Spotting junk traffic
Return to Page 2
Junk traffic can look just like legitimate traffic. In many cases, such junk systems will even set cookies or hold session IDs. This makes it almost impossible to spot directly. Such activity can be inferred by the presence of weird data in your web metrics reports, provided your web metrics systems give sufficient depth of reporting.
A clear sign is people who access your form's "Thank You" page without ever seeing the form. In some cases this page may start to appear in your list of Entry Pages. This indicates junk emailers are accessing your form processing script directly. A possible indicator is people for whom your form is the first page they visit. On the other hand, they could simply be repeat visitors who bookmarked the form to complete it later. Both these signs require that your web metrics system shows visits in detail, or the ability to track user paths through the site, or contains an Entry Pages report. Many popular reporting systems simply don't hold this level of detail.
Another indicator of possible junk emailer subversion is when the number of contact forms (or similar) being filled in exceeds the number of emails coming out of the website.
How bad is it?
I seriously doubt if there is a single web server in the world that is not being probed by junk emailers with some regularity. The chances that your forms are being regularly scanned is also high. If you have very robust security in place, and your forms and form processing systems are secure, you may not be getting statistically significant numbers in your reports. "Secure" in this context means both forms and server systems were updated for security within the last year.
Top-level numbers, such as unique visitors and page views are unlikely to be out by too much. The busier the site, the less impact junk emailers will make on these figures. The conversion rate can be a vulnerable figure.
In some cases I have seen junk emailers cause an apparent doubling of the conversion rate. This can be extremely dangerous because it generates a misleading impression as to the sales performance of the site. In one case I saw plans to rebuild a poorly performing site shelved because it appeared the conversion rate was rapidly climbing.
At this point in time it is impossible to say how bad the problem is in general. Junk email attacks which imitate browsers cannot be detected with ease, and there are no systems to filter this from web metrics systems.
The junk emailers are ahead of us in this game. At this stage the impact is merely the possiblility our numbers could be out by a few percentage points. However, I can remember the days when junk email was so rare I could respond to each one in person and tell them to stop. That was 1992. 15 years later and the vast majority of all email sent is junk. In 2006 junk emailers appearing in your web metrics cause only tiny variations and we can live with it.
What do you think the picture will look like in five or ten years? I have to wonder if we'll be able to measure legitimate activity at all…
Brandt Dainow is an independent web analytics consultant and the CEO of ThinkMetrics. Read full bio.