FOR IMMEDIATE RELEASE:
January 3, 2003
FOR MORE INFORMATION CONTACT:
Brian Moloney
Web site measurement A-B-C's
Web site statistics are a nightmare. My background is in marketing, not
statistical analysis. Talk to me about reach and frequency, not about
hits and IP addresses served. Unfortunately, the proliferation of Web
site statistics is the result of the online advertising industry and
their struggle to justify a return on investment. "Sure the click
through rate on your ad was 0.000003%, but look at the number of unique
IP addresses that resolved to Malaysia!"
There are two Web site statistics that provide a good indication of Web site popularity. Coincidently, they match up with two very important metrics in other media - reach and frequency. Reach, in the Internet lexicon, is equivalent to unique visitors and frequency indicates how many time they visited during a particular time frame, usually a month.
How many unique visitors did my site receive last month? This is determined by counting visitors only once. Even if one person visited 500 times, that person is counted only once and equals one unique visitor. This provides an indication of reach.
How many visits did my site receive last month? This is determined by counting the total number of user sessions. If that one person visited 500 times, that is counted as 500 visits. For example, a site with 25,000 unique visitors and 25,000 visits indicates that people are visiting only once during that month and are not returning (indicating a frequency of one). On the other hand, a site with 25,000 unique visitors and 100,000 visits indicates that people are visiting the site an average of 4 times per month, or a frequency of four.
True, there are other metrics that can be useful - page views, length on site, exit page, etc. But for purposes of site popularity, nothing beats the basics of reach and frequency.
It turns out that most statistics are not really accurate anyway. They are best-guess estimates, since the actual data is not available to accurately determine most statistics. For example, no statistics program can tell you how many unique visitors your site received. It's really a count of individual IP addresses that accessed the site, masquerading under the name of unique visitors. IP addresses are numbers assigned to computers when they are attached to the Internet. Your computer has an IP address. Sometimes it stays the same, sometimes it changes. A count of IP addresses is close to a count of unique visitors but different enough to make the measurement questionable. Think of how many individuals use a home computer, or a computer in a college computer lab or Internet café. What if you access a site from work, then again at home? If you have a dial-up connection, each time you connect, your computer may receive a different IP address. Visit a site from home on Monday, then again on Tuesday and you are considered two unique users.
Is your brain spinning yet? Let's talk about caching computers. There are special computers on the Internet designed to make complete copies of Web sites in order to speed up download time. This is called caching and AOL is famous (or infamous) for it. For example, once a night the AOL caching computers might copy your entire site and serve it up over and over again to AOL users without anyone actually visiting your site and being counted. If your site is small and growing, your unique visitors may falsely decrease as your site gets noticed and cached by these computers. There are advanced calculations for estimating traffic lost to caching computers but they require a slide rule and a chalkboard to fathom.
Way back in the middle of the 90's the Internet was touted as the most measurable medium. Wasn't it great? You could really tell who was doing what and when. Besides the creepy big-brotherish undertones, this claim has caused too many good people to become lost in the details. Just because statistics are available doesn't mean you have to analyze them all. Leave that for the pocket protector crowd. Stick with the fundamentals — like how many first-time physician appointments were scheduled online, or how many people registered (and paid) for a stop smoking class through the site. There's where the real value lies.
There are two Web site statistics that provide a good indication of Web site popularity. Coincidently, they match up with two very important metrics in other media - reach and frequency. Reach, in the Internet lexicon, is equivalent to unique visitors and frequency indicates how many time they visited during a particular time frame, usually a month.
How many unique visitors did my site receive last month? This is determined by counting visitors only once. Even if one person visited 500 times, that person is counted only once and equals one unique visitor. This provides an indication of reach.
How many visits did my site receive last month? This is determined by counting the total number of user sessions. If that one person visited 500 times, that is counted as 500 visits. For example, a site with 25,000 unique visitors and 25,000 visits indicates that people are visiting only once during that month and are not returning (indicating a frequency of one). On the other hand, a site with 25,000 unique visitors and 100,000 visits indicates that people are visiting the site an average of 4 times per month, or a frequency of four.
True, there are other metrics that can be useful - page views, length on site, exit page, etc. But for purposes of site popularity, nothing beats the basics of reach and frequency.
It turns out that most statistics are not really accurate anyway. They are best-guess estimates, since the actual data is not available to accurately determine most statistics. For example, no statistics program can tell you how many unique visitors your site received. It's really a count of individual IP addresses that accessed the site, masquerading under the name of unique visitors. IP addresses are numbers assigned to computers when they are attached to the Internet. Your computer has an IP address. Sometimes it stays the same, sometimes it changes. A count of IP addresses is close to a count of unique visitors but different enough to make the measurement questionable. Think of how many individuals use a home computer, or a computer in a college computer lab or Internet café. What if you access a site from work, then again at home? If you have a dial-up connection, each time you connect, your computer may receive a different IP address. Visit a site from home on Monday, then again on Tuesday and you are considered two unique users.
Is your brain spinning yet? Let's talk about caching computers. There are special computers on the Internet designed to make complete copies of Web sites in order to speed up download time. This is called caching and AOL is famous (or infamous) for it. For example, once a night the AOL caching computers might copy your entire site and serve it up over and over again to AOL users without anyone actually visiting your site and being counted. If your site is small and growing, your unique visitors may falsely decrease as your site gets noticed and cached by these computers. There are advanced calculations for estimating traffic lost to caching computers but they require a slide rule and a chalkboard to fathom.
Way back in the middle of the 90's the Internet was touted as the most measurable medium. Wasn't it great? You could really tell who was doing what and when. Besides the creepy big-brotherish undertones, this claim has caused too many good people to become lost in the details. Just because statistics are available doesn't mean you have to analyze them all. Leave that for the pocket protector crowd. Stick with the fundamentals — like how many first-time physician appointments were scheduled online, or how many people registered (and paid) for a stop smoking class through the site. There's where the real value lies.
