Saturday 17 March 2007

More into Analytics

I am continuing my research into Analytics and changing the way I think about web development to make my choices based on real ideas and not just what I think might be nice.

I guess it is part of the internet growing up at the moment. Some people taking a step back and looking at the work they are doing when it comes to building new pages and spending money on them.

I have worked for so many people that just want to make a page the way think looks good or will do what they want. And if they are paying and not interested in anything else then that is what I will do. But if there is a chance they will look at what was working before and measure the change. Or work out a brief or business case for the change and write it down.

Rant over for now, it just becomes so much more interesting when you track real metrics rather than web stats. Web stats show you what computers are doing with your site. In my view Web Analytics start to look at the people.

I have been looking for articles on or books or anyone to give me some sort of education or direction on how to take things to the next level and away from a techie hobby and into real valuable work. I had found a nice website on Usability which is the area i have been improving my code on to be better for users and SEO. In the site he had taken about 5 modeled people and taken their views and use of a website to give guidance as to making the we work for them. The best bit or the twist on most imagined people articles, is that these are Personas for real people. Modeled to represent the strange needs and attitudes of real people. So it is not person A uses IE on a PC and person B Firefox on OSX, it was more like. Person A likes to check out the latest sports news but as he is not very technical and bought his computer 3 years ago he has dated software. Person B has bad eyesite so bought a bigger monitor and set everything to appear about 170% larger (text, menus etc...). This got me thinking about making my sites usable for people with disabilities and using things like screen readers. But I also made the leap to thinking about Personas as a way of breaking down the audice for a website into sample composite people that represent the different groups.
If you look at the users on most sites they can be broken into communities or common character traits. I do not think you need to cover every visitor to a site, but if you get maybe 4 or 5 Personas that span the range of your site, you can watch the effect of changes across your site. You can plan your sites improvements and changes with statistics and against their descriptions. You can also track statistics against each profile to watch for changes in your audience that you would not see in overall trends.

After reading though this Useablity report and making a list of things that could be done for the site i was working on I stumbled upon the FutureNow.com website which connected my ideas for Personas with a way of looking at web stats to get Analytics. They have some interesting samples of the work they do for their clients and one page that lets you download some Excel spreadsheets that gave a series of trend indicators. They are quite simple and I understand they use a wider range of martics themselves but running the last few months figures for 2 different sites into these sheets stated to give out some more interesting numbers.

They do things like break down the number of page views per visit into smaller groups. 1 page, 3-10 pages and 11+ pages (well i added the middle one.) This way i can get a figure as a percentage of 1 page visits / total visits to see if that has gone up or down against the other two. As simply looking each month at Total Page Views and Total Visits tells you nothing about how your site is getting these values. If i add a great page to the site and Views goes up, are people just looking at that page or are they also looking at other pages? With the break down you can see if you are converting the massive amount of search generated traffic that give you only one quick page view into longer more profitable visits.

I am still only stating to get enough data after 5 months to work out where next with my findings, but that I can get to the where next is a great step forward. The only other thing I will say qickly is that you can not get any of these useful numbers with a non-cookie/javascript statics reporting tool. I have enjoyed using Google Analytics and for the larger site WebSideStory HBX, they start giving more usable figures as they follow people not just group hit results.

My next stage is to add forecasting in and then start A/B split-testing to meet and beat those targets. The other interesting thing it has brought to my attention the the massive amount of 1page/10second visits to the site and I believe that SEOing your site and Search Engines in general generate this figure. Most of the traffic are not getting the right thing or the right call to view and find more on my sites. This then becomes the thing to focus on converting.

A. by reducing this scanning traffic so bringing up the % of valued visitors.

--> But this makes the site seem smaller in raw data terms. And you will increase the people who are not so interested in you when you increase the traffic.

B. by accepting this number may always be high on a content not community site but targeting at it to interest those visitors more and increase Committed and Heavy users of the site.

--> If you have interested them to come to your site it is much easier to interest them further.