April 2006

Metrics 101: Count What Counts

How to make the most of Web analytics tools

By Bob Kane

 This article is the first of a series.
Read the second installment.
Read the third installment.

In past sales and marketing training, you've probably heard the statement, "Sell benefits, not features." And the advice makes sense, particularly in the world of high-tech, where potential customers can be overwhelmed by the abundance of features (and don't get me started on the jargon that usually accompanies them).

Marketing organizations face increasing pressure to quantify the results of money they spend on various campaigns. And a variety of Web analytics tools now offer a way to better understand and measure the effectiveness of these efforts.

However, Web metrics specialists can easily blind you with report after report of meaningless charts that still don't tell you what you need to know. Most companies track more numbers than they really need to assess the performance of their marketing programs and few know how to interpret the results they gather. Before you begin tracking every possible performance indicator, ask yourself three all-important questions:

• What exactly am I measuring?
• Why am I measuring it?
• How will I use the results?

The answers depend on the kind of business you run and the goals you want to achieve. For example, a transaction-oriented consumer site may need a different set of metrics than a business-to-business site that aims to inform and influence key decision makers. Whether your goal is to increase loyalty, attract new customers, drive online sales, or a combination of all three, you'll want to understand how data is collected, what it can tell you, and how you can interpret the results to maximize every marketing dollar you spend.

All Metrics Are Not Created Equal

Some sites have unique characteristics that require special reporting, but in general, three indicators matter most:

1. Page Views: Simply put, this number reflects how many times people view a particular page on your site. Whether 100 different people each view the page once or one individual views the page 100 times, the measurement would amount to 100 page views. And the more page views a given page receives, the more important it is to your website visitors.
2. Page Visits: Visits are different from views in that they measure the number of unique individuals who view a given page. Using the previous example, 100 different people viewing a page would count as 100 page visits, while one person clicking on the page 100 times would count as only one page visit. Visits are also defined by a particular time window that can vary, depending on the reporting system that is used. A visit typically lasts anywhere from 30 minutes to 2 hours, meaning that a visitor coming to a particular page in the morning would be counted again if he or she returned to the same page late in the evening.
3. Previous and Next Pages: This information can help you track patterns of activity on your site, by indicating how people are getting to particular pages and where they go once they are ready to move on.

Other results—such as time spent on page, click stream analysis, clicks to page, and conversion analysis—are often available, but may or may not be relevant, depending on your site architecture and business goals.

Know Your Metrics Tool

In order to generate meaningful website analysis reports, it's important to understand how data is captured from your site. Two different methods, client-side tracking and server-side log analysis, can produce different numbers for the same metric. For example, if you're using a client-side tracking program, a user who visits the page in question, navigates to another page, and hits the back button will be counted for one visit and two page views.

On the other hand, if you're analyzing log files from the server, the previous example would yield only one visit and one page view, (since the page served when the user hits the back button would come from the browser cache rather than the server). By understanding the capabilities and limitations of your Web analytics tool, you'll not only better understand the results you get, but also know which questions to ask and what type of answers to expect.

Time Is on Your Side

Many companies mistakenly look only at month-to-month performance to monitor the success of their Web programs. This ignores the role of seasonality. For business-focused websites, the week between Christmas and New Year's is usually dead—you may see only half of your normal weekly traffic during that time. To get a better sense of how your site is performing, you should compare the results to the same time period from a year earlier. You may also want to look at performance over a quarter, and compare the results to those from the previous quarter, when setting performance goals for your site.

Also, when doing any comparisons, it's essential to choose comparable time periods. For example, in looking at the performance of two different Web pages for a five-day period, you may end up comparing a Monday through Friday performance with a Wednesday through Sunday performance. Since Internet traffic is usually lower on Saturday and Sunday than during the work week, the five-day comparison would not be valid. Holidays that occur during reported time periods can have a similar effect. For these reasons, quarterly and year-on-year performance windows offer the best comparisons.

Metrics are supposed to help guide your decision-making about what content works with your goals. Don't get bogged down in the details of particular clicks on particular days for particular pages. Instead, take a long-term approach to understanding your site's current performance, and make changes that you can measure over time to make sure you're moving your site in the right direction.

 Read the second installment of this series.

About the author:

Bob Kane is Tendo's editorial strategist.