Web Strategy 1: Web Analytics

Dear reader,

I thought I’d start an irregularly written series of postings dealing with various aspects of web strategy. This first blog post handles with web analytics – a subject that absolutely no-one in the field of web can bypass. Here goes.

Web Analytics

Analytics in the realm of web generally answers to the need: “We want to know what our visitors are doing in our web services”. It measures the visitors online behaviour and seeks to enable a company’s web team to make educated guesses on the analytics reports, and thus improve the service, drive new leads in, analyze and improve Return on Investment (ROI), etc.

There exists an expression “half of the marketing budget is wasted, but we don’t know which half”, which is not entirely true anymore when it comes to web marketing. Web analytics makes the gathering of information and success rate possible based on a company’s set needs and goals.

With most analytics tools it is possible to apply or define filters (e.g. one might want to filter out their own company’s visitors from the general visitor metrics) and goals (“I want to know how many visitors came from Google and ended up buying our product”). With most tools it is also possible to investigate visitors’ navigation paths in a visualized manner. Other information of interest include e.g. the visitors browser and platform information, screen resolution, company name (can be sought by ip number), etc. Some of the analytics applications are more information rich than others – the tool to use should always be decided after defining the metrics and process.

A good analytics solution gathers information on different aspects of the visit, including:

  • Content analysis
    • Top Content
    • Enter page, Exit page
  • Visitor / Visit analysis
    • Path analysis
    • New visitors
    • Returning  visitors
    • Read pages per visit
    • Duration of visit
  • Traffic (source) analysis
    • Search engines
    • Directories
    • Keywords
    • Links
    • Direct
  • Technical analysis
    • Browsers
    • Platforms
    • Screen resolution, etc

Other good traits in an analytics solution are:

  • Customizable dashboards
  • Automatic and customizable reports e.g. by e-mail
  • Export to different formats (pdf, XML, Excel)
  • Integration to Customer Relationship Management software (CRM)
  • Integration to marketing tools such as Google AdWords
  • Lightweight user rights management (eventually you’ll want others to just see the reports but not mess up the dashboards, for example)
  • Visually pleasing and accurate graphs, pies, etc.
  • Comparison to history data
  • Benchmarking or anonymous comparison between similar sites
  • Realtime monitoring
  • Fast reports
  • User friendly GUI

How ever feature rich the application is, bear in mind that you should only acquire an analytics application that suits your  business needs, matches with your budget and so on. If there are no inhouse resources or know-how (as often is the case) I heartily suggest using a professional consultant company to help define the needs and processes for your business.

The definition phase

Before buying any analytics software, one should always first define: 

  • Key Performance Indicators (KPIs) to be followed (i.e. the needs of information to be followed continuously)
  • The process of what happens when a certain metric’s set threshold is exceeded
  • Who does what when a certain metric is succeeded (i.e. who ‘owns’ each metric)
  • Whether to use log based or page tagging analytics software, or both

Different analytics solutions

The main difference between analytics applications is how they actually measure the visits and hits. There are log based analytics tools (often called log analyzers or log parsing tools) and tag based tools (often called page tagging tools).

Most experts think that the most accurate results can be achieved by utilizing best of both. This of course requires careful definition, know-how and resources. Some of the analytics applications are able to utilize both tags and logs. The essential differences between these two measuring methods are:

Log analyzers

Log analyzers, as their name suggests, create their reports from web server logs. This means that:

  • Search engine spiders or crawlers get logged as well (unless delibe-rately filtered out of the results)
  • It is a bit more easy to measure downloaded files
  • In a multi-server farm, there are multiple log files that the solution should be able to combine
  • If acquiring a log based solution, one already has the history data as-suming old web server log files are available
  • Log analyzers cannot be seen e.g. by competitors from page source (as opposed to page tagging)

Page tagging solutions

Page tagging analytics software are inherently different from log analyzers. Some of the features of page tagging are:

  • One has to embed a couple of rows of JavaScript in the end of each page (the tag itself)
  • Page tagging analysis reports are usually read from the vendors web service (e.g. Google Analytics). This might be a showstopper for some security aware (or phobic) companies
  • Page tagging is often used in conjunction with cookies, meaning that it will provide more accurate results and can, to some extent, recognize old visitors more easily
  • If a visitor has blocked JavaScript and / or cookies, their visit will not be logged by the analytics software
  • The same can apply if the user exits the page without letting it load completely, because the JavaScript tag is usually put at the bottom of the source code (just before the closing </body> tag)
  • Page tagging is currently thought to be more accurate measuring method by most experts

There are numerous web analytics software out there ranging from free of charge to pricy applications and from log parsing analyzers to page tagging software (see differences above). Some of the most prominent web analytics solutions are:


2 Responses

  1. Good overview. Few comments:

    – Path analysis: I haven’t really had success with it. It is always promoted as a great tool, but in practise I don’t see very much use for it. It is cool, but not very insightful when you really want to get some improvement ideas out.

    – Log based tools: I know that combining log based analysis with tag based analysis is the “best way”, but not many companies really need those “perfect numbers”. Page tagging has become the ‘de facto’ standard for web metrics so maybe we should stop talking about log based tools? (it confuses many beginners and it is not a big deal in the end)

    … only real strenght of log based analysis is the more accurate tracking of downloaded files but that’s about it.

    – How to start developing your analytics program: I know that every consultant says that you should first develop your KPIs, but I somewhat disagree. In many cases the best way to start would be just to install Google Analytics and start collecting the data and reviewing the reports. Maybe deliver some light-weight analysis and recommendations based on data. When the ‘desire to know more’ is there, then the consultants should step in and start defining KPIs and delivering analysis.

    After all, web analytics should be about building a data-driven culture and building a culture should not start by locking some KPIs and choosing a measuring system.

  2. Hi Perttu,

    Thanks for leaving another thought provoking comment! I’ll add a few lines below.

    Path analysis:
    Yes, to some extent I do agree with this, and e.g. Avinash Kaushik (http://www.kaushik.net/avinash/2006/05/path-analysis-a-good-use-of-time.html) has also pointed out some of the possible flaws in using path analysis. However, as Avinash also comments, it can also be really helpful e.g. in an e-Commerce environment, where there aren’t countless of possibilities by which the users can navigate from point a to point b.

    In one of my previous job’s we used path analysis to quickly see the navigation paths of our competitors that seemed to come swarming to our website during a new product launch. It was helpful as it showed us in a compact form what it was that interested our competitors and how information thirsty they really were.

    Log analyzers:
    This is also true to some extent – small to middle sized companies would rarely need to utilize both log analyzing and page tagging methods nor have the resource in doing so. However, in big b2c companies it might actually make a difference. Another big issue would be for companies that do not want to have their analytics data on any external server (such as Google does).

    Another experience I’ve had, is — as you stated — the need to analyze downloads of files. In companies whose core business lies in e.g. software development, file sharing and therefore analyzing its success might be invaluable. Sure, some page tagging software can do this, some require additional link tagging, but usually its easier just using log analyzers.

    Also, starting up a page tagging analyzing tool will start to be informative in comparative data only after a year or so (or, if you wish to compare annual data, the time frame is >2 years). This is another thing where log analyzers would be of help as they can collect all the historical data if the log files still exist.

    Nevertheless, as I was writing an overview, I wanted to include both methods anyway but I must also confess on using mostly page tagging methods myself these days. 🙂

    Developing one’s analytics program:
    This of course varies from customer to customer, and for some (very) small scale websites I’ve created, that’s exactly what I’ve done — putting up Google Analytics for the sake that perhaps one day someone will come up to me and say they want more out of it.

    What I mean is, I don’t think its always mandatory to come up with a litany of really hard-core KPIs. A few well thought-out simple meters that are bound to the company’s general strategy will do just fine (at least in some cases). But if nothing is defined, there is a risk of one not knowing what to look for in the reports, what to do when something critical happens, who will do what and thus sooner or later one’ll stop looking at all.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: