Let’s talk about metrics, shall we? I mean, isn’t it one of the elements you use to judge the performance of your site on? Yes, and no. You should not do that completely. Metrics are nice, and a very good tool to get a quick oversight in performance, but it does not say anything about the quality, or even if it meets the expectations.
I like Google Analytics. I have a lot to complain about Google, but I like analytics, or the old Urchin. Keep in mind, in multiple tests that we ran on different sites ranging from two years ago to two weeks ago, we constantly see that Google is recording between 20% and 40% lower than the actual metrics. So, those numbers are not completely reliable, or completely not reliable. Is it a bad tool then? Not at all! Because information from metrics is best taken with a grain of salt. What I usually do is hold a server session that counts the amount of actual numbers, then compare that with the number of Google and I will interpolate all the extra information from Google accordingly. So if Google tells me I have 10 visitors, and 5 are from the US, but my server tells me honestly I had 16 visitors, I might assume that 8 came from the US.
All that information is fun for the first time your site picks up usage. You like to know where people come from, how long they stay etcetera. But it is not correct either. Take a bounce-rate for example. This means how many people leave immediately your website, so they ‘bounce’ off the homepage. This is a apparently bad thing, but does not have to be. Say, in a week, you have one visit a day. Six visits in that week are bounces. You can say then that your site performs horribly because your visitors do not want to stay. But what now if you also have, over all the week, one unique visitor. Then suddenly the perspective changes. This one person comes back every day to check if there is something new. He is anticipating that there will be soon. You have one very loyal visitor, which is excellent. Suddenly your bounce-rate is telling you a good thing, even though your bounce-rate for the week was 86%!
But this information is becoming more problematic once it is handled in the marketing and advertisement world. When you used to buy advertising space in a magazine, especially with a subscription magazine, you knew that your advertisement had an audience. And the advertisements in the front of the magazine would be seen more then the ones in the back. And also with television you would know at least how many people would be watching – roughly. But now with internet, numbers do not account for that anymore. Sure, sales people will try to differ, but since every pageview that they can summon to give you the impression that every page is visited, and visited well, a lot of fraud is being committed, also by the most trusted companies.
And to be honest, it is very easy to increase the numbers of visitors on your site. The oldest trick in the book, is to visit your site yourself. If you do that through a proxy server or a relay, it actually may look like your visits are from all over the world. But if you know even a little bit of basic programming, it is easy to deliver a site many, many visits, all from ‘unique’ users. This is also why you should never buy a visitors-subscription, in which companies promise you 10,000 or more visits on your website in a month. It is easy to make that happen.
But it is not only because of fraud, but also because a lot of advertisements are loaded up, but never shown. For example, FireFox has a very popular plug-in called Ad-Block. 40%+ of all FireFox users are using this plug in which simply blocks out all the banners. On a site like www.arstechnica.comthat has 40% FireFox users that simply means that 16% of the visitors guaranteed will never see the banners. Probably even more. The webmaster does not know this, since his servers still send out the ads to these FireFox browsers, only FireFox never shows them. So you would pay for those impressions, but they were never there.
A little while ago we had to deal with a major television broadcasting network that provided us with excellent metrics on the site; an average of 7 pageviews and more than a 23 minutes as an average on the site. This sounds amazing, especially for marketing purposes and selling advertisements, until you sit down and take a good look at the site. It’s main feature was viewing aired episodes online form very popular tv-series. If you would like to view one, it took you through 7 pages to get there. There were 2 kinds of shows, the half-an-hour ones (20 minutes) and the hour ones (42 minutes). Suddenly the metrics of the whole site became less special. Because it showed that the pages where the video was shown got the most attention and longest time span. But they sold advertisement on all the sites that you had to browse through before you could see the video. And since you would only receive the site’s averages, you would not see how long people were spending on the page you would advertise on. When I asked about the information for certain pages, they told me they could not deliver that information. Well, any webmaster also knows that this is not true, with even the most basic packages you can deliver the information per page.
If it is about choosing the right site for your advertisements, it might be wiser to do a little research. Don’t be tempted by a lot of visitors. One client I work for has a banner on the top of the homepage of a major web site, and all the 16 portals of this website, generating between 500,000 and 1,000,000 views per portal per day. These views are verified. But this client has only 2,300 visitors per month. He tries to change the advertisements, tries to be angry… but nothing helps. I try to convince him that the audience is absolutely incompatible. The people who are usually interested in the product that he sells is not the same as the audience of the website. It is like selling groceries on playboy.com .
Use your mind, find out if the site you want to advertise on has indeed the right audience, but also if the site ‘feels’ good and then just try it out. Don’t try to go for the most views of your advertisements. Try to go for the most compatible views from your audience. Don’t fall for all the marketing sales speeches. Those numbers mean absolutely nothing.