Why resourcing is difficult for Web Analytics?

Lately, I had to take sessions on Web Analytics to multiple user groups. The types of questions that were asked and the focus from the groups opened my eyes on a problem that we'd never realized till now - identifying the right ‘fit' of people for Web Analytics.

Come to think of it: When you have a Java project, you look for a Java resource. When you have a .Net one - you get a .Net person. But, when you have a Web Analytics implementation project, what do you do? The straight forward answer is - get a person who knows Web Analytics analytics. The question is, what is meant by *knowing* Web Analytics?

Web Analytics - Congruence of Technology and Business
The main problem in narrowing on Web Analytics skill set is that it is inherently different from the traditional technologies or Verticals. To implement an analytics solution, say for a client like Amazon.com, understanding of just Amazon's business model would not be enought. On the other hand, understanding of just the Web Analytics solution is also not enough. A knowledge of any tool will just provide details on how the tool by itself works.

To actually understand and work with Web Analytics, we'd ideally a need a candidate who is:

1. Good with one of the Web Analytics tools
2. Has a business tilt of mind, especially in identifying what things if measured add value to a client

The second point is nothing but an ability to uniquely identify the KPIs for the client's business being driven out of the Web Site. The KPIs could be anything from number of people buying a book (Amazon) or number of people signing up for download of a new White Paper (Lead Generation by News Distribution Sites). A person who can match the tools provided by the Analytics vendor with the measurement requirements of the customer is what we should be looking for.

Current issues
The issue we have now with Cognizant is that we have three different sets of people none of whom are really fitting into the role. We have:

1. Developers who are completely focused on the web analytics tool itself - the way it works and the parmaters it provides
2. Business Analysts who know the client's business but, are unaware on what the Analytics tools can and cannot do
3. Architects who lose themselves into the intrcacies of implementation of the tool itself rather than providing customer solutions

(I was bombarded by implementation of Omniture Web Analytics product by architects in one of the session when the session was all about how to use the tool!)

So, the problem that I foresee is that we have people who have a view on just one aspect of the solution and none of whom are able to envisage the complete analytics package providing value to customer. This is the need of the hour and something that needs to be addressed

My Suggestion
To help overcome this issue, my suggestion would be:

1. Train a batch of freshers and BAs on the Web Analytics products - focus on what the tool can do initially. Then, train the freshers on the implementation aspects.
2. Try and identify people who envisage solutions and train them on designing Analytics solution. Ideally, these are the hands-on exprienced associates - people who have ideally had a stint at onsite and are in the position of tech lead for the projects. The reason for this category of people is two-fold:
1. They have seen how customers think and understand how to think along the lines of customer
2. They are sufficiently hands-on and know what is the problem being solved rather than diluting the problem definition.

Well, I've not got a batch to train as of now in Bangalore but, I do hope to get this type of group whom I can build for future projects!

How to design website to be web analytics friendly?

-Akshay Ranganath

Last week, we received a few requests for implementing Analytics code on over 1000 pages of a client web site. It was while breaking our heads over the sheer amount of mundane task that we realized on the potential of making this process a lot more easier.

Background
Now a days, almost everyone wants to have analytics code implemented on their site. Given that Google is offering it totally for free, who would not want to use it?

So, first thing to do is start with the assumption that your site is going to have an analytics implementation at some point of time. If you see various products out there, the basic things of most products are:

1. a small javascript code asking for insertion of a vendor provided script. This will be s_code.js for Omniture SiteCatalyst. For Google Analytics, the file will be urchin.js


2. another piece or a same piece of code that follows this initial include line for capturing the variable that needs to be reported on. For this purpose, you have multiple option, depending on the product being used:
   
   
   
   1. Omniture SiteCatalyst: Use the custom variables
   
   
   2. Google Analytics: I am yet to learn this!
   
   
   
   

I am assuming that you want to implement a basic analytics solution on the web site. This is the case of most people. (I know it is a generalization but, since this industry is in a nascent stage, there is a lot of information that can be gleaned even with this level of implementation).


To help accommodate this, the pages of the web site should have two important features:

1. A include line at the top of the file, just after the <body> tag. This line in turn can insert the necessary JavaScript code needed for Google Analytis or Omniture SiteCatalyst.


2. A shorter code snippet for pulling in the various custom parameters.

To explain this concept, I'll use the example of Omniture SiteCatalyst. Suppose, you have a news publishing website and you need to measure the following:
1.Page name – s.pagename
2.Section Name – s.channel
3.Headline, if applicable – s.prop1
4.Name of author, if applicable – s.prop2
(Paramters after '-' are the Omniture SiteCatalyst variable)

What I would suggest is a simple implementation of placing the necessary information in a DOM accessible format on the rendered page. So, for example, if the details could be designed as follows:
1.<title></title> can contain the correct page name
2.rest of the details could be put in <div></div> tags
So, my final rendered page had the details in a format like:

<div id='section'>Latest news: Sports</div>
<div id='headline'>Man U win Championship Leage – yet again!</div>
<div id='author'>John Brown</div>

Then, we could write a generic JavaScript to pull in the values. The possible pseudo-code would look like:

function getArticleName(article) {
 if article requested is 'Section' then {
  check if element called section exists
  [document.getElementById('section')]
  if yes, then return this value
 }
}

The final page for could be rendered in a very generic format:

s.pagename = document.title
s.channel = getParameterName('section');
s.prop1 = getParameterName('headline');
s.prop2 = getParameterName('author');

and so on..

Since this is a generic function, it can be easily added in all pages as it is. To ensure the correct usage, it would be a good idea to insert this code just before the tag.

Implementation example
Suppose you are using .Net then the master page concepts fits nicely into the scheme of things. Add the line for including the Javascript (s_code.js or urchin.js) in the master page.

Then, include this master page in all your ASPX pages. And that's should get you a foundation.

Then, design a short javascript having the four lines mentioned about and include it in all your pages just before the end of body tag. As a worst case, the javascript could be empty and do nothing. Still, the pain of updating each page in future is resolved.

The other option could be to use a master footer page. This option would make life a lot more simple.

Conclusion
So, you see implementation of analytics, at least the basic one can be made a lot more easy and painless with a few bit of code designed into your system. Trying to do this after everything is done could be a bit of a pain and unnecessary wastage of effort.

7 Web Analytics Sins - White Paper by ClickTracks

-Akshay Ranganath

Read this nice article on the 7 Web Analytics Sins from ClickTrack. The 7 sins are:

Sin #1: Simple Visitor Counts
Learn the factors that can potentially skew visitor data.

Sin #2: Search Term Popularity
Understand why marketers must concentrate on the quality of visitors a keyword delivers, rather than the quantity.

Sin #3: The Linear Funnel
Learn the reasons why traditional sales funnels can lead to dangerous assumptions.

Sin #4: Data Overload
Know why it’s important to be able to separate interesting information from actionable information.

Sin #5: Relying on Absolute Number
Understand the reason why it’s more important to concentrate on trends instead of absolute numbers.

Sin #6: Relying on Top 10 Lists
Learn how getting stuck in your top 10 referrers can cost you long tail opportunities.

Sin# 7: Technicolor Report
Understand the reason why the way that information is displayed can have a huge impact on ease of use and perception.

For more information visit ClickTracks White paper at http://www.clicktracks.com/downloads/7-web-analytics-sins.pdf

Case Study - How to analyze a Web Analytics Report?

Akshay Ranganath

What to see in your Analytics Report?

Once you have for the Analytics Code onto the site, what do you start to measure? Here’s a short article on it with Google Analytics and a Blog on Ubuntu Linux as an example.

The site used for recording is our own groupMAGNET blog, http://groupmagnet.blogspot.com/.

Reports from Dashboard

Visitor Count

The very first report on the site reports on the number of visits to the web site. (See article on definition of Visit).

So, in the above screen shot, on the 16^th of October, I received 89 visitors to my website. For the range from Sep 28^th to Oct 28^th, this is the highest number. For the period of I’ve also got 213 pageviews.

The question that should come in to mind now is: Why such a sudden surge?

Content Analysis

When I see that that on October 16^th there were so many visitors, I checked the Top content report. It showed something like this:

This page shows that for the 213 page views, 128 were received from just one page. This page is having a URL ending in “3-Months after Ubuntu”. So, this is the page that has created such a huge surge in the page.

Since I know that most people landed here, I want to now know if they actually found the page useful. To verify this, I invoke the report for the specific page by clicking on the first URL shown in the sceen shot above. This results in a page of the following format:

So, this page is telling me that:

• On an average, people read this page for 3:32 minutes. This is a very good time considering that the article is really small.


• But, it also tells me that 99.21% of the users bounced. This means that after reading this article, the visitor to my site navigated to some other web site. This means:




• I am offering something that is of use to a lot of readers (the huge number of views) BUT


• My site is not offering a range of solutions to keep the users hooked on.

Hence, if I were to run huge ad campaigns, etc for some other customers, it could not be a big advantage.

The next question that comes to mind is: How did people land on my site?

How do people reach the site?

To answer this question, go back to the first page – the Dashboard and look at the following report (TrafficSources Overview):

This simple report has the details that shows the mechanism by which people are landing on my site.

So, it says that the most number of users landed on my blog via Referring Sites. A referring site is any site that has a link to this blog. (Sites like Google, etc are treated as a special case and reported in the Search Engines).

Hence, my web site is famous not because a lot of people reached through Google Search but, some particularly important source is referring to my site. Who is this site? To see this detail, click on the view report link.

Here the details shows that the top traffic sources are Ububtuhq.com and Digg.com sites.

Conclusion

From the above discussion, we see that the article “3-months after Ubuntu” has drawn a lot of viewers from the sites digg.com and ubuntuhq.com. Knowing the history of what had happened, I can now conclude that:

• Digg.com and UbuntuHq.com attract good quality viewers for topics related to Ubuntu Linux


• These sites (digg.com and ubuntuhq.com) also have ability to target users who are specifically interested in a particular topic (Ubuntu Linux)


• If I have anything to say on Ubuntu, it is probably a good idea to link the article from digg.com and ubuntuhq.com since it gives me viewers who are actually reading my material. (Coupled with the fact that I got around 10 comments, it also means that they actually read the contents and try to understand it too!)

PS: Google Analytics is a free tool. Anyone with a Gmail Id can get the necessary Javascript code for implementing Google Analytics.

3 Months after Ubuntu...

-Akshay Ranganath

Its been three months since I installed Ubuntu 7 (Fiesty Fawn) on my laptop. Looking back at the way I've used it and issues faced, here's my take on the system:

1. User Interface
The UI is definitely simple and easy to use. Quite nice, fast and robust. Maybe not as sleek as Windows but, it gets the job done. I did not have time to explore various other themes - so maybe, there are some better options out there.

1.a. Missing shortcuts
One thing that I definitely missed (or never managed to learn) was creating shortcuts. For example, in Windows, there is the option of "Send to desktop". I could not find any such options easily.

This meant that everytime I had to open specific directories, I had to go through the entire file structure to locate the folder that I wanted. A simple shortcut would have really helped a lot.

1.b. Locked up menu-bars
This was a real irritant. All of a sudden, menus would become locked. You cannot resize or close windows from the menu. The only way to close a menu is to right click on the tab in the screen and then choosing the close option. For someone used to closing windows by clicking on the standard X option, this is an irritant.

Another area that it really bugs is the difficulty in re-sizing windows. Of course, if window is locked, there is no way to do it. In case where the window is not locked, the option is to re-size using mouse but, somehow it is not intuitive or easy to use. Sometimes it re-sizes nicely, sometimes it does not. Maybe, this can be improved.

2. Application software
The default installed list of applications are quite helpful to get most of the work done. The OpenOffice suite is surprisingly easy and nifty to use. Yet, there were some areas where things felt a bit jarry.

2.a. Document Viewer
This is the standard PDF viewer that comes along with Ubuntu 7. It is quite light weight and does the job of displaying PDFs. The places where it caused problems were:
Selection of text - When text was selected and pasted, the formatting was lost or things like spaces were simply gobbled up. So, words would be joined and I would have to go through the copied contents and re-format the same to get the content in order.

The second area where I had a problem is that the document viewer does not allow you to choose areas as images. So, if I like a graph in a PDF that I want to copy, I'd end up rebooting in Windows, opening Acrobat and then working with the PDF. I was not really happy to do this, but it was the only way to get the job done.

2.b. OpenOffice Writer
I am not sure if I explored it but I could not find the Document Review tabs in OpenOffice. So, if I have to produce a collaborative document, I would still need to use the Windows word. Maybe, I need to search further but, this is a first impression.

Apart from this, I was quite surprised by the entire system. One area where I am really happy with is the fast boot time that it takes. On an average, Windows XP and Ubuntu take about same time to boot. But, once booted, Windows would suddenly start all sorts of Virus scans, updaters, etc which would simply hog the memory and make my system slow. So, I would have to start my system and wait for almost 5 minutes before it got to a stage where it was usable.

With Ubuntu, if I want to get something real fast, I am able to get hold of it almost the moment the system completes the booting. Way to go!

Well, that's about it for this edition of my review on the system. Please let me know your thoughts.