The Challenge

Rambles and Riff Raff about all this and that

The Quest for a Meaningful Metric

Published by Esteban Glas on December 10th, 2006 | This post lacks all category except for: Metrics, Web

During 2006 we’ve buried page views. The value of this has been in doubt for as long as metrics have been of importance to anyone. Visits are still important, but it’s not the prima donna any longer, since not every visit is as important as the next one. Conversions are key nowadays, tracking them can be tricky, and is one of the most important things to implement while deploying a metrics software, of course a sale is a conversion, but there are all kinds of “successful” conversions in complex sites (software downloads, product registration, successful support, video/webcast watching, the count might be endless). And even visits that don’t “convert” (I feel like a middle-ages witch hunter when I use that word… Convert, brother visitor or thou shalt burn in the flames of a useless visitor pyre) might be meaningful for a site. Say you are into sales, and a certain visitor goes through the sales process and decides he wants to call and buy in instead of purchasing through web, isn’t that considered a success? On the other hand someone might complete a purchase but might cancel the order afterwards, that would not be a conversion.
As hard as metrics software try they can’t read user’s minds (yet), thus expect to have a certain “noise” level in your key metrics and conversions. (An interesting challenge for metrics and analytics guys might be to determine an estimate of how much noise there is on a certain site). Bottom line: Metrics wont ever tell you everything you need to know and much less “the truth, the whole truth and nothing but the truth”. Hence comes into the game the analyst, that obscure guy locked up in “the cave” (wink, wink, Jim).

This guys work is to read the data and make some very important “fill in the gaps” job. A responsible web metrics analyst will contact call centres and find out how  many people were driven there by the web, how many web orders were cancelled and get every piece of non-web-but-related-to-web numbers he can get his/her hands on. Then it’s time to run those numbers against the ones coming from the tracking software. It is then and only then that you can have a more close to reality look at what’s actually happening.

But do not despair, brother web-guru, metrics are key, still, to understand what’s going on on the web and how successful campaigns, adverts and other stuff are. But you must make certain you understand direct metrics are only part of the picture!

Since I don’t want to extend too much on this post (so I have the excuse to further write about metrics in the future and drive you back to this blog) and that the Sunday beer is taking it’s toll on my understanding of life, the universe, and everything I’ll stick to web metrics for the rest of this post (thus still taking a look at only part of the picture).
There are some meaningful and useful metrics out there. For web marketing guys ROI-related metrics are core. If I spend X bucks on google adwords and those come back as X bucks in purchases you can easily make the math and understand how successful that was (of course I’m minimalist on this, you should also take into account what I’ve already stated about success out-of-the-web). That kind of campaign-related metrics will remain relevant.

But to further understand the effectiveness of campaigns, web design and architecture there should be some other things to take into account. How many clicks / pages / time did it take to convert? (metric suggestion) Time is a tricky one, since users might be really interested in your site’s content and read a lot, or your site might be slow and/or hard to navigate: that’s the analysts job to discern. (Usually if only a small portion of visitors take “too long” and the rest can complete the process considerably faster you can be somewhat certain your site is ok).

Means are mean. Averaging in metrics is as useless as it gets. On most (if not any) important metrics you need to segment things in order to understand the difference between user segment A and user segment B (and C and D…). “We’ve had an average 2% conversions the past few months”. So what? out of that “other” 98% how many were not going after that particular conversion(s)? How many where after it but decided otherwise? (and why?). How many converted outside the web? Where did they come from? Out of that 2%, how many “deconverted” (yes, I’m killing English language, thank you very much) afterwards? How long (in terms of page views / clicks / time / visits) did it take them to convert?

Thus, really meaningful metrics are only a result of crossing two or more basic metrics (and outside information), since standalone metrics say pretty much absolutely nothing.  You need to segment and compare in order to get an approximate picture of what’s actually going on. I’m sorry guys, if we want to be of any use to whoever we work for we need to work much harder.

A  very special thank you to Mr Hazen, since he introduced me to all this stuff and taught me 98% of what I know… the funny part is he thinks I helped him!

The Challenge | WP Theme by HSO MM based on Gathering
Site prodly powered by WordPress