Chapter 8

Published on May 2016 | Categories: Documents | Downloads: 50 | Comments: 0 | Views: 495
of 37
Download PDF   Embed   Report

Comments

Content

chapter 8:

Traffic Building

“ “ “

A wealth of information creates a poverty of attention.
Herbert Simon1


The Price of Words

Just give me some leads that don’t come out of a phone book, huh, you give me something hotter than that and I can close it. It’s just a streak. I’m gonna turn it around.
Shelly in Glengarry Glen Ross2

” ”

Mesothelioma is a terrible disease. It is a form of cancer, almost exclusively caused by exposure to asbestos. Each year approximately 3,000 Americans contract mesothelioma. Most are construction workers, shipyard employees, or others who worked with asbestos decades earlier. Once diagnosed, life expectancy is less than two years. Patients suffer increasing pain and disability prior to death. Mesothelioma is also big business. The tight connection between asbestos exposure and the disease makes mesothelioma a prime candidate for class action lawsuits against businesses that produced or used asbestos. Per patient awards in such a suit, which often end early in a settlement, average around $1 million dollars.3 Law firms representing clients receive approximately 40% of such awards. Lucrative awards make attracting the attention of patients a top priority for these law firms. The Internet is a major competitive battleground. The long latency between exposure and disease makes it likely that many of those exposed will have changed jobs and addresses, making it difficult to reach them through phone or mail. A better strategy is to rely on sufferers self-identifying. Online sites are a primary source for a majority of Americans seeking health information4. Attracting visits results in some of the highest prices paid for online advertising. The preferred choice for this advertising is pay-per-click search advertising. Search advertising has grown rapidly in the past few years. While

this chapter
Nothing happens on a web site without visitors. Web content is growing faster than web users. At the same time, much more accountable campaigns result in more rational management of traffic-building campaigns.

Topics covered in this chapter include: > > > >
The Price of Words Traffic as Value Search Engine marketing Traffic by Assocation

it lacks flashy graphics and brand-oriented imagery, the simple text-based ads that appear on Yahoo!, Google, and other sites are accountable and efficient. By connecting advertisers with searchers revealing a specific immediate interest by their search terms, at a price determined by a never-ending auction, search marketing firms are using the power of DNI capabilities. Table 8.1 illustrates the very high cost of a keyword ad for the phrase “peritoneal mesothelioma” using the Yahoo! search engine. Firms pay as much as $100 per click to generate a visit to their web site. Whenever someone types this phrase in the Yahoo! search page, sees an ad displayed to the right of the normal search terms, and clicks on the ad, the advertiser must pay Yahoo! a fee.

Table 8.1
Bids and Costs of Keyword “Peritoneal Mesothelioma” Ad Position 1 2 3 4 5 Max Bid $100.00 $99.99 $99.98 $30.01 $25.00 Cost per click $99.99 $99.98 $30.02 $25.01 $20.01

Source: Overture bidding tool, 2/18/05

Even though all of the ads generate a link to a web site, the location of the ad matters for attracting attention and generating a visit. On the Yahoo! system, ads appear according to their bid ranking.5 If you want the highest ranking, you must bid the highest. The second ranking bid gets second position, and so on. The maximum bid of an advertiser is not the same as the price paid. The actual price an advertiser pays for a click is also determined by the bid of the firm below it in the ranking. Table 8.1 shows both the maximum bids and the resulting costs-per-click of the top five bidders for the phrase. In this case, the top two advertisers are paying much more for their position than positions 3 through 5. A visit to the top listed web site, costing $99.99 per click, presents something of a puzzle. The majority of the page is dedicated to patient information, without any apparent advertising and few direct links to a sponsor. There is little “call to action,” where visitors are asked to do anything commercial.6 This approach requires a two-step conversion from visitor to client. First, the sponsoring law firm is boosting source and information credibility by downplaying a commercial connection and emphasizing health concerns. Only then is the visitor likely to find the links to the law firm’s specific information. Such a soft-sell approach must be balanced against a failure to convert, especially given the high cost of traffic. A little math demonstrates how valuable these visitors must be. If 1% of web site visitors 249

250

Part 2:

Essential Skills

choose the sponsoring law firm to represent them, a very respectable conversion rate for such an initial stage of interest, the cost of client acquisition is $100/.01 = $10,000. A lower conversion rate boosts the acquisition cost accordingly. Paying $10,000 for a new customer can only be profitable for extremely valuable customers. This is where the tight connection between exposure and disease, and the high value of the lawsuits, enters the picture. A law firm expects to receive hundreds of thousands of dollars in legal fees for each successful client it acquires. The efficiency of the keyword advertising system relies on competition driving pricing. If law firms find that they cannot convert visitors, or that settlements cease to be as lucrative, keyword prices will fall. Controlling legal fees and class action lawsuits is an ongoing political debate. One consequence of such a legal change would be a fall in the value of these keywords. Alternatively, web site improvements that boost conversion rates from visitor to client would tend to drive up keyword values. Fortunately for search advertisers, very few words are as expensive as mesothelioma. On Yahoo! and Google bidding starts at a dime or less. The eventual equilibrium price is a combination of keyword focus, alternative methods of customer acquisition, the power of domain names and organic search to stimulate visits, potential customer lifetime value, web site conversion effectiveness, and the intensity of advertising and market competition. The growth of keyword advertising has made a dramatic change in the online world, and has stimulated an emphasis on accountability that is rippling throughout marketing. This chapter looks at the range of tools capable of generating new visits to a web site. Traffic has value, and an effective traffic-building plan is valuable to any site. Naturally occurring traffic is especially welcome, and should be the centerpiece of any approach. Beyond that, cost-effective and profitable paid advertising injects additional traffic in an accountable and traceable way. In some cases more traditional advertising venues, such as radio or television, may be a valuable addition to the marketing mix.

THE DIFFICULT BATTLE FOR WEB TRAFFIC
Value and Scarcity > Online Content, Attention, and the Diamond-Water Paradox
The diamond and water paradox is one of the oldest confusions in economics.7 At first thought, students feel it is obvious that a product’s value must determine its price. Something that provides an essential benefit must be valuable, and something without any inherent benefits must

Chapter 8:

Traffic Building

251

be cheap. An immediate difficulty arises when comparing the relative prices of water and diamonds. Deprive someone of water, and an individual dies in days. Diamonds, on the other hand, have no biological benefit and are difficult to work with. Outside of some specialized machine tools, they have few productive uses. They are almost exclusively optional ornaments. Despite the critical role of water and the inessential nature of diamonds, a diamond is vastly more valuable than water. Of course, scarcity is the key. Water literally falls from the sky, while diamonds are exceedingly rare. Even extraordinary value can be driven down by abundant supply.8 The diamond-water paradox lives online in a variety of forms. Email is highly valuable, but users are free to send and receive to their heart’s delight. Online mapping sites provide efficient driving to unfamiliar locations, but are supported by online ads without any user fees. Online magazines, travel sites, and global art museums all provide entertaining and educational content to all that arrive. Although the number of Web users continues to grow, the amount of material on the Web is rising even faster. The battle for attention and traffic is intensifying, and the true scarcity becomes user attention. Every site faces a difficult struggle attracting visitors. A simple measure of relative scarcity is the number of web pages per user. Figure 8.1 shows the steep increase in content per U.S. Internet user between 1998 and 2004.9 Even omitting nonEnglish pages from consideration, the readily available online material is much higher than just a few years ago. Potential visitors to a site use search tools to find valuable content within the billions of alternative web pages. Even though the top search engines index billions of pages, their coverage differs and their ranking of the best matches varies substantially. First-generation search engines were not particularly strong in ranking results effectively, and high value results could be far down the results list. Despite this, a study of Excite users found 58% of users did not look beyond the first page of results. For the search engine Alta Vista, 85.2% of users only looked at the first page of results.10 Placement on the first page of search results is valuable.

Web Pages in English Per U.S. User 25.00 20.00 15.00 13.01 10.00 5.00 2.93 1.42 0.00 6/98 6/2000 6/2002 6/2004 23.63

Figure 8.1
The Battle for Attention is Getting More Intense

252

Part 2:

Essential Skills

Figure 8.2 highlights how the Yahoo! and Google top 100 search return items vary dramatically for the same search term, in this case “hybrid car.” Only the blue dots appear in the top 100 on both sites, with a line connecting them showing where each appears. For example, the top listing on Yahoo! is the seventh listing on Google. However, the number 4 listing on Google does not appear in the top 100 of Yahoo!, and the same is true for the number 6 listing on Yahoo!. As this tool shows, appearing high on one search engine does not guarantee a high listing on a different engine. Maintaining a top search engine presence is not a simple task as the same approach is not valid for all engines.

> Novelty No Longer Guarantees Traffic
In addition to the high volume of online content, online users’ growing experience makes attracting new visitors difficult. Exploratory browsing falls with expertise.11 Faced with busy schedules and seemingly endless new alternatives, many visitors resist adding new destinations and activities (see Figure 8.3). Attracting new visitors requires motivating a visit from locations they visit regularly. During the early days of the Net, high-quality sites were few and recreational browsing common. The situation was like a small town, where eventually any store gets a view because there isn’t much to see. Early sites received large amounts of free publicity. Lead sites in almost every industry were highlighted in books, trade magazines, news-papers, radio, and television. First movers were able to count on traffic boosts due to these reviews. Now, free media is hard to generate, and there are far fewer opportunities to be the first site focusing on a topic.

> Spending on Traffic Building
Internet advertising and online traffic-building services help overcome the plentiful supply of content and the challenge of attracting visitors to a site. Online spending has surpassed the boom years, and in its first ten years (1994–2004) has grown faster than the corresponding first ten year growth of either television broadcasting (1949–1958) or cable (1980–1989, all in 2004 dollars). The majority of this is traffic building in nature. For the fourth quarter of 2004, search accounted

Figure 8.2
Overlap and Differences in Top 100 Returns for Search “Hybrid Car” on Google and Yahoo!
Source: Courtesy of LANGREITER.COM; Yahoo! Logo Courtesy of Yahoo! Inc.; Google Logo Courtesy of Google Inc.

Chapter 8:

Traffic Building

253

for 40% of the total. These are exclusively devoted to building traffic to a site. Traffic building also drives much of the spending on display ads (18%), classifieds (18%), sponsorship (6%), and referrals (3%) (see Figure 8.4). There is also a rapidly developing service bureau industry devoted to search engine optimization, search advertising, and other methods of building traffic. While it is possible for small- and medium-size business to do many of the steps themselves, there is a viable outsourcing option for a wide range of traffic-building activities. The major search engines develop networks of advisers, creating advertising “ecosystems” Figure 8.3 familiar with their tools and approaches. Search marketing has dominated the Scarce Time Can Lead to Avoiding Novelty recovery in online advertising, primarily Source: © The New Yorker Collection 1993 Robert Mankoff from cartoonbank.com. All the result of highly accountable paid Rights Reserved. search engine listings, as seen in Table 8.2. This is broken into three categories. The first, organic search engine optimization (SEO), is the spending to ensure that search engines rank a site highly on the relevant search criteria and searchers “organically” find the listing and the site. Search engine marketing (SEM) technology is rather small, and refers to technology spending to

Annual $ Ad Revenue Growth – First 10 Years $10,000
$8,859 $9,626 $8,188 $7,267 $6,010 $4,621 $5,030

$8,000 $ millions

$8,087

$7,885 $7,134

$6,557

$6,000

$4,000
$2,787 $2,162

$3,698 $3,180 $2,495 $1,920 $1,190 $745 $1,580 $1,853 $2,080

$2,000
$1,012 $358 $147

$0

$55

$295 $267

$907 $499

Year1

Year 2

Year 3

Year 4 Broadcast

Year 5 Cable

Year 6

Year 7

Year 8

Year 9

Year 10

Internet

Figure 8.4
Rapid Growth of Internet Advertising, Much of it Traffic Building
Source: Internet Advertising Bureau

254

Part 2:

Essential Skills

Table 8.2
Search Engine Marketing is a Multi-Billion Dollar Industry 2004 Advertiser Spending (Million $) Organic SEO SEM Agencies In-house 492.1 178.8 313.4 Share of Total 12.0% 4.4% 5.8% 36% 64% Share of Tactic

SEM Technology Leasing SEM Agencies In-house

71.6 30.0 20.8 20.8

1.8% 0.7% 0.5% 0.5% 42% 29% 29%

Paid Search Advertising Search Media Firms SEM Agencies In-house
Source: SEMPO

3,523.1 3,059.0 181.0 283.7

86.2% 74.8% 4.4% 6.9% 86.8% 5.1% 8.1%

make onsite search more effective. The dominant share of spending is pay-per-click advertising, on major sites such as Google, Yahoo!, and MSN but also a number of additional more specialized search engines. This growth has led the online advertising rebound.

A WEB TRAFFIC PLAN
Traffic-Building Sources
Without visitors the best web site is a wasted resource. Generating traffic, which can be expensive and difficult, is an essential web marketing skill. Some of the most effective and least cost tools require creativity, imagination, and an attention to qualitative areas of marketing such as branding and publicity. At the same time, quantitative approaches such as keyword advertising and a careful attention to metrics can generate substantial traffic and profits. A web traffic plan is a combination of strategic and tactical choices that a company makes to build an active user base. It combines one-time actions with choices that can change daily. Traffic sources fall into five main categories. They are (1) branding decisions, (2) search engine marketing, (3) affiliate networks, (4) online banner advertising, and (5) publicity and word of mouth.

Chapter 8:

Traffic Building

255

> Branding Choices
A strong brand creates a ready source of inexpensive online traffic and a natural starting point for many online users seeking information. Figure 8.5 highlights the various sources of traffic to the Club Med web site. One of the most effective sources for Club Med is the power of its wellestablished brand. The majority of users go to a branded site without first consulting a search engine, external links, or following an ad banner. Rather, they either type the domain name directly into their browser or revisit a site based on a stored bookmark. Both of these are strongly reinforced by the power of the brand.12 Chapter 7 discussed how a well-crafted system of domain names naturally builds brand as well as traffic. An appropriate domain name system makes it much easier to include URLs on branded material and to include web addresses into advertisements, billboards, direct mail, and billing statements. A domain system that reflects how consumers view the brand will be memorable and accessible to a web user (and hopefully not like Figure 8.6).

> Search Engines
Organic search traffic results from a visitor going to a search engine, typing in a keyword phrase, and following one of the unpaid links to a site. It is free traffic, but it is not easy to achieve. The most important success criterion is an appearance high in the list of search returns. This results from strong content, high visibility, and search engine optimization. While each search engine follows its own approach in constructing its return list, there are commonalities across the leaders and important themes to understand.

(a) Domain Names and Branding

(b) Banner (c) Search Engine Listings

(d) External Link (e) Destination Site

(f) Word of Mouth and Publicity

(g) Paid Links

Figure 8.5
Sources of Web Site Traffic
Source: 8.5a, b, e, f Courtesy of Club Med; 8.5d © TangoDiva.com - premier travel magazine for women; 8.5c, g Courtesy of Google Inc.

256

Part 2:

Essential Skills

Figure 8.6
Domains Need to Match Branding
Source: © The New Yorker Collection 2001 Sidney Harris from cartoonbank.com. All Rights Reserved.

Paid keyword advertising uses money to move up the search engine return list by purchasing sponsored links and keyword ads. These ads display at the top right, top center, or bottom of the first return page. Keyword advertising has grown very rapidly in the past few years, as Yahoo!, Google, and others discovered the effectiveness of auction allocated keyword ads. The mesothelioma example demonstrates how keywords can be highly valuable, even without images and branding capabilities. The fundamental benefit is relevance—searchers reveal themselves to be in active information collection mode. The keyword search is a signal of interest. The rapidly developing industry surrounding paid search engine advertising is based on results. Tools and service bureaus help an advertiser identify the most effective combinations of phrases and bidding strategies. Many of these tools optimize keyword selection around the web chain that follows from this traffic. Visitors from each keyword phrase can be followed throughout their web visit, and in many cases their commercial value tied back to the keyword ad phrase that caught their attention. Promising keywords are kept, disappointing ones discarded.

> Affiliate Networks
Affiliate networks are sites that maintain links in order to receive a potential fee for generating traffic. This charge may be a simple per-click fee, or it may be based on additional actions such as registration or purchase. One of the most interesting aspects of affiliate networks is determining which of these possible actions should be the basis of the fee. Affiliate networks may be exclusive or non-binding. Online sites such as fatwallet.com are multiple category, multiple vendor sites offering coupons and discounts to shoppers. They have relatively little content and exist around the affiliate fees as a primary business focus. At the other end of the spectrum are sites that have a few affiliate links as an easy extra stream of revenue to go along with their normal content.13

> Banner Advertising
Online banner ads are the most visible forms of traffic building. Companies can boost their traffic by placing ad banners on ad-supported web sites. A small but relatively predictable fraction of the viewers of these ads will click through to the web site. One virtue of online web advertising is scalability. An activity is scaleable when its size can be rapidly expanded without changes in its performance and capability. Online advertising has a wide

Chapter 8:

Traffic Building

257

range of possible advertising exposure levels. A company can buy a few thousand impressions, or with proper care can select millions. As was the case for keyword advertising, banner ads will vary in productivity by advertising copy and by advertising site. Standard reporting software and ad campaign software creates information tags linking the specific page and ad to an arriving visitor. This permits tracking of acquisition costs and advertising productivity.

> Mass Media Advertising
Under certain circumstances traditional mass media can be an effective online traffic source. Pepsi adopted the Web as an effective medium for reinforcing its branding, partnering with companies such as Yahoo! and Apple that also have very strong appeal with its core market. A co-promotion with Apple is aimed at boosting sales of Pepsi, but also boosting visits to the Apple iTunes site (Figure 8.7). Lucky winners get a free download from Apple on iTunes.

> Word of Mouth and Publicity
An illusive but powerful traffic-building source is publicity, promotion, and word of mouth. Events can make a site famous overnight, with millions of visitors. An amusing example happened during the 2004 Vice-Presidential debates between Dick Cheney and John Edwards. Seeking to make a debating point, Mr. Cheney told viewers that they could go to “factcheck.com” and see for themselves. What he meant to say was “factcheck.org,” a nonprofit site run by the Annenberg school. Instead, the factcheck.com site took fifty thousand visitors within the first hour to a site maintained by George Soros, which strongly critiqued the Bush-Cheney team. Only 200 visitors had gone to the site during the previous day.14 Consumers consider word-of-mouth recommendations the most persuasive and credible information source. Building natural links into email communications, online visits, and other electronic material helps spread these recommendations. As publicity and word-of-mouth activities have a strong social networking and community connection, a detailed discussion of these efforts is postponed to Chapter 10.

Figure 8.7
The Pepsi-iTunes Promotion
Source: Courtesy of Pepsi-Cola North America

258

Part 2:

Essential Skills

Traffic Volume and Quality > Typical Visit Patterns
Each step of a web site visit leaves an entry in a web log. Chapter 7 highlighted the information available from web logs about visit duration, common visit entry pages, and common visit exit points and how that shapes content decisions. Chapters 9 and 10 discuss important opportunities for communicating and personalizing messages with visitors based on usage patterns and choices visitors make as they use the site. From a traffic-building perspective, it is important to spot the broad patterns of usage and to determine the cost and productivity of the visits. Figure 8.8 shows a “wine glass” plot, a useful web site visualization tool. It presents the Customer Funnel ìWine Gl ass” Plot average visit pattern of web site visitors, either Categorization: Attracting Wrong People for all visitors or for a target demographic seg5 ment of customers such as women. Starting with arrival, it shows the proportion of visits that exit 4 1.00 at each stage. An example would be home page only, additional content pages, completed regis3 0.12 tration, access a shopping cart, and complete an order. The shape of the “wine glass” helps spot 2 0.05 important trends and problems. For example, panel (a) has very steep attrition at the earliest 1 0.02 steps. This normally occurs when the traffic building is attracting the wrong type of visitors, 0 0.00 -1.0 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0. 0 0 .1 0. 2 0 .3 0. 4 0 .5 0. 6 0 .7 0. 8 0 .9 1. 0 who quickly realize this and abandon the site. Conversion Rate From Home Page Visitors Panel (b) has a more gradual loss of visitors, but has a very low conversion at the last step. This is much more suggestive of problems with pricing, the costs of shipping, or some other problem Customer Funnel “Wine Glass” Plot with payment and checkout mechanisms. Categorization: Good persuasion, Poor conversion. 5 Building a wine glass plot requires tracking visits from start to exit, selecting the records of 4 users of interest, grouping exit pages into 1.00 sequential stages, and sending the data to a pro3 gram for drawing and analyzing the results (such 0.92 as with a spreadsheet). As seen in Chapter 7, 2 there are many situations where visitors come to 0.52 a site without sufficient identifying information 1 to categorize them into segments. In those situ0.03 ations, the wine plot must be for the full popu0.02 0 lation. The identity of the sequential changes -1.0 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0. 0 0 .1 0. 2 0 .3 0. 4 0 .5 0. 6 0 .7 0. 8 0 .9 1. 0 will change depending on the goals of the Conversion Rate From Home Page Visitors traffic-building plan. For commerce-oriented Figure 8.8 goals, the key steps will focus on shopping and conversion behavior. For branding or persuasionSummarizing Visit Shapes with a Wine Glass Plot oriented goals, stages will focus more on the Source: Concept from Sterne, author data. content selection and actions taken with regard

Steps Remaining

Steps Remaining

Chapter 8:

Traffic Building

259

to the brand or cause. Many tools exist to drill down and further analyze web visits. Figure 8.9 demonstrates a “starfield” tool, useful for spotting problems of over and underexposure of individual product items or desirable actions.15 The size of each box represents its sales importance. In Figure 8.9, showing online sales of IBM laptops, the width of a rectangle is the product price and the height is product margin. For products in the bottom right, many visitors see the product but fail to click through. In the upper left, there is high conversion but relatively low exposure. These forms of visualization help identify outliers, including under- and over-achieving products. More detailed information can be inferred from statistical analysis of clickstream data. These formal models allow much more careful explanation of the shape and trends of visits. Specific controls can be used for pricing, advertising exposures, the impact of repeat visits, and a variety of other marketing tools.16 A combination of visualization tools and rigorous quantitative analysis of web log data can explain the patterns of web site usage.

> Cost-per-Action
Wine glass plots demonstrate the steady attrition between web visit stages. One of the most useful guides for traffic-building analysis is to compute the cost-per-action, defined as the number of visits that make it to a particular stage (the “action”) divided by the campaign costs. Useful measures include the cost per impression, the cost per visit, or the cost per even-

Figure 8.9
Micro-View of Visit Behavior for an Online Retailer
Source: Figure 2 in J. Lee, M. Podlaseck, E. Schonberg, and R. Hoch, “Visualization and Analysis of Clickstream Data of Online Stores for Understanding Web Merchandizing,” Data Mining and Knowledge Discovery 5, Issue 1–2, (January–April 2001): 59–84. Copyright © 2001 Discovery. Reprinted with kind permission from Springer Science and Business Media.

260

Part 2:

Essential Skills

tual purchase. Figure 8.10 shows results for a relatively productive ad banner campaign. The pricing of the campaign immediately gives the cost per impression. Typically quoted as CPM, or cost per thousand (e.g. mille), the cost per impression in 8.10 is $0.015. While 1.5 cents per impression may seem cheap, banners typically have a low click-through rate. The next stage is a web site visit. If the click-through rate is 2% (a relatively high rate), this gives a cost per visit to the advertiser’s web site of $.015/.02 = $0.75. In other words, the average cost of generating a web site visit is 75 cents. Cost-per-action escalates at each subsequent stage, as the falling success rate appears in the denominator of the cost equation. If the action desired is a purchase, and the conversion rate from web site visit to purchase is 5%, then the cost-per-purchase is $0.015/(.02*.05) = $15.00 per purchase. Cost-per-action is a simplification of the full web chain analysis presented in Chapter 5. While a web chain analysis tracks all eventual possibilities and evaluates the campaign based on its full range of results, a cost-per-action analysis only focuses on the benefits for one specific desired outcome. When the focus of a campaign is primarily on traffic building for a specific goal, the difference will not be large. When the campaign and the tools used have a considerable branding impact, the difference may be significant. A cost-per-action analysis has the benefit of

From Visits to Purchases 0.05

From Impressions to Clicks 0.02

Pool of Visitors

0.95 Cost per Impression: Pool of Impressions Cost per Visitor @ 2% click-thro ug h: $15 per 1000 impressions $.015 each. $.015/.02 = $.75 per visitor.

Cost per Purchases @5% conversion from visitors to b u yers: $.75/.05 = $15.00 per purchaser. 0.98

Figure 8.10
Visitors Decline and Costs Rise at Each Stage

Chapter 8:

Traffic Building

261

simplicity, creating a method of quickly comparing alternative traffic-building campaigns. It may understate the full benefits of the campaign when multiple stages and different benefits result from an online ad or web site visit. The ease of determining campaign costs and campaign results varies greatly between the different traffic approaches illustrated in Figure 8.5. It is difficult to partition the relevant costs for a branding or publicity campaign, and to connect them to the number of visitors arriving at a site. Other sources of traffic, especially search and banners, are much easier to measure and evaluate. Campaigns that permit a cost-per-action calculation lend themselves to much more scientific management, with optimization tools and experimentation leading to substantial lowering of the cost-per-action.

> Web Visibility and Competitive Analysis
One of the notable aspects of online marketing is the number of available tools to monitor and improve a web page or web site’s visibility and online presence. Many of these are a free and open source, allowing inspection of exactly how they operate and modifications for special situations. These tools provide diagnostics of a site's online presence and help evaluate the impact of advertising, publicity efforts, or some news story. External links to a site have many benefits. They are crucial ingredients determining how high in the search return list a site appears, as well as a source of web traffic from web surfers following links on other web sites. Tracking the number and location of sites linking to a particular page provides a manager with a highly useful measure of the strength and nature of their online visibility. The major search engines provide “hooks” into their data, which allow programs to access and report link information. In the sites indexed by Yahoo!, for example, there are 12,200 links to Club Med.17 A Club Med manager can also capture these links into a spreadsheet, and analyze which sites are heavy linkers and where their coverage is sparse. It is also useful to compare this coverage against other benchmark sites. Some product categories lend themselves to external links much more than others. One method of handling this is to focus less on the absolute number of external links and more on the relative level of visibility within a set of competitors. Benchmarking on competitors can highlight strengths and weaknesses, while ensuring that the comparisons are relevant. Services such as Alexa allow a web site operator to compare their viewership performance against other web sites. Gathering data from the 10 million users with the Alexa button Figure 8.11 bar,18 the system provides daily traffic ranking of site popularity. For examUsing the Alexa Service to Measure Site Usage ple, Figure 8.11 shows that Honda Source: alexa.com (brown line) and Toyota (red line) rank

262

Part 2:

Essential Skills

consistently much higher than the Ford (blue) or GM (green) sites. Both Ford and GM should be concerned about the low viewership their sites receive compared to Toyota and Honda. This is even more serious if web usage reflects the underlying strength of the brands. A traffic plan might want to focus on both raising site traffic and determining what is causing such low relative performance. Traffic services can also help spot successful promotions and unusual dips that raise concern. Honda’s September boost is especially notable and provides important information to both Honda and Honda's competitors. Club Med gets much less traffic than auto sites, averaging around 30,000th on the Alexa list. The number 1 site is Yahoo!, followed by MSN and Google. The system also allows rankings by browser language used. For example, the top two ranked sites in Arabic are Google and Al Jazeera. In Chinese, the top three are sina.com.cn, baidu.com, and sohu.com. Chapter 16 covers some of the issues in different types of panels and sample issues they raise.

Traffic-Building Goals
Managers should decide on the goals of a campaign prior to its start. This is good advice for several reasons. One, it helps prevent ex post justifications. There is a natural temptation to focus on the positive results of a campaign when results fall short of the target. A decision to focus on a welldefined objective works against this. Second, it is impossible to effectively test and experiment with different traffic sources or campaign designs without an agreed upon metric of success. Third, it is much easier to contract with outside providers when traffic objectives are well established. At one level, any traffic-building campaign has the simple goal of attracting the best traffic to the site for the least cost. This is true whether the means are mass media, search engines, external alliances, or some other method. The controversy surrounds the most appropriate and actionable definition of the idea of “best traffic.” While maximizing profit is the logical goal for profit-oriented firms, it may be hard to operationalize and may not fit for nonprofit groups. It may be easier to specify and measure alternative goals, such as cost-per-action or the total number of actions accomplished through the traffic campaign. Setting traffic-building goals is complicated by the relative importance of brand building and offline impacts versus online actions such as registration or first-purchase (Figure 8.12). The web chain analysis of Chapter 5 captures all of these effects into a unified value of an impression and a unified value of a visit. When brand impacts matter, or when a visit to a web site leads

Best Traffic with Least cost

Maximize Profit
(long term or short term)

Minimize Cost-per-Action
(subject to volume targets)

Maximize Actions
(subject to spending cap)

Figure 8.12
Possible Traffic-Building Goals

Chapter 8:

Traffic Building

263

some consumers to make beneficial product purchases offline, these should be accounted for when setting traffic-building goals. With the push toward online spending accountability, there is a natural tendency to ignore them. Each of these three approaches has benefits and drawbacks. All three are used by different organizations. For many firms Internet spending has moved beyond experimental money and a lack of accountability, and campaigns are managed for results.

> Maximize Profit
A web traffic campaign with profit maximization as its goal has no ex ante spending caps or volume targets. Rather, it is willing to spend on media and traffic as long as the incremental benefits of the traffic exceed the incremental costs. While for many advertising settings, such as television or radio advertising, this guideline can be very challenging to measure and implement, online marketers can often measure the consequences of campaigns they run. This permits traffic building based soundly on profit considerations. Chapter 5 presented the two essential tools to manage traffic campaigns based on profit maximization. These are customer lifetime value and the unified visit value computed using web chain analysis. Customer lifetime value is the discounted net present value of the expected profits from a new customer. The unified visit value is the full expected benefits arising from a customer’s visit to a web site, which includes any branding and purchase impacts. For new and growing firms, the probability of acquiring a new customer multiplied by customer lifetime value is the largest impact of the unified visit value. For more established companies, the largest impact may come through longer term branding. A traffic-building campaign is not the same thing as an online advertising campaign. Figure 8.13 shows a case where they differ. Far more visitors to the iVillage Health and Beauty page will see the two McDonald’s salad ads than ever click through to the McDonald’s site. The ad is prominent and well designed, and it is likely to lead to both awareness and recall without a visit to the site. Judging this ad campaign solely on traffic building to the web site would be inappropriate. Its ultimate goal is to drive traffic to McDonald’s outlets for lunch or dinner, primarily through longer-term brand impact. For simplicity, the branding impact of the ads will be ignored in most of what follows. The emphasis is on building traffics to web sites. In cases like Figure 8.13, however, the ad brand impact may be the most important part of the ad campaign. Focusing only on actions that occur upon arrival at the web site, profit maximization provides the following guideline:

Traffic-building Profit Guideline (web site branding and other impacts included, ad branding impacts ignored)
> Spend on traffic sources that maximize the difference between unified visit value and cost per visit. > Acquire traffic as long as the cost per visit is less than the unified visit value. Traffic campaigns that produce a cost per visit lower than the unified visit value are profitable. When the difference is large, the campaign is more profitable. Some marketers may wish to simplify the web chain analysis and only consider new customers as a valid result. This assumes that the site branding and other impacts do not occur (have zero probability) or have no value (e.g., zero web site brand impact). This can be a useful simplification if there is widespread agreement on customer lifetime value and acquisition cost measures and the other impacts are felt to be minor.

264

Part 2:

Essential Skills

Traffic-building Profit Guideline (branding and other impacts ignored)
> Spend on traffic sources that maximize the difference between customer lifetime value and online customer acquisition cost. > Acquire traffic as long as online customer value exceeds online customer acquisition cost. The first part of the guideline is another way of saying a traffic campaign should focus on maximizing customer equity, typically defined as the difference between customer lifetime value and customer acquisition cost.19 The second part states that traffic-building efforts are profitable over the long term as long as customers acquired from the activity bring with them positive customer equity. It may be easier for some companies to measure customer acquisition cost and manage their traffic building based on that. This is a simplification that assumes brand building is minor and customers produce roughly the same profits after acquisition. Companies and researchers routinely compute customer acquisition cost. For example, a research paper computed customer acquisition costs for Amazon, eBay, Ameritrade, and E*Trade over a five-year period. As seen in Table 8.3, they range from a low of $7.70 to a high of nearly $400.20 As customer lifetime values for Amazon and eBay were likely to be higher than $10 over this period, these activities look profitable. The threshold for the brokerages is much steeper. Care needs to be taken when using historical averages. The proper customer acquisition cost and customer lifetime value to use in calculations is forward looking, and historical values may be misleading. Even if E*Trade’s acquisition costs were very high during 1997–2002, their acquisition costs going forward could be much lower.

(a)

(b)

Figure 8.13
Online Ads Can Contain Branding and Traffic-Building Impacts
Source: 8.13a The screenshot www.ivillage.com made available courtesy of iVillage, Inc. © 2005 iVillage, Inc. All rights reserved. iVillage and the iVillage logo are trademarks of iVillage, Inc.; 8.13b Used with permission from McDonald’s Corporation

Chapter 8:

Traffic Building

265

Table 8.3
The Wide Variation in Average Customer Acquisition Costs for Four Net Companies Company Ecommerce companies Amazon.com eBay 3/1997->3/2002 12/1996->3/2002 33,800,000 46,100,000 $7.70 $11.26 Time Period Acquired Customers Customer Acq. Cost

Online brokerages Ameritrade E*Trade 9/1997->3/2002 12/1997->3/2002 1,877,000 4,117,370 $203.44 $391.00

Source: Sunil Gupta, Donald R. Lehmann, and Jennifer Ames Stuart, “Valuing Customers,” Journal of Marketing Research 41, no. 1 (February 2004). Used with permission by the American Marketing Association.

> Minimize Cost-per-Action
Some organizations prefer to manage traffic building with pre-set limits on the total activity rather than the open-ended approach implicit in profit maximization. One popular approach is to set volume targets for some action, and then seek traffic-building approaches that achieve those goals with minimum cost. Possible volume targets are: > > > > A certain number of unique new visitors, A certain number of new registered users providing email addresses, A certain number of visitors inquiring about an upcoming new product, A certain number of new customers.

One of the virtues of this approach is its ability to handle more than one objective. A campaign may seek the cheapest solution for generating (for instance) 100,000 new visitors and 3,000 new customers. The goal of the campaign planner is to identify the best mixture of traffic-building tools that accomplish these goals. Marketers should be careful, however, that volume goals are not a substitute for thinking carefully about profitability.

> Maximize Actions
Another popular goal is to maximize the number of visitors completing some action subject to a fixed budget. The marketing department may be instructed to generate as many new customers as possible with an online campaign budget of $75,000. Again, it is possible to use this approach for multiple goals or attempt to reach some balance of new visitors, registered users, and new customers. A goal of maximizing new registrations (for example) requires some form of budget cap to prevent acquisition costs from spiraling out of control. If the goal is simply to bring in as many new registrants as could potentially use the service, more and more advertising will chase ever more difficult to acquire holdouts.

> Tracking Multiple Goals
It is relatively straightforward for a marketer to simultaneously monitor multiple goals and results from traffic-building campaigns. Web hosting and web site tracking firms create management

266

Part 2:

Essential Skills

“dashboards” that present results from campaigns on the daily or even hourly basis. Warnings and alerts appear when there is unusual campaign performance, both on the high and low side. Visualization tools increasingly put detailed results in the hands of managers, with intuitive yet real-time connections to live data provided through the Net. These dashboards may also track competitive activity, as online changes may be a leading indicator of changes in competitors’ product or pricing plans. The next section considers search engine marketing, both for organic traffic and paid advertising. Search marketing provides some of the most accountable traffic sources, and has grown rapidly during the past few years.

SEARCH ENGINE MARKETING
Table 8.2 shows how search engine marketing is a vibrant multi-billion-dollar business.21 The $5.1 billion spent on search in 2005 was 41% of all online advertising.22 This spending spanned the range from individuals dedicating a few dollars to promoting their personal web sites to major corporations spending millions. Almost any web site can participate, overhead is low, and entry is easy.23 Regional and local customization is straightforward as well. Search engine traffic comes in two basic forms. First is organic traffic arising from a search return list, automatically generated by the search engine without any commercial arrangement. Actions taken to boost this traffic are typically referred to as search engine optimization. The other basic form is pay-per-click. The top two keyword search services are Google and Yahoo!, but all major search engines offer some form of the service.24 Organizations bid on keyword portfolios. Depending on their bid and the bids of others, their ads appear higher or lower on the search returns page. Actions taken to drive this traffic are commonly referred to as keyword management. Search marketing and keyword analysis provide a highly useful benchmark against which all other traffic-building activities can be judged. The pay-per-click nature of search marketing establishes a hard number for the cost of generating a visit to a web site. Other media can be judged against this cost, and the benefits of an average visit compared as well. Organizationally, the difference between the unified visit value and the cost of acquiring traffic provides a market test of the web site value. Once an organization decides on its traffic-building goals, it should turn attention toward search engine marketing. The next part considers both search engine optimization and pay-perclick advertising.

Site Optimization > The Power of “Above the Fold”
Newspaper editors have long put their big stories and top news “above the fold.” Prominent items naturally attract attention (Figure 8.14). A similar phenomenon holds for search. Searchers’ willingness to investigate and respond to the search engine return list falls off steeply as an item falls further down the list. This appears as a > A much higher click rate on top-rated search items, > A decay of willingness to look down the list of search returns, and > A low fraction of searchers going beyond the first page or two of returned items.

Chapter 8:

Traffic Building

267

This behavior would not be so troublesome if the Web was small, and normal searches didn’t return thousands or even millions of hits. Googlefight.com provides an amusing example of how normal searches provide many items. Googlefight uses the number of pages found for each of two searches as the score in a contest, complete with animations of their battle. Celebrity names are one common contest, as are sports teams. For example, a “fight” between the terms Boston Red Sox and New York Yankees leads to a Figure 8.14 narrow win for the Yankees. What makes for a good game makes for Top Billing Gets Notice difficult traffic generation for sites Source: © FPG/Getty Images beyond the first few hits. Top placement in search returns is a key to generating organic traffic. Search engine optimizers seek to improve search engine placement by making the main search engines aware of their site, ensuring that content creates the appropriate keywords and details that search engines utilize in their rankings, ensuring that site navigation reinforces search engine algorithms, and cultivating links from other appropriate sites. They provide site guidelines that raise visibility without sacrificing usability and online quality.

> Site Guidelines
> Search Engine Awareness
Any serious site will probably be indexed eventually by the main search engines. Placement services accelerate this process by providing the search engines with addresses and information about the site. Table 8.4 highlights the lags involved in being listed in the main engines. Once part of its listing, each search engine maintains the freshness of its index by occasionally revisiting pages. The importance the search engine gives a page or site partially determines how frequently it is revisited.

> Use Meta-Tags to Flag Important Concepts
Search engines look to the content of the page to determine relevance. Not all content is treated equally. Most of the leading engines weight the page title and keywords listed on the page above words appearing elsewhere in the text. Web designers can assist search engines finding their page with proper emphasis and use of metatags. Kazanlaw.com demonstrates how good design yields highly valuable traffic through organic search results. Kazanlaw.com ranked in the top three among the organic search results on Google for the high cost phrase “peritoneal mesothelioma” over the course of several months.25 Rather than pay for high price ads, this top ranking generates very valuable traffic without a fee. Top billing is partially due to the emphasis of the phrase within the title metatag, the keywords metatag, and the description metatag.26 Online tools let a marketer check how a search engine indexes a site. These tools simulate how a search “spider,” the indexing software used by search engines to read and collate Web

268

Part 2:

Essential Skills

Table 8.4
Search Engine Submission and Costs Search Engine Listing Tied to Submit? Submit Limit Pages Appear Overall Freshness AllTheWeb No No limit 2 to 4 weeks 1 day to 4 weeks AltaVista (Yahoo!) Yes No limit 4 to 6 weeks 1 day to 6 weeks + Google No, but may help No limit About 4 weeks 1 day to 4 weeks Inktomi n/a n/a About 4 weeks 1 day to 4 weeks Teoma n/a n/a n/a 1 day to 4 weeks

Source: SearchEngineWatch.com.

information, reads all the metatags and normal content. Several powerful versions are free and provide a valuable check on programming oversights and missing information. Among the items search engine spiders record are date of last modification, size of page, metatags, image, and textual descriptions.27 Web authors must decide which keywords and phrases merit inclusion. This requires an understanding of how consumers view the product or service, and which terms they will use in the search engine. In a health-related site, these keywords will naturally focus on the disease. In other cases the choices are more complex. There can also be legal challenges. In 1998 Playboy Enterprises sued Terri Welles for her use of the keywords “playmate” and “playboy playmate of the year 1981,” claiming that these were copyrighted terms owned by Playboy. The Court found that Welles could use the phrases, as they accurately described her past and were a fair use of the terms.28 Other jurisdictions, such as the European Union, have put stricter limits on the use of brand names within metatags. While a page can be changed at any time, the lags shown in Table 8.4 imply that keyword choices for search optimization must change relatively infrequently. High placement on multiple keywords is difficult. Firms must focus on the choices which best identify the page and stand out from the competition.

> Structure Web Site Content
The design of a site must satisfy goals of usability, clarity of information, ease of navigation, and entertainment. There are also traffic-building concerns. One of the difficult tradeoffs is between an exciting animation to welcome new viewers and a more informational and search friendly landing page. The site for fans of blues legend John Lee Hooker provides a good example. The first stop for surfers typing in www.johnleehooker.com is an animation providing photos and music, giving a feel for the music prior to landing on a more informational page at www.johnleehooker.com/home.asp.29 Search engines, however, do not understand this distinction. As the animation does not contain many of the tags and information content that leads to a high ranking, the site’s visibility suffers. The same type of concern arises with dynamic pages that change with new content.

Chapter 8:

Traffic Building

269

There are also influential aspects of site structure that shape the importance of a site or a page in algorithms used to determine site importance. The PageRank algorithm (see next page) of Google uses page links as a form of importance voting. Much of the importance of a site comes from external links, as described in step 5. Within a site, the structure of links also influences rankings. While somewhat technical, there are tools, tutorials, and consulting services that help web site managers structure links and navigation design to better reflect the true importance of pages within a site.

> Cultivate External Links
One of the keys to high search engine placement is the number and quality of external links that point at a site. Some search engines use these links as a fundamental feature determining the quality of a page (see description on next page). Even search engines not using links to calculate influence change their indexing behavior based on external links. A strong network of external links is at least a partial substitute for paid advertising. An example of a site placing high on many search returns for such popular topics as “stem cells” or “health information” is the National Institutes of Health (NIH). Both Alta Vista and Yahoo! show more than 5 million links to the NIH site. External links provide a snap shot of the connectedness of a web site and the influence it has on the rest of the Web. Each external link is also an opportunity for free traffic. When a site visitor sees a link on a page there is some probability of following it to the web site. Even a small chance on a high volume site can be a valuable source of new visits. External links also provide a variety of clues about fruitful sources of traffic and interest. The search term “link:ford.com” demonstrates that more than 61,000 pages have a link on them pointing toward some page on ford.com. Refining the term to “link: ford.com—site:ford.com” restricts attention to links originating outside of ford.com. Each of these 57,000 pages provides a potential source of traffic to Ford, and insight into companies and organizations that think Ford deserves tracking and monitoring. “Link farms” are one of the more dubious search engine optimization techniques. These web sites create massive numbers of links, and work out reciprocal links from other sites, in order to concentrate importance weighting through algorithms such as PageRank. Link farms often make their money by charging for establishing a link to a site attempting to move up their ranking. As search engines mature, they are penalizing both link farms and sites that join them for attempting to manipulate rankings with artificial connections.

Keyword Advertising
Site optimization and other sources of naturally occurring web site traffic are valuable sources of visits. Many marketers find that it profitable to go beyond these measures and advertise their sites. One of the dramatic online developments since 2000 has been the rapid growth of online keyword pay-per-click advertising. Small, clickable text-based ads, as shown in Figure 8.15, bring in billions of dollars of advertising revenue. Simple and straightforward, with little or no branding and a basic pitch, their performance is surprisingly effective. These ads steer traffic to an organization’s web site, with no money changing hands unless a click occurs. This section describes the process of keyword advertising, what determines a keyword’s value, and how marketers should decide how aggressively to pursue keyword advertising. Many useful online tools make the process easier and more powerful.

270

Part 2:

Essential Skills

PageRank for Marketers

Google established itself by producing surprisingly effective hits at the top of its search results. At the heart of this success is the PageRank algorithm, and a little knowledge of the Google approach clarifies both its success and the power of external links to a site. The PageRank algorithm treats links on a web page as a vote of influence. The goal of the PageRank algorithm is to take all the links (the votes) on all the pages of the Web and compute the importance ranking of each page. The rules of voting are somewhat unusual. The basics are: > Ballots aren’t secret. It matters what page is doing the linking. > You can’t vote for yourself. Links from Page 1 to Page 1 don’t count. Links to other pages within a web site do count. > You can vote for as many candidates as you want. Every page getting a link receives some benefit of the link. > No ballot stuffing. Repeated links only count once. > The order of listing on your ballot doesn’t matter. > Influence is contagious. A page receiving many votes is important, and casts important votes. The most unusual rules are number (3) and number (6). A page with only one link is casting more of its vote to its destination than a page linking to thousands of sites. The

1

3

2

4

Links are the key to Google’s PageRank

PageRank algorithm spreads influence evenly across all outbound links on a page. How much power a page has works through rule (6). The PageRank algorithm is surprisingly effective. At least partially, it captures the entire influence of all the link information of the entire Web. Second, it does this in a way which is mathematically solvable for billions of web pages. Third, it is relatively immune to manipulation. Site navigation and link structure concentrates PageRank influence. While an organization can only partly affect external links, it can make design and layout decisions which raise a key page’s influence. One of the simplest and most important actions is creating links to the home page on all of the internal pages, reinforcing home page value.
Sources: Sergei Brin and Larry Page, (1998), “The Anatomy of a Large-Scale Hypertextual Web Search Engine,” WWW7/Computer Networks. Kurt Bryan and Tanya Leise, (2005), “The $20,000,000,000 Eigenvector,” working paper, January 26.

Chapter 8:

Traffic Building

271

Figure 8.15
Ads for Keyword Phrase “GM Fuel Pump”
Source: Reproduced with permission of Yahoo! Inc. © 2005 by Yahoo! Inc. YAHOO! and the YAHOO! logo are trademarks of Yahoo! Inc.

> Keyword Value
Table 8.5 shows average prices of keywords for major sectors of the economy, compiled by sampling thousands of keywords on a continuous basis.30 In the automotive sector, the average price per click runs about $1.50, while finance mortgage ads averaged $4.52 a click in May 2005. Wall Street and investors seem to view these prices as important indicators of online activity, and there was a sizable impact on the stock prices of advertising companies following the release of the May dip in keyword prices.31 Keyword ads influence billions of web site visits.32 They also establish a benchmark for the value of traffic from all sources. Traffic is relatively cheap if it can be generated for less than the prevailing keyword price in that sector, and should be refocused if the price is higher than this rate. Traffic cost places a burden on web site managers and usability experts to ensure their sites are sufficiently productive to justify the visit cost. A few main factors determine keyword value. A keyword only has value if it leads some viewers to click and visit a site and some fraction of those visitors to convert to a desired action. A keyword ad’s value to a specific advertiser is driven by its productivity in leading to a desired action and the value of that action. For example, if the desired action is signing up a new customer, then purchase profitability and customer lifetime value strongly influence the value of the keyword. The connection between keyword ad and action value is possible due to the inherent traceability of keyword advertising. Each arrival to a site is accounted for, and that specific visitor is followed through the “traffic funnel” from visit beginning to visit end. While there are still some uncertainties about what happens offline or on later visits, this level of monitoring is much higher than for almost any other advertising medium.

272

Part 2:

Essential Skills

Table 8.5
Fathom Online’s May 2005 Keyword Price Index™ (KPI™)
Fathom Online Keyword Price Index™ Month Automotive ConsumerRetail ConsumerServices ConsumerTravel Hospitality $0.64 $0.85 $0.90 $0.97 $0.88 $0.81 $0.93 $0.92 $0.87 FinanceInvesting FinanceMortgage TelecomBroadband TelecomWireless Avg. of tracked industries $1.37 $1.55 $1.66 $1.70 $1.64 $1.60 $1.75 $1.95 $1.67

Sep-04 Oct-04 Nov-04 Dec-04 Jan-05 Feb-05 Mar-05 Apr-05 May-05

$1.54 $1.39 $1.35 $1.41 $1.34 $1.28 $1.42 $1.40 $1.50

$0.32 $0.48 $0.60 $0.58 $0.52 $0.46 $0.51 $0.52 $0.49

$0.54 $0.96 $1.27 $1.36 $1.29 $1.04 $1.08 $1.13 $1.09

$1.76 $1.60 $1.70 $1.76 $1.73 $1.65 $1.86 $2.03 $1.81

$3.17 $4.31 $4.74 $4.79 $4.93 $5.10 $5.39 $6.49 $4.52

$1.89 $1.78 $1.59 $1.63 $1.67 $1.72 $1.85 $2.09 $2.02

$1.09 $1.06 $1.09 $1.09 $0.79 $0.77 $0.95 $1.01 $1.00

Source: www.fathomonline.com

Keyword value is driven strongly by competition. Although the specifics differ between major search engines, keyword ads are priced by a continuous auction process. Even if a keyword is extremely valuable to an advertiser, if no other firms bid, its price will be low. However, if there are a number of firms competing for the keyword, the price can be bid up to levels that become unprofitable for some advertisers. Prices can change minute-by-minute, as new bidders enter and current bidders change their offers.

> Keyword Portfolio Evaluation
The competitive bidding process creates situations where seemingly close keywords can have very different prices, at least temporarily. Keyword advertisers should monitor activity and follow an iterative process of identification, expansion, testing, trimming, and tracking of campaigns. Identify main keywords –> Expand the list –> Test -> Bid –> Trim -> Track.

> Identify Possible Keywords
The best place for a company to start building a list of relevant keywords is by reviewing their own web logs. Web log software captures the search terms used by visitors arriving at the site when they click on search engine result items. These are the terms that current visitors generated on their own, and they provide a snapshot of keyword search activity connected to a desire to visit. Some software systems track these natural search terms on to further conversion steps, and high-performing keywords from this list are an especially useful starting point. Keyword length and alternative phrases are part of keyword list building. Phrases can be of increasing length, and most systems let an advertiser specify near or exact matches. For example, an online battery store selling parts for the Sony Vaio might use any of the following phrases. > 1-word: battery > 2-word: laptop battery

Chapter 8:

Traffic Building

273

(a) Keyword ad (Ex: Yahoo!)

Click through

(b) Destination web site (Ex: bizrate.com)

Conversion

(c) Desired Action (Ex: 1Aauto.com)

Figure 8.16
Keyword Initiated Traffic
Source: 8.16a Reproduced with permission of Yahoo! Inc. © 2005 by Yahoo! Inc. YAHOO! and the YAHOO! logo are trademarks of Yahoo! Inc.; 8.16b Courtesy of ©2005 Shopzilla, Inc.; 8.16c Courtesy of www.1AAuto.com

274

Part 2:

Essential Skills

> 3-word: Sony laptop battery > 4-word: Sony Vaio laptop battery > 5-word: Sony Vaio laptop TR3A battery Most consumer searches use one or two word phrases, with 39% using one word, 49% using two words, and 9% using three words. Fewer than 5% of searches use four or more words. A marketer may want to augment the keyword list with words or phrases that are especially important to a segment of customers. These can be specific brand names, specific part names, technical phrases, or some other special terms.

> Expand the List
Search providers have tools that help advertisers expand their search term list and find other possible keyword phrases. For example, the Adwords keyword tool suggests more than 100 additional specific keywords for the phrase “fuel pump,” such as “mustang fuel pump” or “chevy fuel pump.” There is also a long list of additional related topics, such as “gas filter” and “fuel injectors.” Depending on the testing strategy, the advertiser can start with a very broad list and trim down or start with a more focused approach and build up from high-performing terms.

> Testing Keywords
Keyword ads vary widely in click-through rates, conversion rates, and cost-per-action productivities. It is very useful to test the keyword list and collect some basic conversion rate and costper-action data prior to committing to a keyword plan. Search engines allow advertisers to set daily traffic and budget limits. These permit budgeting for easy and low-risk testing of alternative keyword lists. Bidding and traffic estimation tools provided by major search providers are a good starting point. For example, the tool lets anyone see the bids placed by advertisers for the phrase “fuel pump.” As seen in Figure 8.17, a tester must type the code on the bottom as a protection against software agents constantly checking on prices and overloading the system. Thirty different advertisers bid on this specific phrase, with a range from $.06 to $.47 per click. The system only displays seven ads during an actual search.33 There is an extensive literature in statistics on the best strategies for running test markets and how much information to collect. For keyword advertising, the goal is to collect the type of data shown in Table 8.6. This is a list of keyword phrases with their results. Conversion rates, given by conversions/clicks, vary significantly by keyword. The tests should also reveal how

Figure 8.17
Yahoo! Bidding Tool with Human Verification Step
Source: Reproduced with permission of Yahoo! Inc. © 2005 by Yahoo! Inc. YAHOO! and the YAHOO! logo are trademarks of Yahoo! Inc.

Chapter 8:

Traffic Building

275

Table 8.6
Collecting Keyword Data Keyword Phrases Skylark fuel pump pos. 1 GM fuel pump pos. 2 Buick parts pos. 2 Used Buick Skylark pos. 1 Buick pos. 3 Used car parts pos. 2 Clicks 62 148 1246 8678 362 17265 Conversions 6 18 48 95 15 98 Cost per click to hold position $0.10 $0.20 $0.25 $0.23 $1.37 $0.34 Conversion Rate 9.68% 12.16% 3.85% 1.09% 4.14% 0.57%

productive different keyword positions are, and the prevailing price required to maintain any particular ad position if competitors’ prices remain fixed.

> Bid, Trim, and Track
How Search Engines Charge for Ads
The two main keyword-advertising sites, Yahoo! and Google, use two different rules to auction keyword ads. Yahoo! ranks ads according to maximum bid, while Google ranks ads by maximum bid times click-through rate. There are arguments in favor of each approach. Studies suggest that the Google system is somewhat more profitable for the search engine, as better clickthrough rate ads get “promoted” higher in the list. As the search engine receives money only for click-through, this tends to augment their income by putting low performing ads in less desirable positions. On the other hand, the Yahoo! system is simpler and more transparent. Yahoo! makes public the bids for all keywords, allowing an advertiser to understand what the cost would be to acquire any particular position in the bid list. A keyword bid sets the maximum price an advertiser is willing to pay for a specific keyword. The actual price paid, and the position on the list, depends on other bids from other advertisers. For an example, assume there are six bidders and a minimum bid of $.10. The system first sorts the bids, from highest to lowest: (.95, .82, .56, .35, .33, .13). The system “works from the bottom.” The bottom bidder pays 0.10 and gets the last slot. For an advertiser to show in fifth place instead of sixth it must bid higher than the sixth place maximum bid. Thus, the price of the fifth place is $.14. This continues up to the top. The bids (.95, .82, .56, .35, .33, .13) produce prices paid for six slots of (.83, .57, .36, .34, .14, .10). The system on Google requires knowledge of estimated click-through rates for each ad. Ranking is no longer based solely on maximum bid, but is also based on the maximum bid time of the estimated click-through rate. Key similarities of the two systems are: > > > > > No payment is required until someone clicks through to the web site, Higher bids increase the possibility of a higher keyword position, No clicks will cost more than the maximum bid, Actual price charged depends on the intensity of competition, and The cost-per-click in the marketplace establishes a benchmark for traffic-building activities.

276

Part 2:

Essential Skills

Proper Bidding
A firm placing a keyword ad on Yahoo!, Google, or some other keyword-advertising site must decide which keywords to bid on and how much to bid on each keyword. In order to do this properly, it is necessary to specify the traffic-building goal(s) and to understand keyword specific clickthrough rates and ultimate conversions to the desired actions. Several assumptions make the discussion much simpler. When these assumptions do not hold, the proper rules for keyword bidding are more complicated and more of a trial and error approach may be necessary for many firms. Assume that: > > > > The desired action is new customer acquisition, and all customers have the same dollar value V. Competitors’ bids are known and they do not change in response to the firm’s bids. Ads are ranked by maximum bid (the “Yahoo! system”). We have test market information showing click-through rates and conversion rates for each keyword in each ad position.

These assumptions are obviously very strong,34 and do not hold in many situations. The real goal is to understand the logic of proper bidding before tackling these complexities. The value V in assumption 1 can be customer lifetime value. It might also be the customer acquisition cost through the next best alternative if this is less than lifetime value. The solution works in two steps. Step 1 is to calculate the best ad position for each keyword. Once that is determined for each keyword, Step 2 computes which of these keywords to bid for.

Stage 1: Choosing the Proper Position for each Keyword
Stage 1 is a calculation for each keyword separately. It does not imply that the firm will use this keyword, but provides the necessary information to complete stage 2. If no one else is bidding on a keyword, the firm automatically gets position 1 for the minimum allowed bid. This is currently a dime for Yahoo! in the United States, but in the United Kingdom it is ten pence (or almost twice as high with prevailing exchange rates). If there are competitors, the firm observes competitor bids (assumption 2). As in Figure 8.18, it can compute from these bids the lowest bid it can make to obtain position 1. Likewise, it can compute the smallest bid that gets position 2, and so on for each of the positions. List these as (bk(1), bk(2), ...., bk(n)) for the n spots for keyword k. The conversion rate from visit to Actual Price action for traffic following keyword ad k in position j is 0.83 CRk ( j ). The click-through rate from ad to visit is CTRk ( j ). For each of these n spots, compute the expected 0.57 cost-per-action for that keyword:

Max Bids 0.95 0.82 0.56 0.35 0.33 0.13 Figure 8.18

Yahoo! system

0.36 (.10 minimum) 0.34 0.14 0.10

CPAk(1) ϭ bk(1) Ϭ CRk(1), CPAk(2) ϭ bk(2) Ϭ CRk(2), … CPAk(n) ϭ bk(n) Ϭ CRk(n). Only one position is possible per keyword. It is not right to choose the lowest cost-per-action ad position. Both top and lower positions may be profitable, but top positions may generate many more clicks than lower ranking ads.

From Max Bids to Price

Chapter 8:

Traffic Building

277

What is necessary is to calculate the expected profit per search for the keyword ad in each position: Expected Profit per Searchk (1) ϭ (V Ϫ CPAk(1)) ϫ CTRk (1) ϫ CRk (1), Expected Profit per Searchk (2) ϭ (V Ϫ CPAk(2)) ϫ CTRk (2) ϫ CRk (2), … Expected Profit per Searchk (n) ϭ (V Ϫ CPAk(n)) ϫ CTRk (n) ϫ CRk (n). The position that gives the highest expected profit per search is the best position to bid for. This can result from having an especially low cost-per-action, but it can also result from having especially high conversion rates even if the cost-per-action is somewhat higher. Stage 1 rule: Choose the ad position with the highest expected profit per search for that keyword. If this is position j, send CPAk( j ) to stage 2. The solution in Stage 1 almost solves the keyword bidding problem. If the desired position is j for keyword k, then the right bid is either 0 or bk( j ). Stage 2 decides which of these two alternatives is correct for each keyword. In words, either the firm does not bid for keyword k or it bids just enough to win position j.

Stage 2: Choosing which Keywords to Buy
With Stage 1 solved for each keyword, the problem is almost solved. The Stage 2 rule is actually simpler than Stage 1. Rank all of the different keywords by cost-per-action for the chosen position for that keyword. Buy all keywords with cost-per-action lower than V. Stage 2 rule: Bid bk( j ) for all keywords with CPAk( j ) < V, do not bid on the keyword otherwise. Tables 8.7 and 8.8 illustrate stages 1 and 2. For the keyword “fuel pump,” the data in yellow reflects the test results. For this keyword position 3 is the best location, which is not obvious from either the click-through rate or the cost-per-action. Position 3 wins by combining a low cost-per-action and sufficient volume to generate substantial benefits. Table 8.8 shows the profitable keywords. The top three have cost-per-action below the Max CPA, the other three are above. Proper keyword bids are just the fee needed to hold that desired position for the profitable keywords, and zero otherwise. The Stage I and Stage II logic works for both the Yahoo! and Google system, although for the Google system the cost-per-action formula and bid position assignment details change. Online tools help bidders make these adjustments.

> Complications
Competitive Responses
One of the biggest complications to the bidding process is the competitive struggle for desirable keyword positions. The advertiser previously in position 3 for fuel pump will find its ad in position 4. This might be fine. It may also lead to an increased bid high enough to regain position 3. Finding the right solution to the competitive bidding process is an area of active research. Unpublished results suggest there are multiple possible outcomes, and search engines can affect this competitive advertising battle by slightly altering their allocation rules.35 If competitors can respond, there may be competitive reasons and search engine rules that might make the best bid

278

Part 2:

Essential Skills

Table 8.7
Stage One for Keyword “Fuel Pump” V Keyword Phrases: fuel pump Position 1 Position 2 Position 3 Position 4 Position 5 Position 6 $15.35 Clicks Conversions Cost per click to hold position $0.46 $0.38 $0.35 $0.29 $0.22 $0.21 Test impressions 10,000 10,000 10,000 5,000 5,000 5,000 CTR Conversion Rate 3.53% 3.05% 3.65% 2.85% 2.53% 3.70% CPA by position $13.02 $12.44 $9.59 $10.17 $8.69 $5.67 Expected Profit per Search $0.0067 $0.0055 $0.0127 $0.0124 $0.0053 $0.0116

821 622 603 421 158 162

29 19 22 12 4 6

8.21% 6.22% 6.03% 8.42% 3.16% 3.24%

Table 8.8
Stage Two for Keyword Choice Max CPA Keyword Phrases skylark fuel pump pos. 1 gm fuel pump pos. 2 fuel pump pos. 3 used buick skylark pos. 1 buick pos. 3 used car parts pos. 2 15.35 Clicks 62 148 603 8678 362 17265 Conversions 6 18 22 95 15 98 Cost per click to hold position $0.10 $0.20 $0.35 $0.23 $1.37 $0.34 Conversion Rate 9.68% 12.16% 3.65% 1.09% 4.14% 0.57% CPA $1.03 $1.64 $9.59 $21.01 $33.06 $59.90 Keyword bid $0.10 $0.20 $0.35 -

for a keyword higher than just the minimum bid to obtain that position. For example, some search engines break ties between equal bids based on which firm was the first in time to raise the bid to that level. Establishing priority might lead to an “overbid.” Practitioners are used to operating without complete guidance and simple rules. Competitive responses force advertisers to actively manage bidding and keyword position choice in an iterative process. After a period of back and forth caused by a new bidder entering the fray, prices will normally settle down to a new set of values. Equilibrium occurs when all the bidders run the stage one and two procedures and no firm wants to change their keyword bids.36

Chapter 8:

Traffic Building

279

Missing Test Data
The Stage I and II process relies on a firm knowing the click-through and conversion rates of the keywords in question. This data is only partially available, and advertisers must actively choose to collect that information. This can be further complicated if competitors are changing bids during data collection. The various search engines provide bidding tools to make this estimation easier and less risky. Better data lets an advertiser compete with more assurance. It is likely that the major search firms will continue to improve their assistance to advertisers and make it easier for a firm to understand the productivity of new keywords and different keyword ad positions.

Value per Action Varies by Customer Segment
One of the most important and thorniest complications occurs when customers with different profitability are present in the market and keyword choice partially determines which type of customer responds. This is also an active area of research. It can affect the Stage I and II decisions. These are not the only complications. Despite these caveats and details, the online keyword ad business provides marketers with a jump in accountability over many other advertising media and is changing the customer acquisition process.

TRAFFIC BY ASSOCIATION
Classic advertising works by association, where marketers put a company’s message in front of consumers where they spend their leisure time, where they work, and where they live. Even search advertising works by putting clickable ads near results generated during information collection. Other traffic-building activities shown in Figure 8.5 work by putting the appropriate banner or link in front of individuals where they spend their time, both on and offline. The full range of traffic-building opportunities and strategies is too wide to cover within one chapter. Chapter 6 focuses on branding, and the strong connection between domain name, brand equity, and getting people to visit a site. Repeat visits and word-of-mouth work by encouraging an existing customer to revisit a site and to spread the word. Chapter 10 covers these, focusing on loyalty-building campaigns online and through email. Chapters 13 and 14 consider retailers and manufacturers use of affiliate networks to drive shoppers to sites. Chapter 16 continues the discussion of online banner advertising, looking at how advertising campaigns find appropriate venues to place banner ads in ways that match desired demographics of online users.

Banner Ads
Banner ads were the initial online advertising format. Introduced in 1994,37 online revenue from banners grew rapidly during the 1990s, fell sharply in 2000–2002, and is once more experiencing growth.38 Part of this is a rebound in advertising generally, part is the growth and expansion of households with broadband, and part is the development of new formats.

280

Part 2:

Essential Skills

Even after 10 years of online advertising development, there con+17% Non-standard tinues to be format innovation and 60 –49% Button #2 changes in the size and shape of the 50 –4% Skyscraper most popular online ads. Figure 8.19 40 shows the rapid growth of “Leader –51% Half banner Board” ads during 2003–2004.39 –78% 30 Micro bar These are a larger format (728 × 90) +5% Button #1 20 than the originally more popular full +19% Square button 10 banner (468 × 60) they are displac–50% Vertical banner ing. Figure 8.20 illustrates the two. 0 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 In addition, Leader Board ads often ‘03 ‘03 ‘03 ‘03 ‘04 ‘04 ‘04 ‘04 use animation tools such as Flash to make them more eye-catching and 70 –552% Leader Board interactive. +281% Med. rectangle 60 Advertisers using online ban+210% Wide skyscraper ners typically pay by impression, 50 +26% Rectangle although large advertisers can 40 sometimes negotiate pricing that +15% Large rectangle 30 blends impressions, click-through, -75% Square (pops) and conversions into the pricing 20 –44% Vert. rectangle formula. Thus, online banner ads 10 have some features of more tradi0 tional impression-based advertising Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 as well as performance-based key‘03 ‘03 ‘03 ‘03 ‘04 ‘04 ‘04 ‘04 word advertising. Figure 8.19 In addition to pricing and graphical layout, a major difference Relative Importance of Different Ad Banner Formats between online banners and keySource: DoubleClick [Rick Bruner], "The Decade in Online Advertising: 1994–2004," DoubleClick word search ads is the context (April 2005): 19. where they appear. Banner advertising occurs primarily at major portals and publishing sites. Visitors are often engaged in consuming content at the site rather than searching for external information. This leads to low attention paid to banners and low clickthrough rates (Figure 8.21). There are also software tools that block banner ads from displaying in the browser, making click-through impossible. Although more sophisticated than search ads, it is still easy and relatively cheap to experiment with a wide range of forms and features within banners. This leads to rapid learning about the important factors causing higher click-through rates. Online advertisers have found higher click-through rates associated with:40
70 –50% Full banner

Ad Impressions in billions

Ad Impressions in billions

> > > > >

Bold colors Top of page placement Animation Call to action Limited frequency of exposure

Chapter 8:

Traffic Building

281

Figure 8.20
Leader Board Versus Full Banner Format
Source: © Terri Miller / E-Visual Communications Inc.

Despite improvements and innovations, the low banner click-through rates mean that banner ads need to be quite cheap in order to be cost effective. For example, according to data from DoubleClick.com, the average click-through rate for non-rich media ads during the third quarter of 2004 was .2%.41 For the same time period, Table 8.5 showed the prevailing click-through price for a search ad was approximately $1.50. For a price per impression to result in the same price per click, the CPM rate would have to be $3 (Per impression: X = .2%* $1.50 = .003, per thousand impressions = $3). The richer and more animated the ad, the more likely there is to be a branding impact as well as a traffic generating effect. In that case, evaluating the banner campaign should normally use a web chain analysis and not a simple cost-per-action comparison. The important insight is that comparisons need to be made at the same level of analysis, and this level of analysis should be able to capture all the relevant benefits of the advertising.

Click-Through R a te for Rich Me di a Interstitia l A ds* for Sn uggle Botanica l Bliss, by Cre a tive Forma t, 2003
Floating ad Commercial break GIF banner 0.11% 1.56% 4.39%

Note: *the two types of Eyeblaster-branded rich media interstitials used were a commercial break (a full-page ad that plays on site entry or before any other site page) and a floating ad (moves within transparent layer over Web page and plays within area up to 500x500 pixels) Source: Eyeblaster/Dynamic Logic, April 2003

Figure 8.21
Low Click-Through Rates for Banner Ads
Source: eMarketer.com

282

Part 2:

Essential Skills

Co-Branding and Placement > URL Placement: Bills, Machines, Bags, and Billboards
> Software Integration
Some of the most profitable traffic leads to cost reductions rather than new revenue. As shown in Chapter 7, there are many opportunities to save money with online customer and technical support. The primary goal of traffic building in support situations is convenience. If the address is available and online support accessible and easy to use, the online site will replace the more costly phone or sales force methods. Software companies develop this approach further than most. The proper technical support online addresses are integrated into manuals, warranty cards, and software programs. Microsoft has done this with auto update and help features in their applications and operating system. In fact, the help file in all of the Office applications automatically links into the Microsoft web page (Figure 8.22). Building the Net into products has numerous benefits. Microsoft’s activation policy works against piracy. Updates help control against security problems. Built-in help moves solutions to the cheapest channel to handle the problem. This is much better than requiring the user to find the address and manually type it into the browser.

>

Consumer Communications

Figure 8.22
Microsoft Office Integrates Links to its Online Support Site
Source: Microsoft product screenshot(s) reprinted with permission from Microsoft Corporation.

There are many opportunities for URL placement and inexpensive trafficbuilding publicity in standard consumer communications. These include shopping bags, billboards, monthly statements, catalogs, television commercials, and store signage. URLs do not have to be prominent to reach some in the audience, as seen in billboard ads used by 21st Century Insurance. With a complicated name, there is not natural parsing into a domain name. The URL on 21st Century Insurance’s billboards

Chapter 8:

Traffic Building

283

provide the translation to 21st.com. The company uses a standard placement of seat belt and domain name in the corner of a number of its billboards, whether promoting safe driving or creating an image of driving the Pacific coast. This makes the domain almost part of the logo, further making the domain name and brand connection.

> Sponsorship and Co-Branding
Specialized portals are sites acting as a focus of industry news, activities, events, and other content. These have emerged for a wide range of topics, such as health care, real estate, and even marketing. Nonprofit organizations sponsor some, while others are explicitly commercial. Important portals acquire a leadership role within the industry, perhaps through the amount of content, a “must attend” tradeshow, or leading industry publication. Visible links and sponsorships within these portals may be a cost-effective source of traffic. A regular column within the portal, links appearing to partner sites, or occasional stories with embedded site links all generate a certain number of clicks and visits. Sponsorship arrangements typically lack any volume guarantees. This is similar to corporate sponsorship of sporting and charity events, where the sponsor seeks goodwill and co-branding in a forum frequented by target customers leads to further visits. Online sponsorship has an added advantage of being only a click away from a visit. Bundled access to a site provides a form of digital sponsorship. Advertising Age provides access to the online site and content of its magazine Point for subscribers to its main site and publication. It uses its advertising portal site, full of relevant marketing content and news, as a magnet for individuals interested in marketing. Rich media, such as the pop-up film clip they use, is much more noticeable with more impact than a simple banner or link. The co-brand and sponsorship can further drive traffic to additional partner sites. All the sources of traffic in this chapter have the goal of cost-effectively driving traffic to the advertiser’s web site. The ability to track these conversions, to learn what works and what doesn’t, creates the opportunity for much more rapid learning than in traditional media. Traffic building is one of the best examples of the move to closed-loop marketing.

ENDNOTES
1. H.A. Simon. (1971). “Designing organizations for an information-rich world,” in M. Greenberger (ed.) Computers, Communications, and the Public Interest. Baltimore: Johns Hopkins Press. pp. 37–52. 2. David Mamet. Glengarry Glen Ross. 1992. 3. Carl Bialik. (2004). “Lawyers Bid Up Value of WebSearch Ads,” Wall Street Journal, Eastern Edition. New York, April 8, B.1. 4. Pew Internet Project. (2005). Health Information Online. April. 5. Google has a slightly more complicated rule. Bid * click-through rate. 6. As of February 2005. This reinforces the factors discussed in Chapter 7. 7. The Diamond and Water paradox dates at least to Adam Smith’s Wealth of Nations. “Nothing is more useful than water: but it will purchase scarce anything; scarce anything can be had in exchange for it. A diamond, on the contrary, has scarce any value in use; but a very great quantity of other goods may frequently be had in exchange for it.” Book 1, Chapter IV. 8. In the language of microeconomics, the initial utility of water is much higher than for diamonds but it falls rapidly. The abundant supply makes the price of water nearly free, while the restricted supply of diamonds makes for a steep price. 9. Web pages are calculated as the number of web pages indexed by the largest search engine, as reported by SearchEngineWatch.com. This is known to be an undercount of the total number of pages online, although it is arguably the right measure for a user relying on the particular search engine. During this time interval the fraction of these pages in English appears to have remained roughly constant at about 72% of the

284

Part 2:

Essential Skills

total. It may have fallen with the expansion of the Google index in early 2005. which is not shown in this figure. 10. Amanda Spink, Dietmar Wolfram, B.J. Jansen, and Tefko Saracevic. (2003). “Searching the Web: The Public and their Queries,” Journal of the American Society for Information Science and Technology 52(3), 226–234. 11. For example, Susannah Fox. (2005). “Internet: The Mainstreaming of Online Life.” Pew Internet Life Project. January. 12. An internal study by metrics company WebSideStory in 2003 found that 65.5% of users either type the name into the browser address bar or use a bookmark. This is an increase from 58% in the preceding year. The fraction typically rises with the strength of the brand, and falls off for more generic sites or lesser known brands. Brian Morrissey. (2003). “Search Guiding More Web Activity.” Clickz News. March 12. 13. See additional discussion in Chapter 14. 14. CNN Money. (2004). “Cheney directs surfers to anti-Bush site.” October 6. 15. For more discussion, see Juhnyoung Lee and Mark Podlasceck. (2000). “Using a Starfield Visualization for Analyzing Product Performance of Online Stores.” EC’00 ACM, October. J. Lee, M. Podlascek, Edith Schonberg, Robert Hoch, and Stephen Gomory. (2000). “Analysis and Visualization of Metrics for Online Merchandising,” WEBKDD’99. in B. Masand and M. Spiliopoulou (ed). Springer-Verlag: Berlin; New York, 126–141. J. Lee, M. Podlaseck, E. Schonberg, and R. Hoch. (2001). “Visualization and Analysis of Clickstream Data of Online Stores for Understanding Web Merchandizing.” Data Mining and Knowledge Discovery 5, 59–84. 16. For example, Wendy W. Moe and Peter S. Fader. (2004). “Capturing Evolving Visit Behavior in Clickstream Data,” Journal of Interactive Marketing 18(1), 5–19. Wendy Moe and Peter Fader. “Dynamic Conversion Behavior at E-Commerce Sites.” Management Science 50(3), 326–335. 17. The Link Harvester tool, hosted at several locations online. 18. Alexa.com. May 30. 2005. 19. Robert Blattberg, Gary Getz, Jacquelyn Thomas. (2001). "Customer Equity: Building and Managing

Relationships as Valuable Assets.” Harvard Business School Press 20. Gupta. Sunil, Donald R. Lehmann, and Jennifer Ames Stuart. (2004). “Valuing Customers,” Journal of Marketing Research 41(1). 21. SEMPO study. 22. Compared to IAB total for 2005. 23. A simple keyword search campaign can begin in under 10 minutes, with a $5 initialization fee. 24. Discussion of providing services, alliances, out sourcing, etc. from the search engine watch page. 25. Number 1 on 2/5/05, number 2 on 6/2/05. 26. Meta tags are part of the HTML description of a web page, and can viewed with the “View Source” command of a browser. 27. An example is the Spider simulator, provided at se-spider.com. 28. Carl Kaplan. (1999). “Former Playboy Model Wins Right to Use Keywords.” New York Times December 17. 29. In June 2005. 30. Data from Fathom Online, compiled monthly. 31. “Google, Yahoo! fade on soft ad-spend,” Financial Express. India. June 8. 32. If we knew the average price per keyword overall all ads clicked, it would be roughly 4 billion/ average price per click. An average of $1 per click. which is probably too high, would mean 4 billion visits generated. 33. June 5, 2005. 34. There are also some additional technical assumptions necessary given the specific rules of each of the search engines, but the essentials of the argument stay the same. 35. Author personal discussion with Hal Varian and Mark Schwartz, both of the University of California—Berkeley. 36. This is then a Nash equilibrium in multiple keyword prices. 37. Recall the first banner ad, shown in Figure 6.3. 38. Doubleclick cite. IAB cite. 39. Rick Bruner. (2005). The Decade in Online Advertising: 1994–2004. DoubleClick. April. 40. Net Results. p. 290. 41. http://www.doubleclick.com/us/knowledge_central

Sponsor Documents

Or use your account on DocShare.tips

Hide

Forgot your password?

Or register your new account on DocShare.tips

Hide

Lost your password? Please enter your email address. You will receive a link to create a new password.

Back to log-in

Close