Lines of site

Web analytics software can help improve how agencies interact with their users

American Customer Satisfaction Index


CMS Watch Web Analytics Report


Google Analytics


Nielsen Online


Open-source Report Magic


Web Analytics Association



1. What questions about your Web site do you want to answer? This step will require major thought and likely extensive consultations with your staff. It calls for consideration of the overall goals of your Web site and expert knowledge about what kinds of questions can be asked and answered.

2. Does your agency have the expertise necessary to interpret Web analytics information? If not, is your agency large enough to develop that expertise in-house? Or should you use outside consultants?

3. Would online surveys help answer your questions? Surveys can reveal user motivations and reactions; numbers can't.

4. Do you want the all-encompassing statistics available from Web logs ' but which might result in large amounts of data ' or do you want to focus on specific aspects of your site, as page tags do?

5. In selecting a program, does the vendor have specific experience with government agencies? Government agencies often have different goals than businesses.

6. Are you doing e-commerce on your site? If so, make sure vendors have experience with that.

7. Does your agency have the resources to run Web analytics products, or would you prefer the vendor to run the software? This determines whether you should license the software and run it yourself or use the software-as-a-service model.

8. Do you want to monitor your site's use of special software such as Flash or Web 2.0 elements such as audio, video, blogs and wikis? If so, make sure the vendor can provide the coverage you want.

9. What specific reports and report formats would you like? Make sure the vendor can generate these.

10. What specific vendor products and services will you need, and what is their cost? Beware of not-so-obvious fees for professional services or extra modules.

Understand whether costs are based on visitor hits, server activity or other metrics.

11. How difficult is it to run the software your agency will be using? Hands-on trials are essential for this complex software.

12. Is a trial period possible? Many agencies are dissatisfied with the results they obtain, so it's important to identify problems before implementation.

When was the last time your organization's Web site went down? Is usage of your site going up or down? As Web sites gain importance ' serving as the primary public interface at many agencies ' these are questions worth answering. You may get some direct feedback from users via email or phone calls, but a much more effective way to monitor your site is with a good Web analysis program.

Web analytics software can not only answer the obvious questions about the flow of traffic to your Web site but also tell you a lot about the experience visitors have when they get there. Among the information you can gather is:
  • Where are visitors coming from?
  • What do they click on?
  • What do they search for?
  • How long do they stay?
  • Are there peak traffic times? Slow seasons?

  • It's true that Web analytics programs have limitations. 'Some basic questions are very difficult to answer, such as why someone visited your site, what they were looking for, and whether they found it or not,' said Joe Pagano, digital media projects coordinator at the Library of Congress.

    Reading the signs

    Still, a skilled analyst can figure out quite a bit about visitors' intentions from the numbers culled by a Web analysis application. For example, Pagano noticed a puzzling situation in which most visitors to a new Web site were not coming from search engines, as expected, but from e-mail sites. 'It turned out that previous visitors were spreading news about the site by word of mouth,' Pagano said. Without that key insight, the search engine information might have been inexplicable or it might even have led the team down a dead end of trying to boost search engine hits.

    You can also use Web analytics as feedback for experiments. Suppose you try a new layout of a Web page. If visitors seem to get their information faster, the new layout could be a success.

    If visitors get their information more slowly ' or avoid the page ' maybe you want to go back to the drawing board.

    Web analytics can also suggest changes you might not have thought of. For example, if a significant number of visitors are searching for a downloadable government form, and that form is buried six pages into your Web site, you might want to promote it to the front page.

    For all the potential of Web analytics software, experts agree that tapping that potential requires a skilled staff. 'There are many Web analytics vendors but not many trained people who can interpret what the numbers mean,' said Alex Langshur, president of PublicInsite, a consulting service for the public and nonprofit sectors.

    Two ways to track

    There are two basic ways to gather Web analytics data: Web logs and page tags. A Web log is the record that most Web servers keep automatically of every transaction. When a visitor's browser requests a Web page, the request goes into the log, as do the records of all Web pages loaded, downloads and so on.

    'This is a huge amount of data, most of it not useful,' said Phil Kemelor, vice president of strategic consulting services at Semphonic, a Web analytics consulting firm. However, one type of Web analytics software parses those logs, sifting them for interesting data, performing statistical tests and creating reports of the results.

    Page tags are scripts that are embedded within the HTML of specific sections on particular pages.

    When a visitor clicks that section, the script sends the pertinent information off for collection.

    The Web log method is more all-encompassing because it includes every transaction. That makes the log method good for tasks such as monitoring Web performance.

    But it is also time-intensive to process vast amounts of data generated by Web logs.

    'If expense is a major concern, Web log analysis is least expensive,' Pagano said. There are several free open-source tools that agencies can try, including Report Magic and Analog.

    The page tag method is more focused, zeroing in on the specific sections you want to examine and producing manageable amounts of data. That makes this method useful for granular analysis. Of course, the results of page tag analysis depend on how intelligently those tags are deployed.

    The development of Web 2.0 sites is a special challenge to Web analytics applications. Neither Web logs nor page tags are good at dealing with Web sites using multiple active frames, nor are they good at tracking whether embedded videos and other content are accessed.

    Some vendors ' such as Coremetrics, Omniture, Unica and WebTrends ' emphasize that they offer Web 2.0 functionality at some level. But Web 2.0 offers such a variety of features that you should inquire about the technologies and actions you want to track.

    Some Web analytic programs rely exclusively on Web logs. Others rely exclusively on page tags. Some applications allow you to collect and use both methods, and some also throw in custom tools for tracking Web 2.0 actions.

    To select the most suitable solution, you should carefully consider what types of information you need about your Web site.

    If you're content with basic traffic numbers, such as the number of visitors per month or the most-used search term, you're in luck. The most basic Web analytics tools will give you those answers for little money and effort.

    Picking the right solution

    But what if your questions are more complex, such as what your visitors are looking for and whether they find it? In that case, you'll need to think hard about how best to tease that answer from the available information. What data would help you answer that? What pages would be the best source of that data? Is it the kind of situation that requires monitoring over time? These are the considerations you need to keep in mind before you even attempt to perform Web analytics.

    Also, bear in mind that nearly all Web analytics vendors offer hosted, software-as-a-service solutions in which they run the software and collect the data. Many vendors have licensed software, where your agency runs the software and handles the data. With a hosted service, your agency pays by the number of page views or server calls.

    SaaS solutions have the advantage of not requiring any software on the premises, which means no administration tasks and lower support costs. A disadvantage is that you don't have direct access to the data or the application, making deeper analysis or customization difficult. 'Agencies may have security or privacy concerns, with data on shared servers, al- though this is becoming less of an issue,' Kemelor said.

    Conversely, if you buy or rent the software and run it yourself, you have more control over the data and the solution. However, you also take on more of the management burden.

    'Different vendors offer different combinations of solution models and data acquisition,' Kemelor said. For example, Omniture, Coremetrics and IndexTools offer SaaS solutions using page tags. Unica and WebTrends offer both SaaS and licensed solutions and use both log files and page tags.

    Keep it simple

    Ease of use is also a major consideration.

    'Often, customers don't even use the product they bought because it's too complex,' said Olivia Sheng, professor and director of the Global Knowledge Management Center at the University of Utah. Some solutions require effort to configure and use properly. Naturally, the tools that provide the most complex analysis require the most setup. Simpler tools, such as those from ClickTracks and Google Analytics, provide fewer configuration and specialization options.

    A hands-on evaluation by users is essential before purchase.

    Price is always a major factor in considering solutions, and there is an unusually wide range of pricing in the case of Web analysis. Prices are tied to features and capabilities, so you're going to need to make an accurate assessment of what you need to accomplish so you don't pay for tools you don't need.

    At the low end are free applications such as Google Analytics. Many other vendors, such as StatCounter, offer free versions of their analytics programs with limited capabilities.

    IndexTools offers its IndexTools Web Analytics Enterprise for as little as $250 per month, which includes 500,000 page views for the year ' perhaps just right for a small agency.

    ClickTracks offers a $1,200 plan that includes software licensing with support ' you do the work ' or more complete services that start at $9,500 per year. Omniture has a base rate of about $1,000 per month, but you'll probably want add-on modules and extra consulting to achieve enterprise-level analytics. WebTrends offers a variety of options for an average of $27,000 per year.

    Unica and Coremetrics are on the high end but focus on enterprise-level solutions. Expect to pay an average of $50,000 for Unica services that include as many as 100,000 page views per month. High-traffic sites could exceed $150,000 per year. Similarly, Coremetrics Marketforce can range from $70,000 to $110,000 per year, depending on traffic.

    The pricing models for Web analytics can be complex, largely because every Web site situation is different. You need to be clear on all the various items that might be necessary to answer your Web site questions, including software, storage, hosting, automated services, live experts, special configurations and reports.

    And be aware that vendors aren't always specific about how functionality is split across add-on modules, often leaving clients unsure which components they really need.

    DeJesus ([email protected]) is a freelance writer who writes regularly for GCN.


    • Records management: Look beyond the NARA mandates

      Pandemic tests electronic records management

      Between the rush enable more virtual collaboration, stalled digitization of archived records and managing records that reside in datasets, records management executives are sorting through new challenges.

    • boy learning at home (Travelpixs/

      Tucson’s community wireless bridges the digital divide

      The city built cell sites at government-owned facilities such as fire departments and libraries that were already connected to Tucson’s existing fiber backbone.

    Stay Connected