Tracking analytics is a polarizing subject. On one hand, its precise
measurements and metrics can provide invaluable, scientifically-tested
information crucial to fine-tuning your website.
On the other, its very precise nature can quickly escalate into a dim
haze of confusing lexicon – a bewildering amount of metrics, and
complexities that surpass even many so-called specialists.
So why bother? What can analytics provide for me and, furthermore,
how is it worth the steep cost – whether that is in money, time, or
sanity.
A Numbers Game
Analytics is all about the numbers. There are several services out
there that serve analytics metrics, such as Google, Adobe Omniture and
others, but the one thing they all revolve around is numbers. Actually
more like tallies.
In its simplest form, analytics are the grown up, lab-coat wearing
children of hit counters. Tracking services tell you one thing in a
seemingly infinite amount of different ways: how many people visited my
site?
But it gets deeper than that. And do buckle up for this.
Let’s say your website received 10,000 hits in one month. Simple
enough, right? But wait, just in navigating a website a person is likely
to visit the same page more than once. Maybe a single person visited
your homepage several times in the course of their visit. OK, so 10,000
hits isn’t quite accurate enough. You need to know how many people came
to your site, not just how many visits were recorded.
So it’s several thousand people. But wait, how often did a person
visit? It’s one thing if they came back to the homepage multiple times
in the span of an hour, but another matter entirely if they went to the
homepage the next day. How long can a “duration of a visit,” or a
session, be defined?
For that matter, how much time have people been spending on your
site? Seconds, minutes? Milliseconds? If it’s the latter, then you’re
being visited by search engine spiders and bots, not people. That
doesn’t really count as “web traffic.”
But even on the other hand, if all those numbers were of genuine
people, are they spending a healthy amount of time on your website or
are they simply stumbling across it and then ducking back out? Can a
visit of a few paltry seconds really hold weight against a session of
half an hour? The latter is quite a visit, while the former can barely
qualify as one.
For all these people, spending at least a decent amount of time on
your site, are they spending it only on the homepage or are they taking
the time and effort to explore the other sections of your website. How
can you be sure these visitors are truly seeing what you have to offer?
Are they coming into peruse your wares, or essentially window shopping,
practically bouncing off your homepage as fast as they came.
So from just this most basic of levels we have unique visitors,
sessions, duration, and bounce rate. See, not just mumbo jumbo after
all.
Proving Its Worth
Why would a website need to keep track of what browser people use to view it? Aren’t they all the same?
Not quite. Obviously desktop and mobile layouts differ. But did you
know that how the Web is displayed is not truly uniform across all
browsers? The Big Four: Internet Explorer, Mozilla Firefox, Google
Chrome and Apple Safari all display web pages a bit differently. Also,
different versions of those four browsers often display pages
differently from other versions. To make matters even worse, the same
browser running the same version but on different operating systems can
display pages differently as well.
Sometimes these differences are slight, literally pixels off.
Sometimes it is major, like whole portions of the web page not
formatting properly or consistently. This can, and often does, lead to
throwing the entire alignment of the page into disarray. What may look
fine in Chrome and Safari may look off in Firefox. What looks all right
in IE9 may look terribly wrong in IE6. What looks OK on Firefox 12 on
Windows may be only slightly off as seen on Firefox 12 on OS X.
With inconsistencies like this, it’s a wonder anyone can view the Web
at all. At this point, you really ought to effusively thank your web
design team for making sure that doesn’t happen.
This is where tracking really shines: let’s say after looking at your
metrics you notice visitors who use Firefox, Chrome and Safari all seem
to be mostly equal, but IE users are sharply low by comparison and have
a large bounce rate. This is a huge red flag that you ought to check
your site’s compatibility with IE browsers. Chances are, your page isn’t
rendering correctly; perhaps it’s even grossly off. This would explain
why traffic is so low – people get turned off by broken websites.
Testing… 1, 2, 3
In another instance, one of your graphic designers has proposed a new
layout. Some people in your team think the layout is too confusing,
others think that it looks more modern and trendy and will better
increase traffic and visitor retention. Which do you choose?
Obviously either decision is risky and could jeopardize your site’s
traffic. If only there were some way to test which version people liked
more before permanently deciding on one design or the other. But wait,
you can.
It’s called A/B or multivariate testing. Half your incoming audience
sees one version, and the other half sees the other layout. Now you can
judge how people react. If the new design seems to retain more visitors
and exhibit more in-page exploration, then you can easily make a
decision.
Heatmaps can literally show you what people tend to click on the
most. If you discover people are frequenting one portion of the layout
over the other, the question to ask yourself is: why?
Perhaps the other sections aren’t clearly labeled, or are
inconveniently placed. Or, perhaps something about that one section is
just really attractive to your target audience, so why not take that cue
and make your whole website more like that?
Science: Always Worth It
Analytics can be tricky to set up and even trickier to interpret. It’s very much all about statistics and constant fine-tuning.
Nevertheless, it provides an invaluable tool to glean feedback
through user behaviors, detect and isolate potential trouble spots and
perform much-needed testing.
The examples covered here are but the most basic, simplest examples
of what web analytics can do. Like any instrument in the hands of a
master, it can manipulate your medium to no end, limited only by the
imagination. It can perform tasks you may not have always conceived of,
but now surely cannot live without.
What are your thoughts on analytics? Quantum mechanics voodoo,
over-hyped statistics, or something no website at any scale can live
without?
Vince Ginsburg, Post from: SiteProNews
