Jim et al
Page tagging solutions (Google Analytics, Web Trends Hosted, Omniture, Clicktracks etc) will all give you much lower results than logfile tools - even with filtering. Page tagging solutions require visitors actually LOAD an HTML page, not just an element, script or file from your server . . . AND stay on the page long enough for the Javascript or external image code to be triggered.
Logfile analysis is extremely problematic because unless you have set up your server to specifically exclude the logging of particular directories on your site (esp scripts), then not only will then over-inflate your figures as a result of spiders and bots, they will also over inflate your page views as a result of scripts. This is increasingly important if you are paying for a logfile solution like Web Trends which now charges on by page views.
However, NEITHER page tagging nor logfile analysis will capture activity that occurs beyond your site . . . . Nor are they a comparative measure of how you are performing against your 'competitors'.
I'm currently working on a paper for MW08 in Montreal which proposes a new combinatory methodology for analytics in the museum and cultural sector so i am thinking a lot about these issues at the moment. (http://www.archimuse.com/mw2008/abstracts/prg_335001716.html)
My feeling is that everyone needs to be seriously segmenting their analysis. There is no point reporting some global aggregate figure - it means nothing. A global trend means only slightly more.
If you want to know how successful your VISITING sections of your website are, then you probably are concerned with the volume of visitors that spend a small amount of time on a lot of pages (spending a lot of time on a few pages probably indicates bad navigation and usability problems - especially if they are spending 2 mins on your opening hours page . . . ). You are also probably going to be concerned if these visitors are NOT coming from relatively local geographies. You might also track visits from emarketing campaigns through unique ID tracking from email to your site and back again. You are going to want to know the bounce rates and stickiness. You are also going to want some competitive intelligence.
On the otherhand, you might have a lot of education kits as PDFs. A page tagging solution will only pick up a fraction of the users viewing these PDFs - they would need to visit the placeholder HTML page first to be counted. If your PDFs are well indexed in Google then it is likely that a large number of your visitors will be coming directly from search and will only be visible in your server logs. You might want to know the geographic spread and the number of EDU resolving IPs that download these resources - that might prove a better indicator than anything else.
Likewise, you will probably find that if you have podcasts listed in iTunes then 70% or more of their listener/viewership will not even visit your site - they will grad the podcast through iTunes. It will pop up in your logs but not in a tagging solution. Does that matter? If you want to use your podcasts as brand evangelism and convert listeners to visitors then it definitely does - if they aren't getting to the podcasts via your site then they aren't seeing any of the other guff and branding you have around them . . .
And that's all before we get to social media measurement and notions of 'engagement' . . .
An anecdote . . . I was recently listening to a prominent New Zealand Government representative talking about how they had two public wiki projects. The first one got listed on Digg and got so much traffic and usage in the first 3 days that it had to be turned off and the project stopped. 100K users in a day. The other had barely 50 users a day - and they were the SAME users. It lasted for 2 years. Which was more successful? The second one - the value of the participation from the 50 committed users far outweighed the 300K unique visitors the more prominent project got.
--
The sooner the sector moves away from the notion of global reporting the better. This is going to take a lot of educating upwards to funders, but ultimately a more segmented approach will deliver much better value to your organisation AND more transparency to funders (be they public or private).
Seb
Sebastian Chan
Manager, Web Services
Powerhouse Museum
street - 500 Harris St Ultimo, NSW Australia
postal - PO Box K346, Haymarket, NSW 1238
tel - 61 2 9217 0109
fax - 61 2 9217 0689
e - [log in to unmask]
w - www.powerhousemuseum.com
-----Original Message-----
From: Museums Computer Group on behalf of Jim O'Donnell
Sent: Fri 04/01/2008 9:30 PM
To: [log in to unmask]
Subject: Re: Measuring web site visits
Thanks very much for all your responses.
I've another question about log analysis. One of the graphs produced by webtrends is a breakdown of visit figures by number of pages viewed. For our sites, around 40% of the visitors view 0 pages ie. 40% of visits download files without visiting a HTML web page. Presumably this is requests for images on our server, embedded on other sites, or people downloading pdfs directly via a link on google, say. We get a lot of requests for images from myspace pages and discussion boards. There's also a small number of subscribers to RSS feeds who don't visit the site directly.
Should we subtract that figure when counting the number of visitors to our site? It does include people who are using our content, just not in the traditional manner of visiting the site and reading a web page. Perhaps we need to look at separate services, like feedburner, to keep track of those?
Interesting to note by the way, that if we subtract webtrends' visits from spiders, and visits that view 0 pages, from the total visits figure then the number you get isn't far off the number reported by google analytics for the same period.
Jim
Jim O'Donnell
Senior Web Developer
National Maritime Museum
Park Row
Greenwich
London SE10 9NF
DDI: 020 8312 6517
Fax:
email: [log in to unmask]
P please consider the environment - do you really need to print this email?
We are listed on everyclick.com, the search engine that helps charity. Please go to http://www.everyclick.com/uk/nationalmaritimemuseum and set everyclick as your home page, so you can search the web and help National Maritime Museum. It does not cost a penny, so it's a great way to support us every day.
_____________________________________________________________________
This e-mail has been scanned for viruses by Verizon Business Internet Managed Scanning Services - powered by MessageLabs. For further information visit http://www.verizonbusiness.com/uk
**************************************************
For mcg information and to manage your subscription to the list, visit the website at http://www.museumscomputergroup.org.uk
**************************************************
=========================Important Notice=====================================
This email and attachments are for the use of the intended recipient(s) only and may contain confidential or legally privileged information or material that is copyright of Powerhouse Museum or a third party. If you have received this email in error, please notify the sender immediately and then delete it. If you are not the intended recipient, you must not use, disclose or distribute this e-mail without the author's prior permission. Any views expressed in this message and attachments are those of the individual sender and the Powerhouse Museum accepts no liability for the content of this message. Whilst every care has been taken, the Powerhouse Museum cannot guarantee that the integrity of this email has been maintained nor that the email is free of errors or viruses. The Powerhouse Museum advises all organisations and individuals to undertake their own virus scanning and security measures.
==============================================================================
**************************************************
For mcg information and to manage your subscription to the list, visit the website at http://www.museumscomputergroup.org.uk
**************************************************
|