Editor’s note: This month features the second installment of The Online Advertising Sourcebook for Newspapers, authored by Saxotech.

The Sourcebook is an educational reference tool providing a thorough explanation of the state of online advertising and the technologies and methodologies available to improve a publisher’s revenue using online efforts.

Print copies of the book will be available at Saxotech’s booth (2038) at Nexpo in Chicago this month, with a limited number of complimentary copies on hand. Part One of this series ran in the March 2006 issue of Newspapers & Technology.

Advertisers use three main sources of data in their research of customer activity - internal, primary and secondary. For the purposes of this sourcebook, internal data is considered the most relevant. While there are a few variations of internal data such as non-marketing data (accounting, etc.) and sales force data, we will concentrate on customer characteristics and behavior because that is what the Internet can mine the best.

Internal databases may contain some basic customer information. In the past, the database may have been a salesperson’s index cards with names and addresses of clients. Today, information is kept in a computer database, and the customer information gathered is far richer. For example, visitors to the travel site Expedia are asked to register before using the service. Visitors provide a variety of personal information, from airline seating preferences to favorite destinations. These individual customer files might also include information such as phone calls made to the help center and other calls made for service.

Added to the self-reported files is the observation of the customer online. Online behavior is captured by clickstream (the tracking of movement through the site) and/or log files. Doing so allows the advertiser to develop a combined profile of the customer online.

Basic measurement

At its basic level, the clickstream data will show which pages a customer visits, where a customer clicks and how much time that customer spends on a page. Other information derived from Web log files or more advanced analytics systems might include the exact search phrase entered to find your site, prior to even landing on one of your pages. Geographic locations by country, state and city can be extracted to refine customer profiles. What page a customer came from, first looked at, and the last page looked at before leaving your site are other measurable points. This sort of information allows online publishers to adjust and modify content on a site to create more effective user experiences.

Capturing all this information into a database and continuing to refresh it with new behaviors brings value for the advertiser - and for a savvy Web publisher. In addition, this type of information becomes the foundation for a business to build its online customer relationship management, or CRM. It is important to merge the online customer relationship building with the offline efforts. This reinforces the need for seamless integration of all customer-contact databases and systems so that the business has “one voice” to the customer regardless of the communication channel used.  

Managing all this information requires an analytics strategy that cuts across the whole organization. The analytics data (page visits, clickstream, conversion, etc.) is converted into actionable business intelligence and requires a managed approach. Each department in a firm will have different uses for this information and it will be used to meet the department’s individual Web site goals. Analyzing all the data about your site can be overwhelming. A team comprised of a representative from each department will help control the volume of information and how it is acted on.

Unique visitor

One of the more important tracking benchmarks is the unique visitor. “Uniques,” as they are commonly referred, measure the amount of traffic a Web site receives by tracking the number of individuals who visit a site within a specific period of time. This period can vary from site to site and is determined by settings in the historical log analysis software (or, for more innovative reporting, within the implementation of advanced Web analytics services that measure real time behavior).

Unique visitors are measured by their unique IP addresses, which are similar to a fingerprint. However, an Internet server provider’s proxy server can cause many users to look like one person, and dynamic host protocols can make one visitor look like many. (As an example, a company with 1,000 employees might manage its Internet access via a single IP address. If all of those employees visited the same Web site one day, that Web site could register them as just one unique visitor visiting because all it would see is one IP address visiting 1,000 times.) An inaccurate unique visitor count can potentially skew the accuracy of Web tracking or analytics metrics.  

Fig. 1: Nielsen//NetRatings’ measurement of the top online current events and global news sites, November 2005.
Source: Nielsen//NetRatings

Using cookies

One method of getting a more accurate count is to use cookie tracking. When a user visits the site, a cookie is placed within the Web browser on the user’s computer. This allows the analytic software to track all movement to the site (and throughout the site). However, if a single user accesses a site from multiple computers such as home, the office and/or a PDA the software cannot combine these visits into one visit. Other issues with cookies include the fact that they can be deleted by the user and thus will inflate a user count if the visitor returns. Or, individual users may select privacy settings within their browsers to reject the setting of cookies altogether.

Another method of tracking a unique visitor is through site registration. This type of tracking is based on actual site login. For example, if you go to eBay to carry out a transaction, the eBay registration system identifies you by the cookie it placed on your computer in the past, but it also asks you to verify your login to be sure it has the right user. This way the identification of the user can be verified. The site registration form of tracking is ideal and combined with cookies provides a higher quality of site analytics.

One of the largest independent tracking services, Nielsen//Net Ratings, tracks Web sites by using a large sample of Internet users. Nielsen counts the number of unique addresses of computer systems that are used to query a measured Web site. According to Nielsen, issues can be created by the increased instances of cookie deletion that leads to an overestimation in unique browser statistics. This also results in an underestimation of visitor frequency metrics, since repeat visitors are often recognized as first-time visitors (See Figure 1, page 62).

Segmentation

Marketers make informed decisions about their advertising based on the collected data from internal, primary and secondary sources. By taking this information and grouping it based on similar characteristics, the marketer can identify groups to target. These groups can be of any size, from a few to more than a million. By its nature, the Internet allows for targeting of individuals within these groups.

Traditional marketing segmentation is usually broken down by geographic, demographics, psychographics and behavior. The same can be applied to online marketing. In fact, the more online and offline information that can be combined into the same database the better.

Geographical difference

Geographic segmentation may seem important to a business whose galaxy of customers traditionally lives within a few ZIP codes, but the Internet opens up a whole new universe for many businesses.

Suddenly, a medium exists that allows a small antique-toy store in Duluth, Minn., to reach a potential customer in Paris. Opening up the whole world to marketing possibilities invites even more segmentation variables. While that may be appealing to a business that can ship products or deliver services anywhere in the world, it’s not attractive to a local dry cleaner or pizza shop. A firm’s product and/or service distribution becomes the lead thrust behind geographic segmentation.

As such, most marketers want to be able to target specific geographic regions. A company may want to test market a product, as an example, in one particular city. In this instance, a local newspaper site may make sense for the advertiser - or any site that caters to local-only visitors.

A more sophisticated site may be able to target advertising geographically by IP address location. Take, for example, the Iowa bank that wants to advertise its state government employee services on Yahoo! but doesn’t want to advertise worldwide.

By targeting a particular proxy server (a gateway to the Internet combining many computer users to appear as one), in this case the state government’s proxy server, the bank can reach its intended target. Thus, the bank’s Yahoo! ads are only served to the government workers within the state, ignoring others coming to Yahoo!

Demographic segments

Early on, the typical Internet user was a younger male with a higher income and college degree. Over time that changed.

Today, the typical Internet user looks more like the typical American, except that lower-income and less-educated Americans tend to be absent. With such a broad range of the population now reachable on the Internet, marketers can identify the appealing marketing niches within the demographics. Using statistics gathered about age groups, occupational groups and even ethnic groups are important for segmenting the audience.

A recent study by the Pew Internet & American Life Project (see box), illustrates the important demographics that marketers might consider in approaching an online advertising campaign. This kind of information, for example, can help an advertiser looking to reach online seniors determine that unless it segments or targets its advertising, the chances of reaching an adult over the age of 65 on the Internet is only slightly better than one in four.

Psychographics segments

User psychographics are made up of personality, values, activities, lifestyles, interests and opinions. Instead of counting age, gender, race, etc., as is the case in demographics, psychographics measures one’s frame of mind or mental attitude. This type of information is often a key in the positioning of a product. For instance, knowing that someone is a young Hispanic male can dovetail with a marketing campaign for an Hispanic magazine, but knowing that the person is family-oriented and uses technology a lot at home is another bit of information that delivers a very specific - and valuable - target.

Forrester Research offers a glimpse of how Internet users can be grouped based on psychographics. It measures consumer and business attitudes toward technology with its Consumer Technographics panel. Forrester developed 10 different categories based on attitudes and beliefs:

-Fast forwards - Upscale, heavy users of software and technology.

-Technostrivers - Young and upward moving.

-Handshakers - Not into technology for business usage.

-New age nurturers - Upscale heavy users of technology, family oriented.

-Digital hopefuls - Retirees, not using technology for business.

-Traditionalists - Family oriented, slow to adopt new technology.

-Mouse potatoes - Moderately upscale, heavy use of technology.

-Gadget grabbers - Young, buy low-cost technology.

-Media junkies - Heavy TV users, Internet bloggers.

- Sidelined citizens - Not online.

Behavioral segments

Behavioral segments consist of the information gathered from the site and page visits and other Web log statistics such as clickstream analysis. It also incorporates users’ online and offline behaviors.

Most behavioral targeting uses either predefined or customized behavioral categories. The most common segmentation categories are the benefits sought and product usage. For example, if an auto advertiser wants to reach an intended target, the advertiser may buy an “auto research” behavioral group, consisting of online users who show certain auto-related browsing activities.

Product usage can be applied in many ways. An example could be the difference in users to a newspaper Web site. Some “light” users come only once a week and may check only the local business stories, while “heavy” users may visit daily and check various parts of the site.

Behavioral targeting is widely sold on a cpm, or cost-per-thousand impressions basis, especially on newspaper Web sites. The ability to price based on dynamics of behavior is growing among newspaper sites.

Internet users can be segmented according to how they use the Internet. Some users may browse or “surf” the Web while others may have a specific goal in mind. To further break down these segments, marketers profile users by key usage segments such as home versus office, broadband usage, time spent online, industry-specific usage and usage occasion.

Web at work

Many businesses segment by whether users access the Internet at home or at the office. In a Harris Interactive poll conducted in April 2005, half of the 300 employees surveyed said that they use the Internet at work and that they do so for both work and personal chores. The most popular category while at work: news, accessed by 81 percent of workplace respondents.

The type of connection has a big influence on user behavior. Faster Web connections at the workplace have caused Internet surfers to expect this kind of response at home. As a result, home broadband usage has grown dramatically. Internet broadband becomes a necessity for quicker music and video downloads, to access voice-over IP telephony services, to view virtual tours and to experience rich media.

The amount of time spent online is another segment that marketers watch. A survey from Jupiter Research reports that time spent online continues to increase, and that it’s coming at a cost to time spent with broadcast and print. The study reveals that consumers now spend 14 hours a week online, the same amount they spend watching television. The study also found that the most intense online users are more likely to take advantage of technologies such as streaming radio and RSS (really simple syndication).

Knowledge gained

Unlike other media that are not as measurable, the Internet provides a huge amount of data on page views, clickthrough, unique visitors, time spent online, etc. - collectively viewed as clickstream analysis. Proper use of it can reveal a mountain of information about consumer habits. Understanding segments and behavior adds another layer of insight to the process. Robust behavioral targeting applications can apply this knowledge across differing business models, meeting many types of advertisers needs. Having all the appropriate data and interpreting it properly is an opportunity to drive online strategy, develop marketing materials and measure ROI.