Why is it so hard to measure website traffic?
Today’s topic is a little geeky, but it’s something I’m interested in, and so are a lot of other bloggers, so here goes.
The question recently arose concerning how many visitors and unique visitors come to my blog on a monthly basis. If you’ve ever tried to penetrate the arcane mysteries of this issue, you know it’s an alligator-infested swamp.
There are different reasons why bloggers want to know how many people visit their site. If they’re accepting advertising, the number of visitors is crucial, since advertisers pay by the eyeball. More viewers, higher rates: same as it’s ever been or, to coin a phrase, as it is in print, so it will be online.
I don’t take advertising, but I’m interested in numbers. Everybody who puts a product out there wants to know if people like it or not. Movie producers check the box office; authors check their book sales; and bloggers like to know their visitor stats. You can call that egotistic, but it’s merely human nature.
But how do you get statistics? This is where we step through the looking-glass and enter Weirdville.
There are many services that purport to reliably measure such things as visitors and unique visitors to a website. (“Visitors” is anybody who goes to your site in a given time period. “Unique visitors” is individual people who go to your site in a time period. If I, Steve, go to your site 30 times in a month, I will count as 30 Visitors but only 1 Unique Visitor. This is my understanding of it, anyhow, and I’m sure people will correct me if I’m wrong!)
Most blogs are hosted, for a subscription fee, by a web server that also includes, as part of the package, a statistical service. For example, my web host is a company called Newtek Web Hosting. I can log into their password-protected area and access many different levels of statistics, including my daily and monthly visitors. And it’s been enormously gratifying to see that ever since I signed up with Newtek, more than two years ago, my numbers go in one direction only: straight up. I’m not going to get specific, but let’s just say that for December, 2010 (the most recent month available), my stats were beyond anything I’d ever imagined when I started my litte blog in May, 2008.
Keep in mind, I pay for Newtek; their data is private, so I’m the only one who sees the numbers. Most of us bloggers have private data we pay for. But how do you access someone else’s website traffic? There are a number of free sites where you can do that. One of the most popular is compete.com, which allows you, for free, to compare up to 5 websites at a time and see what their unique visitor numbers are. (The free service is not particularly timely; right now, on Jan. 5, I can only get numbers through the end of November. If you actually pay a subscription fee, I think you get more updated reports.) Now, according to compete.com, through the end of November, 2010, my unique visitors were about 1/12th of what Newtek reports.
One-twelfth! That’s a huge discrepency, and a troubling one for anyone attempting to get to the truth. How do you account for such variation in numbers? That’s what I meant about the swamp.
But wait, there’s more! You can also register with Google Analytics for free to measure your website’s traffic. I just did this for the first time starting Jan. 1, 2011–in other words, five days ago. During that period Google Analytics is showing a number for my Unique Visitors (they call them “Absolute Unique Visitors”) that, extrapolated to a 30-day month, works out to a number that’s almost exactly halfway between Newtek’s and compete.com’s. So what’s going on? Are my unique visitors X, 12 x X, or 6 x X? If you were taking a basic arithmetic test in sixth grade and your results varied by that much, you’d flunk.
There are many reports of compete.com’s inaccuracy. This one, which explains how compete.com “gets data from the surfing habits of two million people”–in other words, from a selected random sampling–also describes how “horribly wrong” compete.com has been proven to be. After interviewing technical experts at compete.com, the author (Matt Marshall) wrote this telling statement: “The Compete guys told us they are more interested in serving general consumer sites (their main customers), and don’t really care about niche sites,” which, of course, is what every wine blog is: a niche site.
Complaints about compete.com’s data litter the Internet, most commonly along these lines: “Compete and similar sites can only estimate traffic and they are pretty much always inaccurate. I personally know websites that receive 10x more traffic than Compete says.”
Another widespread complaint about compete.com is that the free data they provide is only a tease to persuade site owners to subscribe for more extensive, and possibly more accurate, data (which may be why compete.com pitches itself to larger, consumer-oriented websites with bigger budgets). This has led critics to lodge complaints like this one: “Let me be clear here, ALL http://WWW.COMPETE.COM SELLS IS DATA and thus if the data is inaccurate then all they are selling is junk. To pour salt in the wounds of a dissatisfied client….the site has endless links to “upgrade” the use for few hundred dollars. This is a classic bait & switch,” especially if “upgrade” is interpreted as “more accurate measuring.”
After several days of online research, I can tell you that the biggest complaint about compete.com is that their traffic reports are consistently about 1/10th those of other metric providers. As one writer stated, “What’s the best way to determine a site’s popularity? Ask the site’s owner.”
This only makes sense, because a site owner will be getting his data directly from the host, which is the only entity that has full access to the site’s server. A host, such as Newtek, doesn’t have to rely on estimates or polls or tracking or projections or logarithms; it can reliably and constantly track real traffic in real time as it pours into the server.
The reason this is important isn’t just to satisfy bloggers’ egos, or even so they can attract advertising. It’s because if there’s no accepted way to measure how many people are going to a site, then the development of online will be stymied, as publishers, content providers and advertisers stand back, waiting for true metrics to emerge. Measuring website traffic will continue be what it is now: a mystery wrapped in a riddle inside an enigma.
It astonishes me that we–the online world–have gotten to the year 2011, but still have no way of agreeing to something as basic as “How many people visit my website?” This is clearly unacceptable, and a disgrace.