for more. Naming Campaigns and Groups Just a friendly reminder: giving memorable labels to campaigns and groups is part of the process of staying organized. To go back and name or rename an ad group if you forgot to do it at the beginning, you’ll drill down to the campaign level of your AdWords account to see the list of existing groups and click the check boxes next to the groups you’d like to rename. Then, from the gray buttons at the top of that list (Change Max CPC, Rename, Pause, Resume, Delete), click the Rename button. To rename a campaign, go into Edit Campaign Settings and edit the first field, labeled “1. Basic Information—Campaign Name.” Your naming system should be one that will jog your memory later. It might be based on different product lines, different words (fabric, material, cloth), or even different AdWords strategies. I sometimes name groups “experimental,” or even more specifically, such as “low cost experiments,” “developer jargon,” “competitor meta tags,” “unconvincing AdWords keyword tool suggestions,” after any number of nefarious strategies I might use to generate and test innovative keywords. The naming should dovetail with the structure as a whole—it should remind you to group similar keywords together, whatever similar means for you. You’re in Charge: Reevaluate Structure Every Few Quarters As you add keywords to groups over time, you might find your groups becoming too cluttered once more. Some phrases might merit being “hived off” and put in groups of their own. At that 125 126 Winning Results with Google AdWords point, you may want to consider starting new groups revolving around such phrases, especially if you’ve discovered new ones that generate high volumes of clicks—new concepts or terms that might be worth building on in their own right. Writing Your First Ads The first advertisements you write might not prove to be your best, but you do have to get them written to get things rolling. The AdWords interface will ask you to enter at least one ad right at the start. Later on, when setting up new ad groups, you’ll simply drill down to the ad group level of the AdWords interface and click on Create New Ad. As the interface will show, you have a very small space to work with: 25 characters for the headline, and 35 characters for each of two lines for the two-line description. The display URL (the web address that users will actually see below your ad) can only be 35 characters. However, the landing URL gives you plenty of leeway. It can be up to 1,024 characters. Some advertisers have very long tracking codes or complicated URLs for their catalog pages, so this helps. A few tips on format first. For the display URL you’ll typically just put in your home page URL. The user may be taken (by the landing URL) to a specific page on your website, but the URL displayed in your ad needs to be uncomplicated and, hopefully, should look trustworthy; for example, www.legumes.com. Sometimes I experiment with capital letters where appropriate in two-word URLs; for example, pagezero.com versus PageZero.com. I haven’t seen any conclusive difference in user response. Google won’t allow you to misuse capitals. You can’t alternate caps manically just to grab attention (www.GiLoOLy.BiZ), for example. Not that you’d want to make a spectacle of yourself in this way, but you’d be amazed at what some folks dream up. Choose the landing URL with care. You need to ensure that every ad you write uses a landing URL that gets two key things right: first, it must send users to the corresponding page on your site (this is preferably a targeted page that gets users to the information they need without an extra click); second, it must contain the correct tracking code based on whatever tracking nomenclature you’ve decided to use. If forced to do this, I use a unique tracking code for each ad. We’ll return to this, but an example of a landing URL would be something like this: http://www.legumes.com/lentils.asp?source=gaw&kw=23b Always Be Testing Advanced Tip: Don’t prejudge the impact of any element in an ad. What do the consumer response numbers say? The display URL is actually an element to test, if you have high click volumes and the guts to try new things. Some advertisers have the flexibility to try different domains. Others use descriptive keywords in a subdirectory name, such as www .mysite.com/pencils, and Google doesn’t seem to mind if your real URL doesn’t look exactly like that. Another tip you may have completely overlooked is that if you’re running out of characters for the display URL, you can leave off the “www.”! CHAPTER 4: Setting Up Ad Groups Nowadays, though, I rarely use this hand-crafted approach. I’d either use Google’s tracking or use a tracking tool that automates the process of inserting the correct codes into destination URLs using the AdWords API (Application Programming Interface). As for the ad itself: now the challenge begins. You have to squeeze your message into a small space. Some call it a haiku. But it’s really no different from traditional classified ads, except that here, you have the ability to test response in real time. Once you’ve entered that first ad, a maximum bid, and a couple of keywords just to get the account set up and running with your first ad group, you’re on your way—well, almost. Google has some fairly extensive editorial policies to contend with. Editorial Review No one has ever sent me a Google organization chart, but I’ve been fortunate enough to talk individually with dozens of Google staffers and executives at various levels of the organization and thus believe I have a decent feel for the company’s workings. Certainly, as with any industry, there is competition for good people. Google is one of the most prestigious companies in the technology world, and many good people want to work for the company. This works to Google’s advantage. Competitors such as Yahoo also have sufficient experience and prestige to attract topquality staff. I point this out because customer service plays an increasingly important role at Google. That said, in the past three years, it looks like their jobs have been subject to significant standardization and automation. Automated editorial “warnings” stop you from breaking the rules even before you enter mistakes. And written and verbal “decisions” are rare. Today, everything is moving in the direction of opacity, even though human oversight is as important as ever. In the increasingly rare event of editorial disapproval alerts (best visible at the Account Snapshot level), you have the choice to either correct the problem or contact your Google rep to ask for clarification. Often, staff will show you the courtesy of requesting a review by a policy specialist, but there are no guarantees. Most, but not all, of the rules are fairly clear. Responding to Controversial Editorial Disapprovals Sometimes, you can be the victim of a misunderstanding or localized bias of some kind, resulting in an unfair editorial disapproval. To head off potentially condescending responses from editorial staff who may not fully appreciate your depth of knowledge about your own business and the advertising business in general, try making an initial email contact to identify yourself fully, possibly with a brief bio explaining your background. Politely explaining small distinctions about your industry or your campaign that staff may have overlooked may be required, too. With so many advertisers to deal with, staff might simply assume that your case isn’t worth looking at too carefully, since many editorial disapprovals are fairly routine. You might have to remind them of your brilliance and convince them to squint harder at your correspondence. Unfortunately, this is a fact of life in dealing with customer service people who may be accustomed to receiving a large volume of inquiries. 127 128 Winning Results with Google AdWords The online medium can lead to brittle communications, and there is no more uncomfortable feeling than receiving several warning emails that your advertisement has been “disapproved.” I think that’s probably why Google has chosen to sharply reduce the number of such messages in recent years, by offering automated warnings that happen as you enter the ad, and also by burying more of their biases and policies within the Quality Score algorithm. It’s just plain harder to quibble with an algorithm. Many of us have a deep psychological aversion to disapproval. If you work for a larger company, you’ll probably be stunned that someone would nitpick you, given your big ad budget. But consistent policies are obviously better for everyone concerned, even though they might be applied or interpreted too rigidly in some cases. So the best advice is to treat a disapproval as a minor setback, and either adjust your ads to make them conform to the rules, or politely appeal. Why were so many minor editorial rules enacted in the first place? The history is interesting. According to Sheryl Sandberg, an early director of the Google AdWords Select program, Google was from the beginning trying to set an industry standard for advertisers. Pressure to do this increased as Google entered negotiations to form a syndication partnership with AOL, a partnership that they won away from their rival, Overture. A standards-based approach was not implemented simply to appease AOL or imaginary consumers, though. It was also intended to protect advertisers from one another. (The concept is one of a level playing field for all participants.) If an advertiser breaks one of the editorial rules, it may be creating an artificially high CTR at the expense of one of its competitors. Google wants to reserve the right to disapprove ads for that reason as well as for reasons of quality control and consumer protection. Quick Tips The full list of editorial policies is at http://adwords.google.com/support/bin/static.py?page= guidelines.cs&topic=9271&subtopic=9277. Probably the most important idea is mentioned near the top of the guidelines list: “As a basic rule, use clear, descriptive, and specific ad content that highlights the differentiating characteristics of your product/service.” Proper targeting generally makes gimmicks unnecessary. Google does a good job of describing its own rules, so I won’t duplicate everything they say; just a few words about them. Quite a few of the guidelines are basic matters of form—almost like the style guides reporters must follow when writing a news story. No repeated or unnecessary punctuation; don’t use all capital letters for anything except an acronym; spell words correctly. Some punctuation styles that were verboten in the early days are actually now recommended by Google! (How easily we forget how disdainful we used to be about things that actually work.) One example is the practice of capitalizing the first letter of every word in body copy. It used to be banned. Now, it actually appears as a favored example on Google’s editorial guidelines page. This is likely because it increases CTRs slightly in many cases, upping Google’s revenues (without bothering users unduly). It might not work for you. My colleagues and I tend to think this looks unprofessional, but if you insist, we’ll test it. Some guidelines are designed to prevent you from making misleading claims. Some forms of “come-on” might be disallowed—especially if they’re clearly inaccurate. Phony low prices designed to induce clicks might be a waste of your money, so Google might actually be doing you a favor by asking you to reword your ad. CHAPTER 4: Setting Up Ad Groups Google has a set of policies aimed at making life tough for affiliate advertisers. At one time, they were actually forced to identify themselves in ad copy with the designation “aff.” From there, things have gotten a lot quieter, but mainly because Google has launched a deliberately quiet campaign to reduce the numbers of affiliates who advertise. An affiliate is someone who sends referral traffic to, say, a parent company like eBay, receiving a commission if that traffic results in a purchase. They might publish recommendations with their affiliate links (including a code, so they get credit for a sale) in an email newsletter or a website or blog. Or, they might spend money on AdWords clicks, hoping to pocket the difference between the cost of the paid search clicks and the commissions they make from sales. As a result of too many affiliates crowding the page of sponsored listings on some queries, Google enacted a rule that limits the number of advertisers appearing for a given display URL to one (so there won’t be four ads for eBay.com on the same page, for example). But that’s really just the beginning. Manual intervention and automated means of assigning low Quality Scores to affiliate sites are all part of the process, too. Over the past three years, I think it’s safe to say that Google’s evolving policies have been, in large part, designed with an eye to making it harder for affiliates to advertise. If there is the slightest “seedy-looking” thing about certain offers, or if there are too many ads from (let’s say) eBay on a page, consumers tend to react negatively. Examples of Google’s human evaluation methods circulate around the Internet from time to time. These methods—applied, I presume, to both paid and organic search results, and websites vying for rankings on either side—have shown that Google coaches its human “landing page raters” to grade affiliates harshly unless they offer something unique. (For example, human raters might be offered descriptions of “what a ‘thin’ affiliate site” is.) Examples of sites and pages that typical human raters don’t like are fed into the AdWords Quality Score algorithm to train the system to recognize signals of “seediness.” Anyway, on the editorial front, if you feel you’ve been the victim of a gray area ruling that goes against you, by all means reply to Google’s support emails with a polite request for more information, clarification, and possibly an appeal of the ad disapproval. Depending on your geographic location, your reply will be to AdWords-support@google .com, or if in the UK, AdWords-uk@google.com. Or just reply directly to the editorial messages you receive by email. My final word on this subject: don’t spend your life swimming against the current or tilting at windmills. Make your views known, to be sure; then, get back to work! Time Lags and Special Rules Some foibles in Google’s editorial process are never explained in any documentation. There are certain background considerations that need to be kept in mind. In particular, you may find yourself tripped up by delays in getting ads running for new accounts, or new ads within an existing account, or even in unpausing an account that has been dormant for some time. 129 130 Winning Results with Google AdWords Google vs. Search Network Partners Editorial approval delays are sometimes attributed to differential editorial standards at network partners such as AOL. You tend to hear stories circulating around about how “instant” ad activation is on Google.com, and that it takes a bit more time to show up on partner sites, but the routine isn’t really disclosed as far as I know. You should become concerned if your ad isn’t approved and generating impressions on your keywords within 24 hours; 48 hours in busy times, allowing for weekends. Automated vs. Human Review; Ramp-Up Timelines Many people assumed (based on how AdWords was marketed in the early days) that ads get “up and running right away.” Indeed, many advertisers found it exhilarating that they could see their ads running nearly immediately and that editorial review mainly consisted of automated tests to see if ads broke any rules, followed by a post facto checkup by a live editor as Google’s software deemed necessary. Not so fast, though. While the process has often worked basically like this, Google has now deviated significantly from it, especially on popular keywords with many established advertisers. Although ads may be given a few token impressions and clicks (for what reason I do not know—to confuse advertisers further or to throw Google’s competitors off?), it’s been confirmed in support calls that new ads can sometimes sit in an editorial queue for up to a week. The most glaring example of this will be over the Christmas holiday period. After December 10 or so, don’t count on getting any new ads approved in anything less than ten days. Plan ahead to avoid disappointment. Over the years, I’ve also come to suspect that (for reasons I can’t explain) new accounts and new keywords undergo a “ramp-up” process (even taking into account delays in syndication through network partners or content targeting). They’re served only partially at first, gradually increasing in volume, until one day, a week or several weeks later, they finally reach full delivery. Again, no one at Google will confirm this, but plotting the number of daily ad impressions over a period of three to four weeks for any new campaign or ad group containing new keywords would offer some proof of this. The exact pattern just changes from year to year. Now, new accounts and new parts of accounts are subject to an opaque “buildup” phase wherein they are evaluated on all kinds of variables (primarily CTR) in order to establish Quality Scores for keywords. It’s not clear whether that evaluation phase also limits delivery until there is some certainty level about the readings. As if it weren’t difficult enough to project your monthly spend prior to launch, you may have difficulty getting any kind of consistent feel for how much you’re spending until everything has been running steadily for a few weeks. Established accounts tend to have fewer hiccups and “phase-in” issues, but if you’re doing anything involving “turn on a dime” seasonality, I’d budget for a week or more of slow initial delivery. No Double Serving As mentioned earlier, Google won’t allow you to use multiple AdWords accounts to run several ads on the page at the same time on the same keyphrase. Admittedly, this can be a gray area. A large diversified company won’t be able to prevent its various divisions from bidding on the CHAPTER 4: Setting Up Ad Groups same keywords, and if the lines of business are distinct enough, Google likely has no problem with it. However, case studies I’ve seen from companies like IBM have shown that it can be bad economics to bid against other divisions of your own company! Even in smaller companies, there can be allowable forms of double serving. For example, a division might target consumers exclusively with one site, and a separate division might be using the same keywords to send traffic to their B2B site. If you’re trusting, you’ll turn to your Google rep and ask for clarification as to whether this might be allowed. If you’re a shoot-first, askquestions-later type, you’ll just do it, avoiding messy questions, and hope that Google doesn’t discover or have a problem with the double serving. Rest assured, though, that in 99% of cases, Google knows what you’re up to. The principle here is simple: allowing companies to blanket the page with their ads by creating multiple AdWords accounts would be an abuse of the system, unfair to other advertisers, and a bad deal for users, who expect to see some choice in listings. The terms of Google’s arrangement with advertisers do not, of course, include the right to buy up all of the screen real estate devoted to advertising. Evidently, Google prefers the competitive auction process because it provides users with more choice and drives up Google’s revenues. Hey, they’re the publisher and it’s their website, so that’s their right. 131 This page intentionally left blank Chapter 5 How Google Ranks Ads: Quality-Based Bidding A s I showed in early chapters, the manner in which search engines display query results is becoming increasingly complex. Google’s ranking algorithms for the organic web index are constantly evolving. In addition, Google now organizes a lot of different types of information: news, video, local and map results, weather, Google Groups discussions, and much more. To display an appropriate mix of information, Google (like other search engines) attempts to discern user intent through a mix of personalization and guessing at what a query’s intent means in part by going on past data. So if you were to type a common news-related query, Google might show news results above the web index results. For some local vendor queries, Google might show six or more local listings, pushing everything else further down the page (see Figure 5-1)! Google calls this blended approach to showing search results Universal Search. It has changed marketers’ assumptions about what they can expect to accomplish by optimizing websites solely for the web index results. Fortunately for your focus in the paid search realm, the screen real estate taken up by paid search ads hasn’t changed a whole lot. Paid search ads are relatively easy to trigger in the familiar slots in the areas above and to the right of the organic search results. Phew, so all that increasing complexity in the search world won’t affect our paid campaigns, then? Sorry, but the clever engineers at Google have ratcheted up the complexity of their keyword advertising system, too. There is now a whole formula that affects both how high your ad ranks on the page in relation to other advertisers, which affects the visibility of your ad and therefore your click volume, and keyword Quality Score for the purposes of meeting a newly designed Bid Requirement, which can make your ad ineligible to show up at the time of the query. I explain all the generations of ranking formulas, leading up to the present one (released late August 2008), in the current chapter. This formula used to be relatively simple in the period 2002–2005. Even so, it required several pages of explanation in the first edition of this book. The concept of multiplying your bid by your clickthrough rate (CTR) to arrive at your “AdRank” was not intuitive to many advertisers. 134 Winning Results with Google AdWords FIGURE 5-1 Google guesses at user intent, displaying a large number of Google Local listings for the search query “green bay wi plumbers.” Eventually, folks got used to it. In August 2005, a newly announced formula led us down the path towards an increasingly complex and opaque bidding regime. CTR was replaced by Quality Score (QS), and the method of calculating QS has been changing steadily. It’s not entirely unlike the mysterious and changeable proprietary algorithms Google uses to rank other types of search results, most notably the unpaid listings in the web index. In this chapter I will delve deep into the concepts of Quality-Based Bidding (QBB),1 covering the current principles that drive the system; the historical evolution from then until now; and the basic ranking factors that affect your ad rank. To illustrate the practical side of succeeding in this environment, I’ll also offer case studies. If you get lost, well, I’m a big fan of Occam’s razor. So keep in mind that two key principles behind the auction—relevancy or tight targeting, and catching bad guys—can help you simplify all the detail. The other thing to remember when you get lost is that there is considerable CHAPTER 5: How Google Ranks Ads: Quality-Based Bidding continuity in the system’s operation from 2002 to this day. CTR is still a core driver. If you don’t get clicked, it’ll be hard to make your keyword stick. A quick heads-up: what counts as a good CTR is relative, not absolute. A good CTR in ad position 2 is not the same as what constitutes a good CTR in ad position 11. Typical CTRs vary by industry and by keyword, too. In Google-speak, they “normalize” CTR based on such factors as ad position, so don’t get caught up in thinking that you have to be in a very high ad position solely for the CTR benefit. Google’s formula attempts to compare apples with apples. Encapsulating the Concept of Quality-Based Bidding Much of the advertising world is based on loose targeting. Put another way, it’s mass or broadcast advertising, dreamt up by old-school advertising moguls from the days of the TVindustrial complex. These were the sorts of chaps who, if asked off the record, might tell you that they love the smell of napalm in the morning. Metaphorically speaking, of course. If you watch TV daily, not a day will go by that you won’t say to yourself: “Why are they showing this to me?” You might be offended by a given ad because it’s talking about some embarrassing ailment you don’t have. You might even get a bit angry at such loose ad targeting! It’s the napalm of the airwaves. This is the opposite of what Google wants the search experience to be like. So they’ve built some serious disincentives into their auction to discourage loose targeting and to encourage tight targeting. To put it in a single sentence, Google’s ad system is a complex auction that takes into account your bid amount (on any given keyword) along with a variety of other (mostly relevancy-related) factors, which, put together, is called your Quality Score. Quality Scores are assigned by keyword, but ads, landing pages, and account history may affect Quality Scores. So when we speak about Quality Scores, assume that we are talking about an individual keyword or phrase, but don’t assume that the keyword’s performance alone is what influences that keyword’s quality. (OK, so I admit, that was more than one sentence, but you could have stopped reading after the first sentence.) Under this system, the lowest possible minimum bid is actually lower than it has been in the past; it’s one cent ($.01) or the equivalent very low amount in your currency unit. That’s mostly window dressing, however. You’ll likely have to bid considerably higher to reach a desirable ad position. Quality Scores Are Based on (at Least) Three Broad Types of Data Googlers often stress that this process is nearly 100% automated. Decisions on where to rank ads, and whether keywords are active at all, are complex and data-driven. Here are some guidelines to the broad types of data Google looks at when it comes to keyword performance. 135 136 Winning Results with Google AdWords Historical Data Accounts with a lot of past history have generally come through the new era unscathed, especially if they have a history of strong CTRs and other indicators of relevancy. Historical data on a keyword—and on the entire account—make it relatively easy for AdWords to assign a “true” Quality Score to any given keyword. The more known information Google has about keyword and account performance, the more accurate the Quality Score is likely to be. Predictive Data Setting up a new account poses much different challenges than managing an account with a strong history. Google has no data for your keywords. Technically, they don’t know how relevant your campaign is to users because you haven’t run it yet. But don’t think that’ll stop them from trying! They’ll use past search data and data from other advertisers’ campaigns to assign initial Quality Scores to your keywords. This is significantly more precise (and in many cases, less generous) than the old version of AdWords, which assigned new keywords a default CTR based on industry averages. Whereas in the old days your initial ad positions were often skewed a bit to the generous side, in many areas Google’s system now takes a “show me” approach to your account. Google now has more than enough advertisers (and in case you hadn’t heard, enough money to keep the lights on), so your account is going to have to prove its mettle a bit at first. That’s probably going to require you to bid a bit higher than you’d like in the early going. Google began experimenting with a “suspicious outlook” on new accounts at least a year prior to the switchover to full-on Quality-Based Bidding. In part because a minority of advertisers—in particular, affiliate advertisers bidding low amounts on a large number of irrelevant keywords—were wont to flood the system with junky campaigns that would show bothersome ads to users during the data evaluation period, Google took steps to make sure these ads rarely made it live in the first place. For new keywords in new accounts, Google—with the help of AdsBot, which scans your landing page and website for content—now uses a host of means to check out the predicted relevancy of your keywords, ads, and web pages. On the keyword front, Google uses (among other things) historical data from past advertisers’ campaigns to try to guess at whether the keywords you’re choosing have any commercial viability. They also look at how well your keywords, ads, and landing pages relate to one another. Gone are the days where you can try to show your ad on keywords about the daily lottery if you’re trying to sell home equity loans. Sure (as the UK-based advertiser who tried that told a panel at SES London a couple of years ago), there is some logic to this—someone’s worried about money, so they buy lottery tickets, so they might be interested in a loan. Apparently, though, on certain queries, people really just want what they said they wanted—in this case, lottery results. This advertiser would have found he garnered low CTRs on his terms if he advertised against lottery results keywords. Using predictive data, Google disincentivized this advertiser from even finding out. Google had so much data on poor CTRs for similar irrelevant ads showing up against lottery results keywords that it imposed a Quality Score hurdle on any advertiser who wanted to experiment with such loose targeting. You’re certainly free to buy “loosely relevant ads” around the Web—if you do display advertising and want to negotiate ad buys with publishers and networks that are eager for your CHAPTER 5: How Google Ranks Ads: Quality-Based Bidding dollar without regard to how closely your offer is related to their content. Not so on Google Search proper, though. Google currently works on the premise that users are unequivocally looking for ad listings that are relevant to their queries, and Google has found that off-topic “diversionary” ads make some search engine users irate. So Google won’t risk killing the golden goose, which, as I’ve already established, is searcher loyalty. Opinion and Arbitrary Determinations Editorial rules and human assessments of keywords, ads, and websites were always a big part of the paid search auction. For whatever reason, Google has taken steps to reduce the level of human enforcement of rules. By “human enforcement,” what I really mean is that Google has reduced the frequency of direct staff interaction with advertisers when it comes to infractions; and Google has reduced the number of “on-off” and “yes-no” policy determinations. Both trends have resulted in more decisions being moved into the opaque world of Quality Score, allowing Google to set a sliding scale of incentives that raises prices on advertisers to varying degrees as opposed to triggering a confrontation. For example, whereas in the past certain types of keywords might have been banned either across the board or selectively in response to complaints, now Google simply discourages advertisers from advertising on those same words, by making high Quality Scores relatively hard to attain on them. Another example: in the past, Google editorial staff (as always, aided by automation) might have looked for a short list of user experience violations on an advertiser’s website. Pop-ups and a few other things were discouraged. Now, a wide range of such experiences might be included in the increasingly complex formula for determining landing page quality. Google still uses a combination of automated and editorial means to catch “no-no’s.” But because the new system is so slick and opaque, advertisers today sometimes think everything is automated. Not entirely so. Sometimes editorial decisions are made outright about a given company, its products, landing pages, or messages. These can be human-driven decisions. Getting anyone to admit this, or to explain exactly what the problem is, can be like pulling teeth. To be sure, major problems are rare. But they happen. Paid Search Ranking Formulas: Past, Present, and Future The key to understanding the basic workings of paid search ranking formulas today, in my view, lies in recognizing the continuity of today’s model—to be precise, what I refer to as “AdWords 2.7” below—with the previous approach used by Google from 2002–2005 (what I’ll call AdWords 2.0). Here are some brief reminders of what past models looked like leading up to the current one. 137