The Challenge Rambles and riff raff about all this and that

This week’s mandatory reading

Aviansh Kaushik's blog presents "Ten Minutes With….. Mike Moran, IBM", an absolutely MUST for anyone with any level of involvement on Web.

One of my favourite phrases is Mike's reply to "How do you make yourself indispensable to a company?":

I think passion is a big part of it. Find what you are passionate about and go after it–you’ll do a better job. But you also need to find out what is important to your company.

So ABC... yet so overlooked.

What can be considered good figures?

An interesting part of web analytics is to determine if the facts and figures you are getting are good or bad. Of course there are some obvious answers, for example, revenue is pretty much fair and square self-explanatory; the more you sale, the more successful your site is. As a matter of fact Conversions in general might be considered as easy, to understand (higher percentage of conversions - better) although there are some other variables coming into play that should be considered (like what I mention about "out-of-the-web-conversions" on this post).

But the real deal is how to measure all the other stuff going on that gets reflected into metrics. Is a 20 page visit good or is it someone who can't possibly find what he/she is looking for? A long visit depicts interest? If someone goes form point A to point B, back to A, to C, to A, to D and so on, is that user lost, looking for something or comparing? What is the mean time to complete a conversion? The list of questions is endless.

So, what can be done to start getting some answers? Nothing very easy. some very serious "profiling" of a given number of "model" visits needs to be done. There are several ways to achieve this. One way is to have a "user experience" group set up and asking them to achieve different goals and measuring what they do. Of course as with any "sample" you only get an approximate idea of what's going on; but approximate if far better than no idea at all.

The other alternative is to get some sort of understanding of your metrics and as much feedback from the users as you can possibly get. Surveys are a good way to get this; and they can be very useful if you can cross-link it to your current metrics problem since a post of the type "your site sucks, I tried to buy a Wii and failed miserably" is far more useful if you know what the angry customer did before (and after) he gave you the feedback.

Usually customers will give you no clues about what they did on your site, but will tell you what they were after. In order for that non-technical feedback to be of any real use a deep analysis of every negative feedback must be made. You must bare in mind that most people will only give you feedback when angry or frustrated. Use this to your advantage. The other thing is that only a rather small group of people will give you feedback this way. The vast majority of people will simply leave when frustrated.

Another good way to think about this issue is having access to any sort of benchmarks to similar sites as yours. This is something that can't always be achieved, since your competition usually won't be very happy sharing sensitive information with you. In that case you can always try to surf through similar sites and pinpoint key differences in navigational flow / times / ways to access important areas.

One of the best answers, I think, is external consulting. I am not too fond of external -anything- in general if a company has the means to afford having "inside stuff"; but this is one of those cases where the general knowledge a consulting group might have on other sites becomes crucial. Thus a broader view is welcome.

Finally common sense and permanent studying and knowledge acquisition is also key. Since the web in general is in permanent change and evolution, permeated by new trends and technologies what was absolutely true 2 years ago might become totally obsolete and out of date today.

Read, Learn and Listen.