Sunday, 31 January 2010

Filtering out specific known visitors from Google Analytics traffic

Many times you want to filter out from your Google Analytics data, traffic from specific users or computers. The typical example is traffic originating from company employees or web designers working on new pages.
Normally you would create a filter excluding specific IP addresses (of your known visitors), or using the __utmv cookie setting a custom variable (in case the visitors you want to exclude have dynamic IP addresses). To set the __utmv cookie we will have to use the _setVar() method. As of December 2009 the setVar method wont be deprecated but the _setCustomVar() method will enrich the toolbox.
This is great news as the _setCustomVar() method offers a far richer way to create custom variables. The new syntax is:
_setCustomVar(index, name, value, opt_scope)
Where:
  • Int      index       The slot used for the custom variable. Possible values are 1-5, inclusive.
  • String   name        The name for the custom variable.
  • String   value       The value for the custom variable.
  • Int      opt_scope   The scope used for the custom variable. Possible values are 1 for visitor-level, 2 for sesson-level, and 3 for page-level.
This means that we will be able to define several custom variables and create very sophisticated user classes and filters.
Unfortunately the new variables didn’t make it yet to the filters therefore we will be forced to use the old _setVar() method.

Filtering a specific IP address

This is quite straightforward. Just go to the “Create new filter” page, select “predefined filter”, and set the options like in the figure below, entering the IP address that you want to exclude.

Filtering traffic based on a permanent cookie

The process is a bit more articulated.

Step 1 – Create a hidden page that will set the cookie

Create a static html page in your domain that will contain the following code:
<html>
<head> <title>Google Analytics cookie setting for traffic filtering</title>
</head>
<!-- Note: 'filter_visitor' is the content of the cookie remember this for later -->
<body onLoad="javascript:pageTracker._setVar('filter_visitor');">
<h1>Google Analytics cookie setting</h1>
<p>Put any HTML code you want here</p>
<!-- Insert regular Google Analytics Tracking Code used on your site here: -->
<script type="text/javascript"> var gaJsHost = (("https:" == document.location.protocol) ? "https://ssl." : "http://www."); document.write(unescape("%3Cscript src='" + gaJsHost + "google-analytics.com/ga.js' type='text/javascript'%3E%3C/script%3E")); </script> <script type="text/javascript"> try { var pageTracker = _gat._getTracker("UA-XXXXXXX-1"); pageTracker._trackPageview(); } catch(err) {}</script>
</body>
</html>
Of course ‘filter_visitor’ is your custom value and you will be able to set it as you want. You will need to replace UA-XXXXXXX-1 with your proper tracking ID.

Step 2 – Save the page in your server

Name the previous page with a name like GAfilter.html and save it in your web server (of course in the same domain name).

Step 3 – Create the exclude filter

Go to Google Analytics and create a new filter.
Using the settings in the picture. You will have to fill the “filter pattern” field with the value we defined before. In our example filter_visitor.

Step 4 – Set the cookies on the visitor computers you want to filter

Ask all the people you want to exclude to visit the page you created before:
http://www.example.com/GAfilter.html
This will set the custom cookie in their browsers that you will be able later on to filter in your reports. It is a permanent cookie that has a lifetime of exactly two years. Naturally it will be necessary to revisit the page every time the cookies are deleted in the browser.

Saturday, 23 January 2010

10 essential books for your Web Analytics culture

I just launched a book search on Amazon.com with the keywords “Web Analytics”. As of today I receive 1,212 results!!! Just three years ago the same search would have turned less than 150 results.
Web analytics is becoming a hot topic, and as you might imagine authors are churning out books at an amazing speed. No need to say that the quality of all these books is varying widely.
So what are the books that should be on the shelves of every Web Analytics practitioner? I try to read most of the publications that are coming out and I asked many colleagues this same question. The answer (and as well my opinion about it) was almost always the same:
  • If you are starting web analytics there are many excellent introductory books out there. If you have an intermediate level you will still find interesting information in books.
  • If you want the latest and bleeding edge information, unfortunately, you won’t find this in books. By the time the book is published, the information is already stale. You will have to read on-line information like blogs and newsletters. Additionally you should join the Web Analytics Association and register to the Web Analytics mailing list on Yahoo Groups.
Here are my top choices.

Reference or essential books

1 - Web Analytics an hour a day

Avinash Kaushik 2007
Avinash Kaushik is a leading authority on Web Analytics. He is the well know author of the Occam’s razor blog. The book describe how to implement a successful Web analytics strategy and show you insights and techniques that will help you develop a customer–centric mindset.

2 - Web Analytics 2.0: The Art of Online Accountability and Science of Customer Centricity

Avinash Kaushik 2009
The book provides specific recommendations for creating an actionable strategy, employing tactics for truly listening to your customers, achieving optimal success by leveraging experimentation, applying analytical techniques correctly and solving challenges such as measuring social media and multichannel campaigns. A great book.

3 - Web Analytics Demystified: A Marketer's Guide to Understanding How Your Web Site Affects Your Business

Eric Peterson 2004
Eric Peterson is another leading authority on the subject. The book teaches the fundamentals of web analytics in a storytelling way. Even though the book was published in 2004 the contents are still quite actual.

4 - The big Book of Key Performance indicators

Eric Peterson 2006
This is a great reference manual about KPIs. Each KPI is analysed and detailed on the way to use and the person that should use it. This is quite important, as the CEO needs different figures than the marketer. The book is only available in pdf from the author website.

General analytics strategy

5 - Actionable Web Analytics: Using Data to Make Smart Business Decisions

Jason Burby, Shane Atchison 2007
Even though the book targets more Web Analysts working for large enterprises than Web entrepreneurs, this book is great reading for everyone. It introduces strong business methodologies and frameworks that should be used as often as possible. The chapter about “Monetizing Site Behaviours” is particularly good.

6 - The cult of analytics

Steve Jackson 2009
Fantastic book on how to create an analytic driven culture in your company. It includes different methodologies on how to create and improve KPIs and the REAN model (reach, engage, activate nurture). It teaches how to get detached about the tools and how to focus on insights.


Page optimisation

7 - Landing Page Optimization: The Definitive Guide to Testing and Tuning for Conversions

Tim Ash 2008
This is an incredibly useful book. Not only it helps you to understand why the landing page is so impacting on the final actions of your visitors but also provides you with excellent guidance for analyzing your current website and determining how to make it easier for the visitor to use and, more importantly, easier for the user to do what you need them to do.

A/B testing and multivariate testing

8 - Always be Testing: The Complete Guide to Google Website Optimizer

Bryan Eisenberg, John Quarto-vonTivadar, Brett Crosby, and Lisa T. Davis 2008
A well written insightful and action oriented book about testing. Even though the examples are specifically focused on Google website optimizer, the principles can be applied to any other tool. The book is divided in 3 parts: the Why of testing, the What and the How of testing, and the technical aspects of testing.

Usability

9 - Measuring the User Experience: Collecting, Analyzing, and Presenting Usability Metrics

Thomas Tullis, William Albert 2008
This is by far the best and most comprehensive book I have found about measuring usability. The language is clear and written for a broad analytics audience. Everyone who is confronted with usability aspects (and most Web Analyst will, sooner or later) should read this book.

Google analytics

I was a bit hesitant about citing in the list vendors specific books, but Google Analytics is so widespread (it’s free as well), that any aspiring analytics Ninja should work extensively with it. Also keep in mind that tools evolve quickly and books might get outdated fairly rapidly.

10 - Advanced Web Metrics with Google Analytics

Brian Clifton 2008
This book is a gem. Not only it gives a wide introduction on Google Analytics but it also details several hacks that you can use to customize your data tracking. It has also several chapters dedicated to customizing javascript tags to adapt the tool to your specific needs. The book was written in 2008 and lacks few of the latest features of Google analytics that came out in 2009 (like event tracking). It should be nevertheless on every Google Analytics user desk.

Afterthoughts

Do you have any other books you would like to mention? Any different opinion?
Tell me :-)

Saturday, 16 January 2010

BootCamp 3.0 MacBook Pro Sound and microphone problems. Here is a solution.

Starting with Mac OS X 10.4 “Tiger”, apple has provided BootCamp: a support for installing Microsoft Windows operating systems on all Apple hardware equipped with an Intel processor.
Let’s face it; BootCamp quality has always been quite poor, often with outdated and crude drivers. Apple in these last few years has consistently ignored thousand of loyal customers screaming their lungs out demanding for a decent Windows support on Mac hardware. You just have to see the amount of complaints on the BootCamp Apple forums.
Anyway, the new low has been reached with the release of BootCamp 3.0, corresponding to the release of Mac OS X 10.6 Snow Leopard in September 2009. The sound drivers provided are not working with the new line of MacBook Pro (the mid 2009 generation, sporting a CirrusLogic sound chip). Basically the internal microphone is not working and the volume of the loudspeakers is painfully low. To add to the annoyance, the headset connector, that has an integrated optical sound output, is constantly turned on turning the MacBook into an expensive laser pointer. The trouble here is that it is virtually impossible to find generic drivers that will work on the new MacBook Pro.
Almost 5 months later there are no official solutions on sight. Instead there are a couple of well-tested handmade solutions that are working like a charm.
I will assume here that you have a partition with a flavour of Windows and BootCamp 3.0 drivers. Should you use solution 1 or solution 2? Well, it depends. Solution 2 sound drivers are slightly more up to date. With solution 2 you will also get updated drivers for the new Magic Mouse and the track-pad.
If you are new to Windows or you are a bit scared by the length of solution 2,  go with solution 1.

Solution 1 (the easy solution, quick and dirty)

Few folks at Sussex County technical School packaged a set of working drivers into an installer.
You will have to:
1) Start in BootCamp with windows
2) Download the file at:
3) Unzip the file. You will get
  • CirrusAudioVista32.exe
  • CirrusAudioVista64.exe
  • CirrusAudioXP.exe

The Windows Vista drivers will work perfectly on Windows 7. Even though Windows 7 is not officially supported by Apple. I didn’t test personally the Windows XP drivers, but I have been told that they work fine.
4) Double click on the appropriate exe file (according to your operating system) and complete the installation.
5) Reboot the MacBook Pro and enjoy Windows with sound

Solution 2 (a bit more complicated but with additional rewards)

In November 2009 Apple released BootCamp 2.2. A maintenance release of BootCamp 2 (running on Leopard) with upgraded sound, trackpad  and Magic Mouse drivers for laptops.
This sounds great, but installing BootCamp 2.2 over BootCamp 3.0 won’t work very well. We will have to manually unpack BootCamp 2.2 and extract the individual drivers.
1) Start in BootCamp with windows
2) Download the BootCamp 2.2 installation file at: http://support.apple.com/kb/DL967
3) Download and install 7-zip at http://www.7-zip.org/
The BootCamp 2.2 installation file is packaged using 7-zip
4) extract BootCamp_Update_2.2.exe with 7-zip to BootCamp_Update_2.2 on your desktop
5) You will find several files in the folder including:
  • BootCampUpdate32.msp
  • BootCampUpdate64.msp

Extract again with 7-zip the appropriate file according to your Windows version (32 or 64 bits)
6) You will find a folder named BootCamp24ToBootCamp223

Inside you will find several files including:
  • Binary.Cirrus_Audio_Bin - New driver with working audio and microphone for  Macs with CirrusLogic chipsets (the new 2009 MacBook Pro)
  • Binary.MultiTouchMouse_Bin – Magic Mouse drivers
  • Binary.MultiTP_Bin – Improved multi-touch track pad drivers
  • Binary.TrackPad_Bin – improved regular touchpad drivers
Unpack again each of the files above with 7-zip, each one to a different directory.

7) In each one of the directories that you just created you will find an installer: DPInst.exe. Run the different installers in older to install the different drivers.
8) Reboot your MacBook in Windows, now with working sound, microphone, MagicMouse (if you have one) and trackpad.

I hope that my explanations were not too confusing. Let me know if you encounter any problems.

Friday, 8 January 2010

Critical review of The UBC Award of Achievement in Web Analytics

The UBC Award of Achievement in web analytics is one of the first university courses available in the growing worldwide offer of Web Analytics trainings.
I finished the program in November and after a couple of months of meditation I wanted to share my experience about it. My final score was 97% equivalent to an A+ (I just received my official transcripts).

The University of British Columbia, based in Vancouver, delivers the course 100% online. This means that you can follow the course anywhere on the planet as long as you have an Internet connection.
The course and the way it is delivered has many pros and cons. What you will take out of it depends on your commitment, on your learning style and to a series of factors out of your control that I will detail later.
The program is structured in 4 modules:
  • Introduction to web analytics
  • Web Analytics for Site Optimization
  • Measuring Marketing Campaigns Online 
  • Creating and Managing the Analytical Business Culture

Once you take the first introductory course (a prerequisite to the others) you can take the rest in the order that you like. If you can demonstrate a sufficient experience in web analytics you might get an authorisation to skip the first course.

The course timeline

Each course last 4 weeks, in other words depending how busy is your life you might finish everything in a quarter (allowing a bit of course overlap). My advice is to plan and book early, if possible few months in advance, as the classes tend to fill up very quickly.

How hard are they?

It depends. If your background is not IT, technology or marketing you might struggle a bit. If you are very comfortable with on-line marketing, the course will be a piece of cake. You nevertheless will need to dedicate several hours per week of your time to read the online material and go through a little bit of personal research.

The students

The classes are quite small for an online course: 20 to 23 persons. As you will see later this is detrimental for the learning experience. In the courses where I participated there was a great diversity of backgrounds.

The tutors

The tutors are responsible of answering to student questions in forums, animating discussions and correcting the homeworks. As you might imagine a good tutor can greatly enhance the learning experience. Here UBC should do a bit of quality checks because while one of the tutors was exceptionally good few of the tutors where not properly doing their job. I had the pleasure to work with Jeff Young, in course 2 (for sure the best tutor that I had in the program).
Few of the tutors were practically absent, not answering student questions and giving extremely concise comments in the assignments corrections (I was wondering if they were reading the assignments at all).

The textbooks

The material is overall good even though it shows the signs of age. A lot of it (including references and external links) is from the 2001-2004 period. In Web Analytics terms this is like studying advanced microelectronics with a textbook from the 70s. In other words do not expect to learn about the latest and the greatest methodologies from this course. Rather expect to give yourself a decent foundation on which to build later on.
Even though I have several years of experience with web analytics, I have to say that I learned many things from this course.
Since the course is targeted to a very wide audience, the technical aspects (like data integrity and data collection problems) are brushed up very quickly. Additionally very little is mentioned or explained about the mathematical and statistical know how that should be present in the web analyst background.
There is no emphasis on the web analytics tools. This is very good as you are forced to understand the ins and outs of the analysis rather than pressing a button without knowing the internal computations behind.

The classroom discussions

Each lesson has several forums on which the students can ask questions about the classes and engage in discussions with their pears and the tutors. Unfortunately most of the students couldn’t care less and you find yourself alone (most of my comments being unanswered). In the end the same two or three students (probably a bit more motivated than the rest) are left to discuss among themselves. This would be less of a problem if classes were much bigger, allowing a statistical number of motivated students to animate the discussions. Here UBC should learn from other universities that have a successful distance learning program. As an example I can cite "Warwick Business School Distance learning MBA" where online discussions are an incredible learning tool.

The assignments

There are 3 types of assignments:
Discussion assignments: these are literally a joke. You have to discuss about a set problem or a topic in an online forum. The first student will open the discussion with a comment. The other students will build and comment based on the previous forum posts.
No need to tell that the usual two or three students that are regularly discussing in the forums will open the discussion few days before the deadline. The majority of the students will post their messages in the last few hours (some of them in the last 15 minutes). The quality of the comments goes from insightful to irrelevant, especially because the last few students in the discussion have very little to add left (everything has been probably already said) and are pressed by time.
Projects: These are by far the best assignments and the ones where you can learn the most. You are given a topic, a web site or data to analyse, and you research and crunch the data at will, using all the knowledge you acquired in the course. You have a limit in the number of words you can submit; therefore you have to be concise and straight to the point.
Drop box assignments: these are more simple assignments than projects where you will have to apply a very specific set of chapters from the course material. You will have as well to do some additional research. It’s a very good way to learn.

Conclusion: The good, the bad and the ugly

The good:

  • At its present price the course is a bargain as each of the 4 courses will cost you 675.00 CAD (Canadian dollars) roughly equivalent to 435€ (Euro) (1740€ in total). You will get a discount if you are a WAA (Web Analytics Association) member lowering the cost of each course to 640.00 CAD (check UBC site for updated prices). Considering how expensive professional courses are nowadays this is for sure a great plus of the program.
  • The course gives very strong foundation on which to build your knowledge. Nevertheless you will need to complement this course with practical experience and personal research.

The bad:

  • The course is a bit light on technical and statistical aspects.
  • Some of the course material is quite old.
  • Several external links in the course are broken or pointing to paying material.

The Ugly

  • The learning experience can be affected by inactive tutors or unmotivated students.

    Final thoughts

    If you are beginning in Web Analytics and you are looking for a way to build a solid background this is the course for you. If instead you are a seasoned professional looking for a certification, there might be better options around. For example the upcoming WAA certification, that will be launched in May 2010. Alternatively you could be interested in tool related certifications like the Google Analytics Individual Qualification (GAIQ) or the many certifications offered by the big vendors (like Omniture or WebTrends to name a few).
    Probably the whole program would be greatly improved by making the classes longer and increasing the amount of exercises. Another possible thing would be to let the students analyse traffic on live websites, for example from charities or willing institutions.