Benchmarking

I'm currently working on some Benchmarks for some major industries, which is very interesting work. However, it often presents a problem with the definition of performance indicators, such as conversion rates, unique visitors, online sales and so on. Having been responsible for providing data to a particular benchmarking company in a previous role, it is clear that the data can often be under or over stated.

I'm therefore looking at consolidating and producing a set definition for each of the KPIs in consideration. This will allow the data to be compared effectively, as opposed to the data being provided based on the definitions used by the company in question.

I feel the potential for a wiki-webanalytics!

Microsoft Targets Explicit Personalisation

I've been looking at personalisation strategies for a number of years, with the end game aiming to improve online conversion rates and processes by treating customers differently based on who they are. Microsoft have started to look at personalisation by using the large amount of data they hold on keywords and website access. The Microsoft Ad Center Lab Demo shows how they would treat customers searching for specific keywords or accessing specific websites, based on the customer age and gender. This can be seen here.

There are obvious limitations to the use of this tool in its current state however. Unfortunately, it is not always practical to assume a specific gender and age group for a specific keyword (except in the case of the "over 60's female car insurance" search term, for obvious reasons!). Take, for example, the people who use this blog (not that I realised anyone other than me read this!):



On this evidence, my marketing messages would assume that my users are Males under the age of 18. However, only 57% of my users are male, and Microsoft predicts that 27.45% of my users are under 18. Thus this segment actually only accounts for 15.5% of all of my visitors, leaving me with the problem of alienating 85% of my visitors........

Closing point: This is an interesting tool, and I for one now know more about my site than I did. Thus as an analytics tool it works, but I'm now sure you can start taregetting your adwords bidding by demographic on this basis.......

Yahoo! Design Pattern Library

Yahoo! has available to developers a pattern library for describing how it deals with certain customer actions. These patterns are interesting for a number of reasons:

• They show how a major internet player deals with customer experience, and the devotion to ensuring a consistent customer experience through all of their sites
  
• The show how they deal with accessibility issues and their devotion to ensuring that their sites are accessible
  
• They provide a template for design patterns which can be used to assist developers and website managers to improve their sites and more importantly the consistency of their customer experience

This last point is extremely pertinent in some cases. Take for example Lloyds TSB Insurance who provide a number of different insurance products to their customers. Compare the look and feel of each of the products and the general customer experience provided by them, and you'll be sure to find that there is a massive inconsistency in there. By using a consistent set of design patterns, this could be easily avoided and the customer experience of the whole site improved drastically.

I for one have already adopted these principles in my work, and have seen a number of equally good applications, not least the Online Retail 2006 benchmark published by eConsultancy

Why women should be paid less than men........

....at Wimbledon, I must add.

Anyway, there's currently outrage at Wimbledon since the men's champion will be paid £30,000 more than the female winner. Tessa Jowell has apparently written to Wimbledon to complain and to 'urge' the LTA to pay the same for both winners.

Normally I'd agree - men have no right to being paid more than women. However, it's important to consider the following:

• There are 128 people in both the men and women draws
  
• Which means that the winner has to play 7 games to win (excluding a successful qualifier winning)
  
• On average, a set lasts 38 minutes
  
• The men play best of 5 sets, whereas the women play best of 3

On this basis, the following are true:

Total Sets Played

_	Min	Ave	Max
Men	21	28	35
Women	14	17.5	21

Total Minutes Played

_	Min	Ave	Max
Men	798	1064	1330
Women	532	665	798

Total Hours Played

_	Min	Ave	Max
Men	13.3	17.73	22.17
Women	8.87	11.08	13.3

And therefore:

Earnings per hour ('Hourly Wage'?)

_	Min	Ave	Max
Men	£49,248	£36,943	£29,544
Women	£70,462	£56,407	£46,992

So on average, the women are earning £20,000 more per hour than the men......

The question is, if a man working in an office does twice as many hours as a women does in the same week, will we start demanding that she is paid the same annual salary? Or will we in fact start demanding that Wimbledon only pay the female winner £409,328, to even up the hourly 'wage'?

On a different note, I think I need a pay rise!

Bon Echo

As a relatively internet savvy person, I'm also a big fan of Firefox, and was keen to have a play with Bon Echo when it was released. Unfortunately, it keeps breaking :( . On the other hand, Microsoft have improved IE a lot with the release of IE7, in particular the handy little 'x' on every tab. All of which means that those people working for big corporations and being forced to use IE can at least get a little closer to Firefox functionality at last.

Numbers Vs Pictures

I've worked with web analytics for a number of years now, the majority of the time for a major high street bank, and have found that the focus for a lot of effort (especially within marketing teams) has always been the number of times that a link has been clicked. I've more recently been working with the clickdensity software from Box UK, with some interesting results, since the output here is pictures rather than numbers.

Some quick things that I've found are:

• Using numbers shows the number of times a link has been clicked only
• Pictures show you where the click was made (in the case of a hyperlinked picture, whether the click was on a face, words, on the edge, in the middle, and so on)
• Pictures show you instances of people clicking where they believe something should be hyperlinked and isn't - something that numbers alone would miss completely
• Senior managers get much more excited by pictures than they do by numbers

So it appears that there is some value in using this form of analysis within an eCommerce business......

Working recently on some analysis for a museum website, the third point above became extremely pertinent. The museum was interested by the number of clicks they were getting to there navigation toolbars, but in fact the majority of clicks were being made on the banner image directly below. By hyperlinking this image, the museum has manage to improve click through rates significantly.

Of course, no business can be run without the numbers to support decisions, but it seems that there may be some very good supplementary tools available to give web analysis an extra level of detail.

About me.....

I'm currently working as a web strategy consultant in Cardiff, allowing me to become involved with a wide range of websites and customers. This blog is aimed at sharing some of my findings, some of my personal (web-related) thoughts, and maybe even some interesting anecdotes every now and then!