WASHINGTON, D.C. -- Getting the 11 million members of the United Methodist Church to engage with its website is no small task for United Methodist Communications, the church's marketing arm.
With 250,000 current visitors to the site each month, leveraging Web analytics is a prerogative, particularly given its current one-person staff.
"We're just getting started, using Google Analytics, establishing some basic metrics," said Poonam Patodia, the electronics marketing manager with the Nashville-based organization. "We need to establish what we want the site to accomplish."
The organization's situation is a common one, said Bill Gassman, research director at Stamford, Conn.-based Gartner, during a presentation here at the Gartner CRM Summit. In fact, in organizations without a formal Web analytics program, Google's free tool is a good place to start.
"If you're in that situation, let chaos reign," Gassman said. "Let them use Google. It is valuable enough for a lot of people, so when you get serious about [Web analytics], they're ready."
As organizations progress, they'll want the deeper functionality of established Web analytics vendors like Omniture, Coremetrics, Web Trends and Unica, he said.
The market for Web analytics has progressed significantly from its heyday during the dot-com boom and the subsequent bust.
"Back in those days, companies were spending $200,000 or $300,000 on software, but the problem was IT was buying these products and no one was using them," Gassman said. "In the last five years, we've seen a real rise in usage, and best practices are pretty well known."
Now, an organization can spend $250,000 per year on both people and tools to do Web analytics correctly, he said. And it's relatively easy to get a return on that investment. A 1% to 2% increase in conversion rates is enough to pay for the program for a couple of years.
Yet many organizations are still in the initial stages of Web analytics maturity. It takes about a year per level of maturity to move from basic awareness to pervasive use, Gassman said.
According to Gartner surveys, 95% of organizations have reached the first of five levels -- where the project is driven by IT and reports are created, but few decisions are made as a result. Thirty percent are at the second level, where the project is business driven and metrics on changes to the website are tracked. About 40% of organizations are in the fourth level of maturity, and less than 5% have reached the fifth level where Web analytics are rolled up to the business intelligence (BI) group for activity-based tracking and balanced scorecards.
Balancing privacy against analysis when evaluating and deploying Web analytics
Organizations need to take several factors into consideration when deploying Web analytics -- and privacy is an important factor.
"Ideally, a marketing person wants to know every click every person ever made on every site," Gassman said. "Users are saying, 'I don't want you to see that.' "
In addition, new Web browsers from Google and Microsoft include a "privacy mode" that prevents websites from tracking their behavior. With more sophisticated users clearing their cookies regularly, Web analytics is facing some new challenges.
"If you want people to keep cookies and save them, then the answer is to add value," Gassman said. "If you give users enough value to not remove [cookies], you can track it."
Things like the personalization Yahoo provides are reason enough for users not to clear their cookies, Gassman said. However, too much tracking can create mistrust, he added. Users become uncomfortable when a salesperson calls them to ask if they're ready to close the deal just after they get off a product page on the website.
"What has to happen is there has to be a set policy," he said. "What do you track? Why? How long do you keep it? Who uses it? Where's the contact information if a user wants to be removed?"
Evaluating and buying Web analytics
Some companies start their evaluation with the decision to buy or build the tools -- and the answer is definitely to buy, Gassman advised.
"The problem is there are a lot of products to buy," he said. "Think of your website as a big data warehouse."
There are tools for gathering attitudinal information, operational information and competitive information.
Most organizations select a Software as a Service (SaaS)-based tool.
"It is the right fit for about 80% of the market," Gassman said.
With SaaS, there is a broader choice of vendors and better user support, and it demands fewer IT resources and less commitment. Beware of changes in funding, however. Where IT may have initially funded the project, with SaaS this often gets transferred to the business side. Also, by sending information over the Web, there are concerns about privacy and data ownership because the vendor has the data in his data center. But those concerns are often overblown, Gassman said.
"Generally, people who switch vendors throw that information away," he said. "It gets stale fast."
Companies purchasing a Web analytics application should begin by asking what questions they want answered first from the program. Then they can prioritize the tasks required to identify the right vendor.
They should also have a strong C-level executive who gets the value of Web analytics and makes sure it happens. Bring in outside expertise. Consultants will help, but the vendors can usually provide enough information. Companies should also institute a strong training program and a governance and expertise team.
"Understand it's a specialized skill and you need a specialized team," Gassman said. "Think about cross-channel opportunities, and start thinking about pragmatic integration."
Another new challenge with Web analytics is the emergence of mobile browsers and marketing to mobile users.
"There are very few features that work with WAP with all the new devices," Gassman said. "The days of WAP are numbered. What you see is vendors can identify the browser, the screen size and the device, but a lot of the devices don't have Java script. You can attach campaign information in the URL. It ends up being a bit of a data model issue."