Deciding What Not To Track / Measure
Posted 30 August 2004 - 11:16 PM
So, my first question is this: in an industry full of buzzwords like ROI, KPI, CPA, CPE (cost per event, like newsletter signup), rankings, PPC, dayparting, usability and engageability, which of the buzzword metrics do not track and measure, and why did you decide not to track them,a nd which metircs do you track as background?
This all started when we had the discussion on engageability a whiles back, and I couldn't for the life of me see how engagability could be either measured, i.e. a metric applied, nor what such a measurement or metric would do for the proverbial "bottom line". If I measued engageability, how would it change what I do?
IMHO, what you choose not to track can play as great a role in creating a successful online presence as what you choose to measure. This became extremely clear to me when I met with a client that ran super detailed logfile reporting monthly, that created megabytes and pages of reports.
Having created this gargantuan report, nothing about the site ever changed, because although they had all this data, they never actually related any of this back to decision making or indeed tangible changes.
So, for my part, what I rarely, if ever, track is "visit" based metrics like Unique visitors, total page impressions, page views per visit, paths etc etc. The reason for this is that, as an SEO who is usually tasked with driving traffic, these metrics rarely, if ever, tell me anything useful in the scope within which I work. I also don't deal with design metrics, as again, this is outside my scope in 99.9% of cases.
What I do tend to measure varies from client to client, based more on the client's willingness to divulge. SE related traffic metrics are the key factor, although ROI and CPA feature prominently when clients don't mind divulging this information. Some clients like to keep their traffic levels fullstop secret (amongst a variety of other things), and in these cases, for obvious reasoins, very little is tracked.
One SEO mertric that I rarely ever measure is rankings. IMHO, rankings tell you very little that can not be uncovered via other means. What's worse, ranking reports can create the false impression that a site is very healthy when in fact it is under delivering considerably, and visa versa.
What does everyone else avoid measuring?
Posted 01 September 2004 - 08:50 PM
There is quite a lot of information that you can collect about a site, but if it's never used, why bother.
I like to see pathways through sites. I like to see where people arrive, and where they leave. I enjoy figuring out why they came (which words they may have used in a search engine or which link they followed).
But, the most important thing the site can do is to get people to call, or send an email, or to write, or fax, or make an order online. A web site is the start of a conversation. We count how often it enables us to have conversations.
Posted 01 September 2004 - 09:34 PM
A web site is the start of a conversation.
Nice Bill, very nice.
Posted 02 September 2004 - 07:48 AM
If you have a good analytics program, there is very little you need to do. Information from unique visitors to ROI can automatically be tracked.
Some of my favorite metrics are:
Bounce Rates, which are the percentage of entrances on the page that resulted in exits without viewing any other page on the site.
Referrals, I love seeing where people are coming from, helps me understand what to deliver to them. Within that, search terms are fun to look at of course.
Location of visitors, where they are from, etc.
Conversions on all types of levels, from transactions to filling out forms. To pin point information such as the ROI of a specific keyword at a specific PPC engine is powerful. To see the time or the # of sessions to a transaction is also important. To see if PPC, Organic, Email, Direct or Referrals are performing as compared to each other is awesome. I can go on forever.
I feel tracking should be done on all levels, why exclude anything? If you don't want to see the data today, you might want to see it tomorrow. So track and then decide what to use to improve your site.
Posted 03 September 2004 - 11:36 AM
Posted 03 September 2004 - 05:06 PM
What does everyone else avoid measuring?
Well, I think this depends on what type of web site I'm looking at analytics for. I find that most metrics have some sort of value, but it's often a matter of not having enough time to really look at everything, so what do I leave out in order to focus on what's really important.
If it's an ecommerce site, I usually don't spend too much time worrying about page views, page views per visit, geography stats (if geography doesn't matter), browser type, OS type, screen size, all that kind of stuff. I tend to stay more focused on things like ROI on marketing campaigns, revenue by referrers, revenue and ROI from marketing campaigns, conversion rate, revenue per visitor, exit pages, conversion by entry pages, and things that usually have a direct influence on the bottom line.
However, if the site has an advertising income stream, than page views and those statistics do become more important.
Essentially, I think the best strategy is to look at metrics that:
A) Directly reflect the bottom line.
B) Are metrics that you can actually work to improve. Sure, it's helpful to know everything, but what good is monitoring data that you can't do anything about improving?
Posted 04 September 2004 - 08:59 AM
So first step is to define your KPIs by site objective. (if you need more info my friend Jason Burby has done a great job explaining them all on the site objectives. He is also co-chair of the KPI and Standards committee for the newly forming Web Analytics Association)
Second each site objective has key reports that you want to follow. (I just completed the basic, intermediate and advanced reports retailers need. I will complete the other site types shortly.)
Now your metrics will fall into 3 parts - counts, ratios and KPIs. If you want more information on some of the key ratios, Jim Novo and I wrote about them in The Marketer's Common Sense Guide to E-metrics a couple of years ago. It still sells for $108 on CafePress, but if you write me directly I will send you a pdf copy for free as being a member of this forum.
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users