Tracking Visitors to Your Site

Let's assume that you set up an attractive Web site, following all the tips and tricks for getting visitors to come early and often. You hope to build an electronic mailing list and make sales to a certain percentage of your visitors. But how will you gather those e-mail addresses? And how will you know what part of the site visitors find the most interesting? Or which part compels visitors to make a purchase? One wonderful thing about the Web is that simple software programs exist to gather all of this information.

When advertising in traditional media, it's often impossible to accurately track customer response. When Bob's Cigar Shop takes out an advertisement in the local newspaper, nobody knows how many readers actually look at that ad or for how long. If Bob pesters every walk-in customer, asking where they learned about his shop, he'll only get a rough idea of how many are making purchases based on his ad. If Bob wants to analyze his ad statistically, he'll have to invest a lot of time calculating how many readers the newspaper says it has, how many customers said they made a purchase in his store after seeing the ad, and the average amount of those purchases. It's hardly an exact science.

Telesales organizations and catalogers have long tracked what advertisement customers are responding to by asking them, "Where did you hear about us?" or asking for a customer identification number from the back of a catalog. But the problem with this method is the same one that Bob has­many customers don't remember or care. Since they're the ones supplying the information, it will often be flawed, because customers forget where they learned about the company. Customers ordering through someone else's catalog, or ordering without a catalog, further skew the information.

But on the Web, customers are tracked accurately without even knowing it. As a company with a Web site, you can find out how many people visit your site, where they come from (another Web site that linked them to your site? a search engine?), how long they stay, what sections they browse, and for how long. If you ask visitors to the site to fill out a simple registration form, you can also tell who's a new visitor to the site and who's a repeat visitor.

This information is essential for companies that want to sell space on their Web sites to advertisers. Advertisers want to know how long visitors stay during each visit, if the same visitors are returning again and again, what sections they're looking at and for how long, and what type of access they have (through a corporate account, government organization, or university). Advertisers pay for Web advertising using the same model as companies that pay for magazine or newspaper ads­prices are based on how many people read the publication.

Web-site tracking software can gather significantly more detailed information about readers than any newspaper or magazine can offer. Print publications can only give the number of subscribers and newsstand sales, not the number of actual readers each month. With Web tracking software, every company with a site can track how many people look at the site each day.

There are basically two ways of tracking visitors using software. The first dates back to the early days of the Web: A typical person visiting a Web site clicks on a page, enlarges a graphic, listens to a Web audio recording, leaves, and then comes back and makes other moves within the site. Each of those activities makes a request of the Web site's host computer to send the visitor's computer some electronic information. Those requests are tracked at "hits." Basically, a hit registers each time a visitor clicks on part of the site. A hit also registers for each graphic or link on a page. So if a visitor merely looks at a page with four graphics on it, that visitor registers as four hits. And if a visitor arrives at a particular site, then clicks on four different pages, plays a short video clip, gets disconnected, and comes back and clicks on two more pages before leaving, that particular visit registers at least 10 hits to the server software.

A site that claims 10,000 hits a week may in reality only have 2,000 visitors a week, each one clicking around the site five times before leaving. Or it could have 8,000 visitors a week, but most only look at the site's first page before leaving. Companies rely on various calculations for translating their hit rates into number of visitors, but nobody knows for sure.

If the information is unreliable, why do companies still use it? One reason is that the capability to track hit rates is already programmed into server software, so Web-site developers don't have to buy extra software. Free software is also available on the Web for companies that want a separate software package offering a bit more detail than provided by some server software. Rather than buy a complex software package or pay an ISP to track the information, companies using hit rates can simply rely on their server software or download the free software to get a general idea of the site's traffic and trend information.

The data is saved in files, which can then be reviewed on screen or printed out. If a Web site's server has a monitor attached to it, someone sitting at that computer can watch the hit-rate information being gathered: If three people are looking at the site, the monitor would show columns of data that track the visitors' progress through the pages. That information can be collected into a report as often as the company wants­daily, weekly, or most common, monthly. Some tracking reports are stored in a form compatible with database software, so they can be analyzed more easily. You can set up hit-rate data files to show whatever information in whichever order you choose.

Some companies have adapted hit-rate tracking software to offer even more detailed information. In addition to hit rates, they can determine what browser software visitors are using, what site the user just came from (so a company can judge how many visitors come from a link from another site), what modem speed the visitors are using (so the company can put fewer graphics on the site if most visitors are using very slow modems, for example), and what type of Internet access accounts visitors have (commercial access or from a university or government organization).

The software will also track error messages. For example, if the computer is overloaded when a visitor tries to play a video clip, and the visitor gets an error message saying that the clip is currently unavailable, the company sees that message on the hit-rate log. So the company will know if there's a peak time when lots of error messages go out and can look for problems with the site's capability. Even with this additional data, a company that talks about visitors in terms of hit rates is still working with very vague information.

With either type of traffic tracking, the names and e-mail addresses of the visitors themselves usually remain anonymous unless they voluntarily provide it. If visitors provide that, they might also be willing to provide demographic information about themselves, such as where they live, their income level, and a few details about their family's purchasing decisions. Gathering this data at Web sites is becoming more common as the Web grows. Companies need this information because it helps them focus the site to visitors' demographic profiles. E-mail addresses are important because mailing lists are valuable as a way to attract repeat visitors and recruit new ones by word of mouth.

Some companies require visitors to register before they can even access the site. Basically, it works like this: Visitors to the site see an introductory page, showing what information and entertainment the site offers. On that page, a notice explains that in order for advertisers to continue to sponsor the site, or in order for the company to customize the site to visitors upon their return, registration is necessary before the visitor can delve further. At that point, visitors can either leave or continue with the registration process.

The registration form is typically only a page or two and usually doesn't ask for full names and mailing addresses. Most Web users don't want to provide that much personal information. Instead, surveys ask for general information like age, interests, income, and family statistics. Once the form is filled out, visitors get a password to access the site.

The first company to require visitors to register was Wired magazine at its Web site, HotWired, in 1994. Visitors who registered were issued passwords to access the site. To return, visitors had to remember those passwords to gain entry to the site. HotWired was able to gather tracking information on repeat visitors­what they browsed at the site and how often they returned. The magazine used that data to become one of the first Web sites to sell advertising, making thousands of dollars per month from advertisers like Audi and setting a Web money-making precedent.

The biggest risk to registering visitors is that they may decide just to skip the site altogether rather than fill out the form. Many Web users dislike being tracked and will boycott sites that try. It's also very easy for visitors to throw off statistics, and they gleefully do just that. At sites that require registration in exchange for a password, visitors fake their demographic information. Knowing that they're being tracked, they click on information they don't really want. And once they leave, they give their password to others. For some time, Hot Wired's most popular visitor was "cypherpunk," a name linked to a password passed among countless users. That name topped Hot Wired's hit list month after month.

Another risk to requiring visitors to register is that the data isn't completely accurate. Since visitors are volunteering the information, they're likely to make something up when faced with a question like, "How many times in the past year have you bought copier paper?" To get the most accurate results, keep the form simple and short.

It's not difficult to imagine how useful this tracking information can be. Imagine placing an ad in a wide-circulation magazine or newspaper. Imagine that every time somebody looks at that ad, a database tracks what they do. It tracks what the person looked at right before the ad, which tells whether the placement of the ad is successful. It tracks how long they looked at the ad and what parts of the ad they read, which tells how compelling the ad is. And it tracks what the person does immediately after looking at the ad, if they have taken action in response to it. The kind of information can help drive a company's marketing decisions and product research. Gathering visitor feedback via e-mail supplements this information. Companies have even set up virtual roundtables with customers online to discuss future products or services or gather feedback on existing ones.

Tracking information can also work for a site immediately, before the company even sees the printouts. That information can be automatically tied to advertising at the site, whether those ads are for the site's products and services or for sponsors. Here's how the process works: A visitor to a real estate company's site fills out an online form asking for information about vacation property on Cape Cod. After clicking on the form and receiving in return a list of brokers in that area or properties for sale, the visitor might also see an ad for a mortgage company that specializes in vacation property financing or an ad for a vacation time-share association. If the visitor clicks on those links for more information, the real estate company can track that process and show the mortgage company just how many visitors are being sent to them. That's one way to establish the value of those ads.

Not only is Web-site tracking information useful, but it's essential for companies that want to make money on the Web. Is your company ready to profit from marketing on the Web? While it's important to learn as much about the medium as you can before investing in it, that doesn't mean you should wait until Web advertising becomes as popular and costly as a 30-second spot on Super Bowl Sunday before you test the waters. With the cost of Web sites so low and the size of the potential market so huge, it will pay to jump in and get your feet wet now.