Your Online Presence > Strategy & Development

Reading Web Statistics

By Lasa Information Systems Team

Having a website is essential for any organisation these days. Hang on a second – it isn’t just having a website that is essential. Your website must be successful, or it just isn’t worth it – but how do you know your website is successful?

You yourself can make judgements of how beautifully designed the site is, or how well the content is written, but if people aren’t visiting the site, reading your material and taking some form of action then was the effort worth it?

Measuring success is an important part of having a website.  It can answer many questions:

• How popular is your site? 
• How does this change over time (perhaps in relation to real world campaigns and what is in the media)?
• Which parts of your site are most popular?  Are these the parts most important to you?
• Who is linking to your site?  Are people visiting your site from places you didn’t expect?  Could you be targeting those people?
• What search terms do people use to find your site?

All of these things can give you ideas about how to make your site work better in achieving the goals you have for it, and allow you to experiment with changes to the content, organisation and design.

How measurement works

There are a number of ways to measure the success of your site.  All of them are based around the logging features built in to all web servers.

Every time someone visits your site, the web server (probably Apache or Microsoft Internet Information Server (IIS)) has to send the files that make up the pages to their computer.  Each time it does this, the web server keeps a record of what it sent, and where.  This is known as the access log.  (Actually there are usually at least two log files – one logging errors, but I digress.)

Log files can grow very long (frequently larger than all the content of your site!), and defy human interpretation.  However, there are a number of tools which analyse access logs - spitting out reports you can easily understand.

These tools break down into two categories.  The first are software tools you run on your own computer (often the web server itself).  The second are online services, which take over the logging facility for you, usually requiring that you include a small image or snippet of JavaScript on each page on your site.

There are benefits and drawbacks to both:

Software you run yourself

These are generally more flexible than online services – you can customise the reports so they only include the information you need.  The downside is that you have to set up and maintain the logs yourself.

Online services

Conversely, online services take out much of the hassle of setting up the logs, but you are stuck with the kinds of figure they offer.  For some people the downside is that a company has access to their statistics, and some visitors to your site might consider the inclusion of tracking images and scripts to be an invasion of their privacy.

As with most things, you have to balance out the benefits and the costs.

Of course, there is nothing stopping you using both methods.

Software

There are several packages available.  I am going to concentrate on two free and open source packages I consider the best options: Webalizer and AWStats.

Both take your log files and spit out web pages giving a wide variety of statistics.  Log files must be in Combined Logfile Format, which Apache creates by default – you will need to configure IIS if that is your web server.

Both of these packages are configured through a text file, and are run from the command line (in Unix/Linux) or a DOS prompt (in Windows).  You can also schedule them to run at certain times of day, meaning all you need to do is visit the resulting web pages.  Since the output is web pages, you will need to set up the packages on a web server in order to view the results.  If this is a public server, make sure to password protect the results page – otherwise you will find most of your visitors come from porn and Viagra sales sites seeking to boost their Google ranking – strange but sadly true.

Webalizer and AWStats are very similar.  How to choose?  I find Webalizer to be easier to configure, however AWStats reports are definitely superior, and it seems to be updated more frequently by the developers.

NB:  Many web server hosting packages will include either Webalizer or Awstats (or perhaps another system), in which case you won’t have to worry about these things – but don’t stop reading – skip ahead to where I tell you about interpreting your stats!

Online services

The easiest way of getting stats (if you web host doesn’t provide them) is to set up an account with an online statistics service.  These require you to insert an image or some code onto your website, which generate log entries on the service provider’s web server – they then analyse the results for you and present your results on a secure web page.  You can set these up in minutes.

Google Analytics

Google have an excellent service called Analytics which gives nice graphically rich results showing things such as where in the world your visitors came from.

Another nice touch is that you can create goals – pages on your site that you specifically want people to visit.  For example, you might set up a donation page as a goal.   You can also integrate the system with third party e-commerce tools, meaning the stats actually show how much money your website raised, and via which route through your site.  Analytics also plugs in to Adwords https://adwords.google.com/, Google’s online advertising system, allowing you to see how much value Google generated for you.

NB:  Charities can apply to Google Grants for free advertising.

To get Google Analytics you will need to sign up for a Google account if you don’t already have one.

For me Analytics is fantastic value.  You pay no money – the only cost is that Google also get to see all this information (duly anonymised, one would hope) and some visitors may resent their visits being tracked by a multi-billion dollar company.  These caveats apply to any online service – you should include information about any third party web log analysis services you use in your website privacy policy.

Alternatives

There are a number of alternatives to Google Analytics, both free and paid for.  A search on Google for “web stats” will uncover many.

Some sites and blogs that discuss your options are listed at the end of the article.

Reading those statistics

So now, you are all set up with your stats package - but what do all those figures actually mean?

Actually, some of them are quite confusing:

Hits

Since people started measuring how their websites perform, everyone has spoken about how many hits they get.  A hit is a single download of a single file from your web server.  That seems fair enough – however, consider that a single page may be made up of several files – images, stylesheets, scripts.  Each one counts as a hit.  So one page with ten pictures will register as 11 hits.  Don’t be fooled – hits don’t tell you much.

Page views

Most web stats packages try to overcome the misleading hits figure with a figure called page views.  What this means varies from package to package, but generally it is a measure of how many HTML pages have been viewed.  This needs configuring so that the stats package knows which files should be considered pages.

Visits or sessions

A site visit or session measures visits from a single computer, no matter how many pages they view.  Usually there is a time limit, after which a visit from the same computer will counted again.  After all, you want to know if someone came back to your site the next day!

Unique visitors

This will tell you the number of individual computers that visited your site over a given period.

All of these figures can be used to measure how popular your site is.  It is particularly interesting to see how these figures change over time.

Your web stats package will show you these figures for your whole site, different pages, and other assets such as video files.

Important measurements

Other information is gathered that can be useful.

Browser type and screen resolution

You can see which browsers people use to visit your site.  Many web designers focus on Internet Explorer as it is the most popular browser on the internet.  Your web stats can tell you if it is the most popular browser amongst people who visit your site.  It is also possible to tell what screen resolution visitors use – some sites are designed to fit in an 800 x 600-pixel window.  Your stats might show that only a fraction of a percentage of your visitors use this.  This sort of information can inform your decisions when it comes to design.

Visitor location

It is often possible to see where in the world visitors come from.  This is of course very interesting in itself, but can also give you ideas of whether your site is getting across to the right people.  Note that any visitors who use AOL as their ISP will unfortunately show up as coming from the USA.

Referrers

These are the web pages where visitors clicked on a link that brought them to your site.  This is good for tracking coverage in the media, creating relationships with related websites, as well as measuring how valued your site is – a link is worth 1,000,000 hits!

Search terms

What did people type into Google or Yahoo! to find your site?  Knowing this can highlight interesting trends, or show what on your site people are most interested by.  This might influence the future direction of your site.

Return visits

If people found your site compelling and worthwhile, they will come back.  Having a lot of loyal visitors is more important than lots of people that look once then never return.

Visit times and depth

You can see how long people spent on your site in a single visit, and how many pages they visited.  Don’t be surprised to see that a very large number visit for only a few seconds and never go beyond the front page.  They weren’t looking for your site or anything like it and moved on.

Robots and spiders

People aren’t the only visitors to your site.  Search engines use programmes called robots or spiders to visit your site and catalogue the content.  Some stats packages will let you see when this happened.  If you aren’t regularly visited by important search engines such as Google and Yahoo!, you might want to find out why.

Sources of further confusion

Caching

Most ISPs keep a copy of any pages its customers visit in a cache.  This means they don’t have to pay for delivery of pages every time someone visits it.  For you this means you will not be able to count every time someone looks at your site.  This is a fact of life, and a good thing.  The internet would be a lot slower if ISPs didn’t do this.  However, it means you should only use your hits, page views and visits as a guide.

Comparing figures from different sites/systems

Different web stats packages measure different figures in slightly different ways.  If you change stats packages, don’t be surprised to see a large difference in the figures.  Likewise, if you make a radical change to your site, particularly moving from static pages to using a CMS, expect a discrepancy.

The lesson of these two issues is that the most useful part of any web stats is the ability to track trends over time – not say with certainty that 1,245,639 people visited your site in October!

Now that you understand a little more about web stats, you are empowered to really take advantage of the web.  Some good sites for further reading include:

Occam’s Razor – A blog discussing different approaches to analysing web stats

Web Analytics Demystified – Site and blog by the author of the O’Reilly book Web Analytics Hacks

Conversion University – part of the Google Analytics site, selling you both the concept, and the (free) product.

A Few Good Web Analytic Tools - article from Idealware, six American nonprofit internet experts outline what analytics tools have worked well for them.