Everyone should love web traffic stats

Tuesday, 01 September 2009 19:20

A couple of news items this morning had a theme of web traffic statistics, hence the theme of today's post.

A lot of people we speak to can't answer the question "How many visits do you get to your website?" - much less know what keywords they used to get there or which is the most popular content on their site. If you have a site, it's pretty much critical that you know what people are doing on it.

One of the first challenges is choosing a measurement tool. Different measurement tools can report wildly different results. This is a big problem for large companies who sell advertising based on audience numbers or any organisation that relies on high traffic metrics as proof of ROI. And as the saying goes "There are three kinds of lies: lies, damned lies, and statistics."

But for many businesses it's not necessarily about the raw numbers - but how they translate into tangible business results. Statistical data should play a key part in any measurement and improvement activity, so pick a tool that gives you the information you need. Test a few out.

We've not used all the tools out there but I've used a few, big and small. Until we come across something bettter, we recommend Google Analytics because a) it's free and b) it gives as much information as most small to medium businesses will be able to do anything with.

In the meantime, we'd recommend an alternative to Google for bigger organisations if:

  • You want to track technical performance factors that Google doesn't cover like server loads and bandwidth usage.
  • You have a problem with Google owning your data - tends to be a bigger issues with bigger organisations for whom this data is considered commercially sensitive
  • You get more than 5 million page views per month per account.

Despite the problems with collecting web stats, they do tell you something, so we recommend clients pay attention to:

  • Overall visits as a trend - is it upward or flat.
  • If you do any kind of promotional work did it increase the numbers?.
  • Are they new or repeat visitors (which is better depends on your business model)
  • Bounce rates if you've got it - if you have a high bounce rate, people are coming - then going again. This is generally a problem
  • Page views and visit durations - tells you how long people stay on your site and whether they're having a good look around. High page view numbers aren't necessarily a good thing if those views are scattered all over the place - it could be an indication that they can't find what they're looking for.
  • Short time-on-page durations are a clue.
  • Top keywords - this tells you what people are looking for when coming to your site.
  • Make sure they get it.
  • Popular content - for larger sites, weeding out the content that no one looks at can reduce your content management overheads, although SEO experts would probably tell you to leave it there but you'll need to balance out whether the SEO benefits from keeping content and the associated links are worth the effort and the impact on visitor experience.
  • Traffic sources - Who is sending you traffic, and which has the highest bounce rate. This will tell you what sites are giving you business - particularly important if you are paying to advertise on them.

One of the things that may become apparent very quickly, is that the site may not be setup in such as way that makes it easy to analyse.

This can be caused by:

  • Obscure URL's. URL's that are search friendly should be easy to read anyway. If they haven't it will also make it a pain to interpret your most popular content if you have to figure out what content the url www.yourdomain.com/18upc_ab_content/holder/special/productidl456ddd_search -72698.html relates to. (Note - this point doesn't refer to dynamic vs static URL's and SEO. Google can handle dynamic URL's, but they are a pain to read).
  • Content lumped together on one page. If a site has a 'services' page with all the services on it, you can't tell which one visitors are interested in Ads that don't link to anything. We've got clients that run ads (either graphical or text) that don't link anywhere (a page, a site, a pop-up) which makes it impossible to get any stats to show whether a visitor had any interest in the product or service being promoted.

At the end of the day your analytics tool - no matter how powerful - will not make decisions for you. The power of analytics is in the decisions you make as a result of the statistical data.

Tags:
< Prev