As the old saying goes 'if you don't measure it, you can't improve it'. So, any website owner worth their salt is going to have some sort of website performance tracking. Google Analytics is popular. Omniture is good. Yahoo Web Analytics is also good and free like Google's Analytics.
Chances are the website owner will be marketing their site to the hilt these days. Paid adverts, email, social media (Twitter, blogging etc) are all examples of online marketing that take time, effort and money - investment in other words. Likewise, the sophisticated website owner may well be using affiliates to drive traffic and running A/B or multivariate tests to maximise bang per buck. Yet again, if you don't measure the effectiveness of these efforts, how will their impact be quantified?
Glance at the list below. All these commonly used techniques will more than likely have their own tracking mechanism:
- affiliate
- A/B & MV testing
- paid advertising
- social media links
- mobile usage
Tracking codes normally use javascript and typically involve some sort of invisible tracking pixel. A 1x1 px gif is pulled from a central server with a query string payload to report data back to the relevant mothership.
All the javascript involved with these tracking codes can start to get a little bit unwieldy very quickly. Unwieldy and posing some risk too.
First of all, your code can get messy. Easy fix - abstract the code out into one minimised & externalised javascript.
Okay so far? What about the latency of all those 3rd party tracking gifs? Only 1x1 px is tiny but you still have to open HTTP connections to a potentially broken 3rd party...hmm?
Even worse if their tracking code is javascript based and either a) it's broken or b) you broke it! How annoying (and damaging to your business) is a javascript error caused by 3rd party javascript?
Again, the abstracted, minimised & externalised library is a good move here. You can do clever things like start using new versions of tracking codes without having undue pain (http://code.google.com/apis/analytics/docs/tracking/asyncTracking.html).
Having said all that, there are a more nightmares waiting for you. Having spent time and effort on Google Analytics implementations, getting the code working for a wide range of measurements, the final step is to calibrate the tracking. When you get 4 unique visitors, 6 onsite searches, 3 sales, 2 bounced visits and a partridge in a pear tree (seasonal blog post!) on your site, you need to be 100% certain that your analytics code is reporting the data accurately. Not confusing bounces or missing transactions or messing up key outcomes. You need to be sure. I'll make sure your GA tracking is spot on - did I mention that already?
Anyway, having got your tracking code accuracy AND precision nailed, can you be sure of the same QA on third party tracking? Do they even work in the same way? Are sessions recorded in the same way? Is the view of the data consistent and properly calibrated?
Does it even need to be?
I say no. ALL third party tracking nonsense should go straight in the bin and be replaced with one solution to track them all. There are so many similar technology formats that a common tracking protocol is entirely feasible. For sure, any affiliate/adwords/email marketing house should be able to offer a custom report/segment/profile & filter that will achieve the required reporting without extra bloat/risk/hassle/cost.
Until such a beast available we'll have to make do with properly implemented and tidied (abstracted, externalised and minimised) javascript libraries and insist on published and verifiable calibration data from the third parties.
Performance tracking should not eat into our margins.
This effort will continue.
