Back to basics with Google Analytics

So I’ve been using Google Analytics for quite a while now but like most local government types all we produce are the pageview/visitor reports the bosses want. It’s time to change that and use some of the more advance features to analyse visitors not just visits. Here is where the usual trial and error method of learning a product falls short; time for a more organised approach and I hope writing it into a blog will cement the knowledge somewhere in my leaky memory.

Step 1: get away from the office; the place is too noisy for any kind of study. My boss has kindly agreed to a Wednesday afternoon at “Google University”.
Step 2: get organised; Google provide quite a bit of online study material.
Step 3: get sidetracked; holiday, sickness, new projects, the day job, all conspire to derail the bet laid plans.
GoTo Step 2!

Having managed to get a bit of introductory work done before Step 3, I’ve spent the last day or so organising myself a course based around Google’s “Analytics Learning Centre” – it has all the free resources you need to get started with GA. If you want to go on, they point you toward advanced courses you can buy.

So, tonight it’s reviewing the set up process. It seems like going back to baby steps but worth the effort I think; I remember being frustrated because one of our original properties had no profiles that tracked search data.

Get straight the difference between:

  • Accounts, (Your access point for Analytics)
  • Users, (You add users to an account)
  • Properties, (One or more Resources of the same type – no mixing apps & websites)
  • Profiles, (a distinct view of data from a property)
  • Resources, (Website, mobile application, blog, etc)

(This list took two drafts to get right!)

Relationship between Analytic entities

On the work account, I gather data from a half dozen or so resources. Most people call them web sites (we have no apps yet). Some ofthose resources post data to dedicated properties, some we combine into a single property. The split should be logical but as I mentioned at the start, we set things up in a pretty haphazard way so this is not entirely the case. This is an early lesson and I will map out the proper relationships shortly & see what we may need to change so our reporting is properly organised.

Profiles also got out of hand in the original set up. At the time it seemed a good idea to produce a profile for every consumer of analytics and turn them loose to play with their data. While this may have been useful the one or two more savvy users, most just opened the dashboard and either read off the desired statistic (usually unique pageviews) or ran off a standard report from the same dashboard element. Most profiles were a waste of effort.

Invaluable advice is to have one unfiltered profile. It is too easy to create a property which gains a default unfiltered profile then create more profiles and in a clean up delete that unfiltered one, because “you’re not using it are you?”

After that there are only one or two profiles per property that are worth setting up:

  • External traffic only
  • Profiles to separate out traffic to particular areas of responsibility
    The sensible method of doing this is to organise the site using directories but of course modern content management prefers arcane URLs only a computer could love. For the sake of the user experience and to make your filtering possible, do all you can to get URLs that a human (and usefully search engines) can read.

The only sensible way to create profiles is with filters. At this point, geriatric Perl developers rejoice! Filters make extensive use of regular expressions (regex).

At this stage you might want to generate a set of standard filters:

  • Lowercase the URLs
    Amazingly some people insist on using UPPERCASE or CamelCase in URLs. GA is like Unix and respects case so you’ll get multiple entries in assorted case for the same resource unless you do this.
  • Exclude internal visitors
    Easy to do by excluding your own IP addresses
  • Remove junk
    If your stats are littered with admin type requests

Just blogging while reading the online material has been useful & I’ve got a few things I want to go over tomorrow at work.

Next up is conversion and goals. Something I really hope will make Google a more useful tool so we:

  1. Sit down and set out some concrete goals for the web site
  2. See how well the site is doing delivering them.