Analyze this

IMAGINE A 24/7, REAL-TIME FOCUS GROUP — one that gives you access to a customer’s every move. Imagine sitting upon the shopper’s shoulders as he views your line of goods, perhaps eschewing jeans in favor of shorts, and tagging along during the checkout process. And imagine doing this for all your customers and logging these experiences into a comprehensive analysis with little more than the click of a mouse.

Well, imagine no more.

Welcome to the Orwellian world of Web analytics, where marketers can gauge precisely what shoppers want by tracking their virtual journeys — or, in Webspeak, click paths — across their company’s Web pages. Using cookies, small files often deposited on consumers’ computers by the Websites they visit, Web analytics programs can identify not only where your customers have come from but also what they’re searching for, what they’re looking at right now, and most important, whether they’re buying.

Throw in the ability to gauge the effectiveness of online advertising, marketing, and site design, and you’ve got, as Mike Sidders, director of e-commerce and new customer acquisition for general merchandiser Fingerhut, calls it, “a slam dunk.”

“Let’s put it this way,” says Sidders. “A good Web metrics package doesn’t have any difficulty achieving a substantial return on investment.”

And that means an investment beyond the basic analytics packages that Internet service providers (ISPs) include as part of their service. Such offerings “probably won’t be adequate for a full-scale Web analytics program,” says Dmitri Eroshenko, CEO of Clicklab, a McLean, VA-based software services provider. Depending on the size of your site, the amount of traffic, and the analytics required, an add-on metrics solution can cost anywhere from $60 a month to more than $50,000 a month.

Playing tag

“Historically Web analytics revolved around counting visitors and pages,” Eroshenko explains, “but for a program to be successful, you really need to track many more variables — namely, being able to isolate sales from a particular traffic source, because depending upon where your visitors come from, conversions may be different.” What’s more important than counting pages and visitors, he says, “is to provide usable and actionable information.”

The “counting pages” type of data collection, also known as log-file reporting, is a server-based method used to capture static metrics such as hits, visitors, and abandonment rates. It is the precursor to the more customizable page-tagging technology, a client-based platform that captures richer metrics, including customer value, retention, and attrition. While log-file analysis might track visitors as they move from page to page, page tagging allows marketers to track events that may happen on a single page, such as the addition or removal of an item from a shopping cart.

In the case of San Mateo, CA-based Web analytics software vendor Coremetrics, client Websites are tagged with lightweight JavaScript page tags (also known as Web beacons) called SmartTags. According to Coremetrics founder Brett Hurt, the average time investment for the client depends on the size of the site and ranges from 5 to 15 business days.

“Catalog retailers could greatly benefit from this advance in Website traffic analysis,” says Karen Hudgins, market specialist for Sane Solutions, a North Kingstown, RI-based company that markets the NetTracker line of Website traffic analysis solutions. For example, she says, “if an online catalog features several items within a Web page that requires a visitor to scroll down in order to view all the items, the use of page tags will allow the Website owner to see if visitors are scrolling down far enough to see all the items displayed.”

Ultimately page tagging can give marketers a portal through which to “understand how effective their merchandising and promotions are,” explains Hurt. “What creatives for merchandise are working? If they put something on the home page, is that smart use of that real estate? What are the abandonment rates for different merchandise located at different areas? Is the onsite search working effectively? We sit in the center of all that activity and measure the effectiveness of all that.”

A fix for failed searches

Minnetonka, MN-based Fingerhut sells a wide variety of products, including electronics, jewelry, and home goods. The cataloger, which closed its doors in early 2002 and reopened under new ownership in November of that year, had been using an ineffective Web analytics platform, according to director of e-commerce Sidders. “When the company relaunched, we had a very rudimentary onsite search engine, and we were leaving a lot of money on the table and leaving our customers with a very bad search experience.”

Using Coremetrics’ Online Analytics package, Fingerhut saw conversions increase 26%. “We identified problems that could be fixed within our rudimentary search,” explains Sidders. These problems included plural and singular conflicts (for instance, if someone typed in “widgets,” the search engine wouldn’t call up products tagged “widget”) and common misspellings.

“We saw dramatic increases and conversions for that area and corresponding profitability,” Sidder notes. “At the end of the day, it’s all about conversion, it’s all about ringing the till.”

Further, after learning how people navigate your Website to ultimately find what they need, you can apply that knowledge to the design of your catalogs or even stores. Because the demographics of online and off-line customers are becoming very similar, multichannel marketers can ultimately learn from their Website what products and creative might work best in their print catalog. You can learn what product assortments are preferred or how people navigate to find what they need, and place all of that information into the design of your catalog.

“Nearly all the insights that you learn about within your online channel can almost be immediately applied,” says Hurt. “Part of the challenge in heading in that direction has been the silos between the online and off-line teams, but those are very quickly breaking down because the online channel, especially for catalog companies, is becoming as high as 60% of their overall business.”

Paid-search applications

These insights can also help you fine-tune your paid-search campaigns. San Antonio, TX-based Ace Mart Restaurant Supply, another Coremetrics client, has used Web analytics data to reallocate funds to the most productive ads and campaigns. Ace Mart sells restaurant equipment, furniture, and accessories. “We’ve been able to slash marketing spending by 25% while maintaining the same levels of online sales,” says Webmaster Ryan Rodkey. “For example, we were able to eliminate certain keyword purchases that just weren’t converting. Our sales and profitability are steadily growing while our competitors buy up all of the nonproductive keywords without understanding the impact of their purchases.”

Coryon Redd, owner of Los Angeles-based Cellphone Battery Warehouse and a Clicklab client, was able to remedy a similar search-engine roadblock. The company, which specializes in batteries for cell phones and cellular accessories such as car chargers, hands-free kits, headsets, and antennas, had a category on Google AdWords for one manufacturer of phones that was performing very poorly.

“Our pricing and selection were good, but our conversions were always low,” Redd says. “After analyzing the Clicklab traffic log, we added negative keywords to the category based on specific searches on Google.”

During the course of Cellphone Battery Warehouse’s research, the company realized much of the confusion resulted from people searching for the manufacturer’s cordless phone line, for which the marketer does not stock products. To stop attracting such shoppers, “we add ‘cordless’ as a negative keyword to all of our campaigns because we do not sell cordless phone batteries,” Redd explains. “We also ‘negative out’ keywords such as ‘free’ and ‘keypad.’ Because it is very difficult to guess the variation of user searches, we used our traffic logs from Clicklab to find other keywords that are not related to our site or just perform badly.”

The result? “After a period of time, we were able to eliminate most searches that were matching our general search phrase but not our product selection,” Redd says.

Cellphone Battery Warehouse saw the cost of advertising decrease by 55% while its revenue increased by 75%, “because we were better able to match searches with the correct landing page on our site,” Redd says. “We saw the positive change immediately and have continued to see the category do well.”

A beautiful thing

Indeed, these kinds of results illustrate the beauty of Web marketing, says Fingerhut’s Sidders: “You can see places where you can stand improvement, go in and make the change, and through a Web analytics package, see the change occurring, see the benefits instantaneously.”

Multichannel marketers might think of Web analytics as the new and improved square-inch analysis — only the “pennies per book” mindset has morphed into what may be, in this virtual world, more appropriately tagged, “pennies per look.”

But, again, that’s only if you’re really out there watching.


Massapequa Park, NY-based freelance writer Dina Santorelli has been published in 1to1 Magazine, Long Island Newsday, and Do!, among other publications.

The nitty-gritty of selecting a solution

The Web analytics market is diverse and well populated, with myriad products and services available, each varying in its reporting capabilities, data collection methodology, and delivery. Some solutions are delivered as a service, for which you pay a monthly fee, while others are sold as a software product, which you buy only once. Some analyze Web server log files; others use page tags to gather their Web information.

“Before looking for a Web analytics vendor, a company needs to understand what they need to track and what their set of requirements is,” says Dmitri Eroshenko, CEO of software services provider Clicklab. “Simply knowing that you need Web analytic services — well, everyone provides Web analytic services. But will the service have the functions you need to track your different sets of data — that’s the big question.”

“With b-to-b, typically the majority of your sales comes from a handful of customers, and you really need to understand at a deep level what those customers are doing,” says Brett Hurt, founder of Web analytics software vendor Coremetrics. “You want to get this information into the hands of your sales team, who is serving those large accounts, to say, ‘Here is what that customer or prospect is interested in and doing,’ not just, ‘Here’s what they’re buying.’ From the [b-to-c] perspective, you’re trying to determine for a massive amount of customers, most of whom have some level of brand recall from receiving your catalog in the past, what products they’re looking at and how they’re using your different site tools.”

Overall, when selecting a Web analytics platform, you should investigate the following:

Reporting: Does the software allow you to do drill-down or roll-up analysis? An example of drill-down analysis would be looking at a chart and seeing something interesting and then “drilling down” to view the details. Conversely, with roll-up analysis, you start at a granular level and then back out to see the bigger picture. “Interactivity in the reporting and in the interface are critical,” says Chris Berk, executive director of technology for Resource Interactive, an independent interactive agency in Columbus, OH.

Customization: Can you create your own unique reporting based on criteria that are important to you as a marketer? Can you export and share the data?

The administrative end: How difficult is the application to set up and maintain? Can you easily set up security levels, particularly if multiple individuals are going to be accessing the data?

Staff: Do you have personnel on staff who are qualified to conduct the data analysis?

Cost: It varies — widely. For example, Clicklab starts at $60 per month, which includes 60,000 pages, while Coremetrics’ Web analytics services start at $2,000 a month for very small catalog companies, with pricing based on the amount of site traffic. At the top end, Coremetrics charges more than $50,000 a month for some of its largest and highest-level clients.
DS

Analyze This

Null hypotheses and chi-squared tests may seem like relics of your college days, but you can’t afford to bury them if accurate organizational measurement is your objective. For corporate research to be of value, statistical rigor is vital, says marketing professor Geoff Lancaster in his new book Research Methods in Management (Elsevier, 2005). Lancaster provides a thorough grounding in data collection, classification, and analysis, taking the time to discuss in detail often-overlooked basics such as the differences between scientific and non-scientific research, data and information, primary and secondary sources, and observational and experimental methodologies. In particular, Lancaster has written a fine chapter on analysis, pointing out the dangers of misinterpreting research results and warning us that “ineffective data analysis can lead to a number of potentially disastrous outcomes with regard to tackling organizational and management problems.” He cites one study in which the researchers missed the effects of group processes on productivity. As a result, the company took the wrong steps to enhance productivity, concentrating on the physical environment rather than on workgroup design and management.