Hack77.Track Demographic Data Using Custom Variables and Visitor Segmentation
Hack 77. Track Demographic Data Using Custom Variables and Visitor Segmentation
If you're collecting any type of demographic data from your visitors using forms, you can easily pass this information along to your measurement application and do some light demographic analysis. It is very common for sites that require registration to ask a few questions about demographics in the process of signing a visitor up for a service (Figure 5-16). The most typical questions are, "What is your gender?" and, "When were you born?" but occasionally sites ask for other information, such as annual income, marital status, or whether you have children. Fortunately for you, since this information is nearly always collected via an HTML form, the results can easily be passed to your web measurement system and collected into custom variables [Hack #31] or used to create visitor segments [Hack #48]. Figure 5-16. Subscription form with demographic data 5.11.1. How to Pass Demographic Data to Your Measurement ApplicationFor the most part, forms like the one in Figure 5-16 use the form POST method in HTTP to hide the fields and values from prying eyes, a good idea when you're asking for personal information. If this is the case, you'll have to use some server-side code to grab the information from the form processing script and pass it to your measurement application. For example, the following form uses the POST method to collect three pieces of useful demographic data: <FORM ACTION="formprocessor.asp" METHOD="POST"> Your gender: <SELECT NAME="gender" size="1"> <OPTION>Male</OPTION> <OPTION>Female</OPTION> </SELECT><br> Your year of birth: <INPUT TYPE="TEXT" SIZE="4" NAME="year_of_birth"><br> Your marital status: <SELECT NAME="marital_status" size="1"> <OPTION>Married</OPTION> <OPTION>Single</OPTION> </SELECT>< <INPUT TYPE="SUBMIT"> </FORM> The script formprocessor.asp will have access to three variablesgender, year_of_birth, and marital_status-the values of which can be passed along to your measurement application. Specific strategies differ, depending on whether you're using a logfiles or page tags as a data source. 5.11.1.1 Pass demographic data to a server log analyzer.Unfortunately, unless your forms are set up to use a GET method, it is difficult to provide this type of information to your measurement application via web server logfiles. The GET will post all of the form data directly into the logfile for analysis. If you're required to use a POST, contact your vendor directly for ideas. 5.11.1.2 Pass demographic data to a client-side page tag.Using the form above, Active Server Pages/VBScript, and WebSideStory's HBX JavaScript page tag as an example, all that would be required to send this information using the page tag would be to use Active Server Page's REQUEST object to drop the form values into JavaScript. //CUSTOM VARIABLES hbx.ci="";//CUSTOMER ID hbx.hc1="<%= REQUEST.FORM("gender") %>";//CUSTOM 1 = GENDER hbx.hc2="<%= REQUEST.FORM("year_of_birth") %>";//CUSTOM 2 = YEAR OF BIRTH hbx.hc3="<%= REQUEST.FORM("marital_status") %>;";//CUSTOM 3 = MARITAL STATUS hbx.hc4="";//CUSTOM 4 hbx.hrf="";//CUSTOM REFERRER hbx.pec="";//ERROR CODES When the formprocessor.asp script is rendered at the server prior to being passed back to the web browser, the ASP code is processed and, assuming the visitor answered "Male," "1970," and "Married," the rendered JavaScript will look like this when it arrives in the browser: //CUSTOM VARIABLES hbx.ci="";//CUSTOMER ID hbx.hc1="Male";//CUSTOM 1 = GENDER hbx.hc2="1970";//CUSTOM 2 = YEAR OF BIRTH hbx.hc3="Married";//CUSTOM 3 = MARITAL STATUS hbx.hc4="";//CUSTOM 4 hbx.hrf="";//CUSTOM REFERRER hbx.pec="";//ERROR CODES When the JavaScript is then executed in the web browser, the page tag sends this data along and it is collected in custom variables one through three. 5.11.2. What to Do with Demographic Data Once You Get ItThere are two basic things you can do with this type of data, one fairly benign and one tremendously powerful. The simple strategy is to simply use the measurement application to count the number of members of each demographic group. While benign, even this type of data can be very helpful to your web marketing group, allowing them to better understand the demographic makeup of your typical visitor or subscriber (whatever the impetus for completing the form is). Simple data such as this can be very useful in helping companies challenge their assumptions. Imagine:
Just having this type of information can be very telling, occasionally forcing companies to substantially change their marketing acquisition strategy. A tremendously powerful use for demographic data collected in this fashion is to build demographic visitor segments from the data. Given the recent rapid maturation of visitor segmentation tools in some of the best web measurement applications [Hack #3] and depending on which application you use, you may be able to:
What's more, the best applications will allow you to easily combine demographic data with more traditional web data and create very complex visitor segments. Depending on your applications ability, you may want to:
The fundamental value in creating these segments is that, usually, once you add a visitor to a specific segment, she remains in that segment from visit to visit (at least as long as her cookie doesn't get reset [Hack #17]). This means that once you're able to establish that a particular visitor is "female, 18 to 25, single with annual income from $25,000 to $34,999," you'll be able to track her habits as she returns to the site and continues her interaction.
Again, depending on your particular measurement application's sophistication and ability to create and manage complex segments, this type of analysis may provide the single greatest benefit from your investment in web analytics. 5.11.3. Things to Keep in MindThere is a handful of things you'll want to keep in mind when attempting advanced segmentation and analysis. 5.11.3.1 It is usually not easy to do.While it's pretty easy to describe, and it sounds easy enough to gather the data and build segments, the reality is that most vendors' abilities to collect custom data and convert that data into useful visitor segments are relatively immature. Either one of these needs in isolation is complex enough; the combination is sometimes impossible. 5.11.3.2 It requires tremendous planning to pull off properly.The recommendation is to carefully define which demographic data you're going to collect and which demographic segments you're interested in well in advance of doing any coding. You should then contact your vendor and share this plan, asking them to explain where they see risks and which reports you'll be able to generate for the segments in question. 5.11.3.3 Most often your sample size will not be significant.Depending on what you're doing to get visitors to complete the form in which you ask the demographic questions, you may never get enough visitors through the form to segment a statistically relevant sample of your population. Keep this in mind when you're setting up segmentation as a critical factor influencing the return you'll get for the work you do. You may want to consider offering some incentive to visitors to complete the form, driving more segmentation. 5.11.3.4 Some consumers are liars when filling out forms.Depending on what you're offering people to complete forms like the one shown in Figure 5-16, you may or may not get truthful responses. Companies that collect this type of data online often report a tremendous number of 18-to 25-year-olds who make in excess of $100,000 annually, data that does not match known populations. You may want to consult with a trained statistician to make sure you have statistically relevant populations before you start making business decisions based on this data. 5.11.3.5 Not every report is always available for every segment.Imagine that you've gone through all of the trouble of creating useful demographic segments to learn that the reports you're most interested in aren't available to visitor segments? Because of the cost and complexity associated with segmentation, it is very common for vendors to provide only a subset of their reports to visitor segments. The only way to know for sure is to ask. 5.11.3.6 Occasionally the payback is hardly worth the effort.Keep in mind that sometimes the great insight you'll get from this type of analysisand sometimes web measurement in generalis nothing more than "yeah, we knew that." Be prepared to not be blown away by the data once you get it. If you end up in a situation like this after having spent hours on planning and implementation chalk it up to experience that you'll use down the road and feel good in knowing you were right in the first place. The fundamental recommendation when trying to build these types of segments is to start small. Instead of collecting every piece of demographic data possible and building 50 segments, slicing and dicing your visitors every way possible, consider collecting simple demographic data (gender is a good example) and segmenting from that. Starting small helps you develop the technical skills necessary to build more complex segments and explore whether the information is truly valuable to your organization. 5.11.4. Simple Demographic SegmentationSome of the currently available web measurement applications are unable to build complex, ad hoc visitor segments and are instead limited to creating simple "move forward" segments. If this is the case for you, don't despair! You can still take advantage of the type of segmentation described with a little more work. The usual strategy for simple visitor segmentation is to require you to create segments in advance. These segments are then assigned numerical IDs that are added to page tags or server requests when a visitor is determined to have joined a group. The extra work is that you need to figure out which demographic elements are assigned to which IDs in advance. For example, if you wanted to track gender and marital status, you might create a segment matrix such as this:
As you can see, you simply make a list of demographic data points and possible states of each, then use your analytics application to assign an ID to each. While this is simple for relatively binary demographic data, consider if you want to keep track of visitors by gender and marital status:
Now imagine that you want to add age groups to the mix! Hopefully, you can see that this strategy is slightly more limited and requires additional thought and planning to pull off. Fortunately, sometimes rigorous planning makes you appreciate the problem more deeply and encourages you to collect only data you're prepared and able to actually use. Fraught with complexity, dependent on visitors who may or may not lie about their demographic grouping, and reliant on vendors who may charge exorbitantly for the chance to learn more about your audience, perhaps this segmentation strategy hardly sounds worth it. Still, given the opportunity to learn more about an anonymous audience, especially the demographic details that can be mapped back to the offline world, and a tremendous body of learning about how different types of people shop and interact with their environment, most marketers simply ask, "Who do I make the check out to?" when offered this type of information. |
EAN: 2147483647
Pages: 157