Saturday 28 June 2008

Short guide to using Web Analytics Solution Profiler (WASP)

Today, I was just looking at something and stumbled a nifty tool call W.A.S.P. Its a plugin for Firefox and stands for Web Analytics Solution Profiler. Freely downloadable from http://wasp.immeria.net, this tool can give you the metrics being collected from your site - and the actual analytics tool being used on that site.

For someone using tools like Google Analytics (GA) this is a big boon. This is because tools like GA did not have any type of debugger. So, if you implemented a solution, you had to wait until the next day to see if whatever you did was right. Unlike Omniture, there was not 'debugger' tool where you could instantaneously see what you was the information being collected and passed for the report generation. With W.A.S.P. all this could change...

Installing WASP
Intaling the tool is just like any other plugin install on Firefox. Just go to the website http://wasp.immeria.net and choose the tool. It gets installed as a plugin on your browser and simply restart the browser to get the tool working.

Using WASP
Once the tool is installed and active (which it is by default), it automatically starts collecting information about the websites you visit. So if you open a page which has analytics code on it, it automatically displays a notice at the bottom. For example, in the screen shown, the site is using Google Analytics. If you want to examine the details, double clicking on it will open the plugin with the details of the site (shown as the left panel in the screen).

The left panel that opens up will display all the parameters being passed to the analytics tool. This itself would be of tremendous help in debugging the implementations. What is still better is the way the variables have been explained in simple language. To use this, just choose some variable and it will display the name of the variable, the value and a short explanation about what information this variable is collecting. For example, in the screen attached, it shows that the variable 'utmdt' is collecting the value from 'Document.Title'. This I feel would help immensely for beginners. You don't have to refer too many documents. All you need to do would be implement analytics code, use the tool and start tweaking the solution.



Conclusion
The tool is nifty to use, easy to understand and install - and light. It can help in getting quick result in detecting what values you are capturing and verifying if it is correct. Of course, the caveat applies that you are using a tagging based solution. If you are using a log analyzer, the issues of debugging still remain. For a tag based solution, this is a great tool for both development teams and analytic consultants.

Sunday 16 March 2008

phpMyVisites: A Short Review

-Akshay Ranganath


Lately, we've been conducting a lot of tools evaluation for a new client. One of the things that we were looking was to see if any Web Analytics tool could help us in readily getting the reports for our client. Here's a short synopsis of our reporting needs and a mapping to phpMyVisites.

Reporting needs of client

Our client wanted to gather some information on the usage pattern for their intranet site. They wanted to know roughly how many people were accessing the site and the machine configuration of the users. In summary, the main needs were:

  • What are the most heavily accessed pages?
  • What is the total no of people who visit the site?
  • What is the Browser, OS and Flash version on the machines from which the users
    are accessing?
  • Trigger emails of the reports at set periods
  • Make the reports available to be displayed elsewhere

Constraints

The main constraint we had was: No data for reporting should go out of the client network. The information being gathered was on an intranet and hence, client wanted nothing to be sent to external vendors. This ruled out the primary vendors like Google Analytics, Omniture.

The second constraint was on cost. Since the reports needed were quick and dirty, there was no outlay to spend too much (or nothing) on the tool to be used.

The third constraint was the ability to tag. Since most of the products being used within their intranet was PHP / Java based, they were confident that a small change to footer or header of all pages was no issue.

Our Solution

Based on the issue, we suggested phpMyVisites. This is a small and nifty tool that provides various statistics. The main features are:

  • Tagging based solution. So, no need to rely on Log files.
  • Page view and Visit statistics. This would provide information on the load and
    the quality of visitors.
  • Top entry and exit pages
  • New v/s returning visitors
  • RSS feed for the reports
  • PDF exporting capability
  • Emailing ability for the reports.
  • "Heat Maps" - a new feature that showed the places on the screen where most users were clicking.

Implementation steps

The solution is very easy to install and implement. It needs PHP, Apache and MySQL. After that, just unzip, create a database, invoke the extracted package as a Web Application - and just follow the screen. The setup takes about 10-20 minutes. Reports can be generated in under 30 minutes. For our test, we did the following:

  • Installing the tool on a Red Hat EL4 server
  • Setting up tracking on PhpMyAdmin tool
  • Tracking the blog software, Pebble

Pain points

The main pain area is its need for graphics library. For the graphical reporting to work, this tool needs PHP-GD2 libraries. As per documentation of PHP, this is bundled with PHP4.3 onwards. Despite having this version, we could not find the module.

Repeated install failed and the debugging has takes us more than a day. While trying to get it working within the constraints of company's intranet, getting any updates or dependencies resolved is time consuming. Hence, it would be better to try the install at home environment and then work in office.

Conclusion

The tool is a nifty piece of reporting that start telling you about your visitors and their browser information in under an hour. Just make sure that this is what the client wants. It is basic in its reporting as well as graphics capability. No Ajax and no Google Analytics like fancy reports and charts. But, for a reporting need, it gets the job done - and fast.

Project status

Currently, we have proposed the tool but, waiting to see if the client accepts the same. If it were purely upto me, I'd have this tool as a starting point on analytics but, switch to something more powerful in due course of time.

References

Saturday 26 January 2008

Help please: Which Linux distro is best for mom and dad?

-Akshay Ranganath


My Problem

Well, as usual, my desktop running Windows XP got infected by virus and died yesterday (Apparantly, AVG did not help). I was about to re-install XP on it when it dawned on me that ever since I'd installed XP, I was having trouble with the machine.

You see, my machine is an old Pentium, 555 Mhz with 192MB RAM. It takes eons to run XP and crashes with Windows defender, anti-virus, auto-updaters and a host of other software that my parents never knew existed. Everytime I am not at home or at onsite, the machine would end up getting infected. My parents who were trying to dip their feet quite hesitantly into the shores of internet would be then frightened. They'd believe that it was them to blame for the crash.. and totally give up using the computer. Hence, the thought, why not install Linux on the desktop and help them?

My problem though is that I am not sure which distribution to use. Before embarking on that let me give you the skill level of my parents

Parents skill level

Basically, both mom and dad have almost never used computers. The only keyboard they were happy to use was the remote for TV - that too to change channel and reduce volume. Now-a-days though, Mom has learnt to send SMS. Dad manages to read the missed call on the Call registry of mobile. Apart from this, the skill level on computers includes ability to switch on, switch off the computer, type on notepad and sometimes managing to send an email (if the email URL is bookmarked). With the childrental (as in parental) guidance, they can chat.

Requirements from OS

Knowing the fear of computers that parents have, important things for the OS are:

  • should be easy to use with a UI similar to XP - at least, the menu should be on left-bottom with On/Off feature being easily identifiable
  • a browser (guess Firefox is the best - i could always use an IE skin to make it look similar to what they know)
  • small program for creating documents (well, this is more of an overkill, but it is definitely something they could learn)
  • a chat program to integrate with GTalk - if I am not in India, I could (just about could) make them use the chat software to talk to them

That's about the requirements from the OS. With this in mind, I want your help in identifying a good disto to use.

Potential options

Ubuntu: I have used Ubuntu but, felt that it needs a bit more of CPU and RAM to work properly. Also, the GNOME interface is a very HUGE learning curve for parents. So it is a definite no-no!

Fedora: Again, it is quite resource intensive and would probably not work on my desktop. Although, I've not used it a lot, from the look and feel, I could configure KDE - again, will it work on my machine?

Help from Internet: I did read this article but, the guy's parents seem to be quite familiar with computers and manged to migrate to a (unnamed) Linux system. Not sure if it'd work in my case!

I also took the Linux Distribution Chooser but - the quiz seems to believe that the person installing the OS is the end-user. Since this is not the case, the results were not all that helpful.

I also read the article Choosing Desktop Linux Distro but, I'd like a second opinion…

Help Please!

And so, I am writing this blog - could you the readers of the blog help me out with this dilemma - what distro would work for my situation? Please let me know which of the one would serve the purpose. Thanks!

Note:

I'd originally put this post on our internal corporate blog. From some really helpful FLOSS enthusiasts, I got the following comments:

..try out Zenwalk Linux distro ( www.zenwalk.org ). It is pretty light-weight and doesn't need a high configuration at all..

PCLinuxOS ( download it..its worth it) or Dream Linux ( should be available with jan LFY issue) ..A new kid is alinux( peanut linux formerly). screenshots at http://alinux.org/linux-screenshots.html

Did you give Xubuntu a shot? I haven't ... but it is supposed to be pretty low on resource requirement, atleast lower than Ubuntu

"Linux Mint 4 " Is the OS which works Out of the box with Gui Better than Xp and Vista,It will have all Proprietary Codecs Installed and Most common Drivers Installed By Default. It's Built with Ubuntu as a base and Some Good Packages Like Mint-Install , Mint-Update which make Automatic updating simpler than Xp / Vista.

I heard of Something Called Puppy Linux For Low end Desktops , try to google for it for its features, and Also Vector Linux Specially made for low end pentium processors and ram

Well, in the lowest memory max usability area, Damn Small Linux is the best. Its all of 50 MB but contains browser, text editors, music players, etc.

..http://featherlinux.berlios.de/applications.htm

i suggest PClinuxOS. simple, does the job and fits well. Even vectorlinux is a better choice for your parents. there are others like linuxMCE, MEPIS if you want to try. If you want to have large application base, go with traditional (G)ubuntu or CentOS.

Thanks for the great suggestions guys! Looking at the help, I had tried PuppyLinux (review follows). Xubuntu would not boot - and when it did (just 2 times), it did not detect my mouse - so I could do nothing much. I keep getting the error - "Unable to load system description tables" and then some out of memory error. After the second error, the comuter just hangs and does not boot at all.

Short Review of PuppyLinux

PuppyLinux seems to be a really light-weight and useful distribution for really old machines. The total download size was just about 98 MB. At about 1/7th of a CD's limit, this was impressive. I did not actually install t - but, I did try from the LiveCD boot. Here's my opinion on the distro:

Pros of PuppyLinux
1. Amazingly small size
2. Comes nicely with a pre-packaged browser and email client by SeaMonkey. Never knew that it was a derivative of Mozilla - so, it was quite impressive.
3. Default install comes with AbiWord, an excel-type program, calculator, themes, wallpapers..
4. Four different games!!! (How on earth did these guys fit all of this in???)
5. Two different X managers - Xorg and Xvesa. Xvesa works like a breeze and makes your old machine feel like it is the top-of-line latest one!

Cons
1. UI is not exactly elegant. At times, feels more like UI for kids...Especially, the icon of a small pup for the 'Start' button
2. Mouse detection seemed to work only during even numbered booting - on 1st, 3rd and 5th time, it failed to detect
3. Distro says it has ability to save session information to make LiveCD boot faster - but, the save feature failed on all times.
4. Not sure of the continued support for project - it looks more like an effort by one person. (my gut feel, not verified thoroughly)

My Verdict
I'd recommend it for some really old system which desperately needs very small and robust OS. For my needs though, it would not exactly suit mainly because of the non-elegant look-and feel.

Thursday 17 January 2008

Tutorial: Analyzing trends through Web Analytics

-Akshay Ranganath

After a long break from the field of analytics, I am back to it! I was just waiting for the data to be collected on my blog to take this further.

In this article, I'll try and explain the importance of trends and how you can derive valuable actionable (and curious) data from the metrics. The way the article is written is how the thought process went on in my mind. So, it's not like a tutorial but a pointer on "how to think analytics"!

Methodology

All the data for this blog has been take from my personal blog titled Life Unraveled. The blog has been in existence for since 2004. The analytics tool that has been used is Google Web Analytics.

Questions addressed

The trends addressed are for the following two main metrics:

  1. Visitors who arrived from Referring Sites (1)
  2. Visitors arriving through an organic search (2)

Two time periods were used for comparison:

  • Period 1: November 1st to December 15th
  • Period 2: December 16th to January 15th

I know the two periods are not the same - but, I realized that only after I'd collated the metrics! So, please bear with me. Ideally, the periods should be of equal duration.

Following is the results from the data collected. Please see the attached excel sheet for the collection of complete data.

Referring Sites

Referring sites mainly have been Orkut and Blogger. So the comparison is for the three:

Metric Overall
Orkut
Blogger
Visits 36 (8) 16 (4) 9 (1)
Pages/visit 2.39 (1.12) 1.79 (1) 2.78 (1)
Avg Time on site 5 m 11 s (7 s) 2 m 39 s (0) 8m 59 s (0)
% New visits 47.22 (75) 68.75 (75) 0 (0)
Bounce rate 66.67 (87.5) 68.75(100) 77.78 (100)

Overall, this category has shown a four-fold increase in the number of visit and about 2 times improvement in the length of stay on site. The loyalty too has increased with fewer and fewer people 'bouncing-off' the site. All of this is good. The questions to answer here are:

1. Blogger has given me 9 visits with an average time of 9 minutes on the site - these guys are really interested in my site. But, who are they? And how did my blog feature on the blogger's site? If I can crack this, I'll be able to increase the volume and my audience.

2. Orkut has started to perform well. The feature of adding the blog RSS feed to the Orkut profile has proved to be a success. Yet, considering that I have almost 100+ contacts and I am still getting just 16 visits from Orkut (with about 10 new visitors), it is not so good. Somehow, the positioning of the feed URLs does not seem to be very prominent or visitors to Orkut are just interested in scrapping me and nothing else. I need to do more marketing of my blog on Orkut - after all these are the people who know me well - and technically should be on my blog for a longer duration of time!

Search Engine Referrers

Search engine and google are synonymous and this was true on my blog too. For the entire period of Nov 1st to Jan 15th, just one query had come from AOL - and this was the one I had tried! All of the visitors who entered the blog through search had used Google.The metrics for users entering from Google looked like this:

Metric 1st Nov - 15th Dec
16th Dec - 1st Jan
Visits 50 169
Pages/visit 1.28 1.45
Avg Time on site 0m 26 s 1 m 50 s
% New Visits 94 93.49
Bounce Rate 90 79.88

The overall number of keywords too had increased from 43 to 133. My blog had suddenly been noticed by Google. The best part is the combination of 'time on site' and 'bounce rate' combination. It shows that a visitor once having landed on the blog spent an average of 2 minutes (at least gave me a chance!) instead of simply moving away - as was happening in the previous time frame. The obvious question was - What were the keywords that were sending so many people to the site and how were they performing?

The search keys and the performance is attached as the excel sheet. If you look at it, I've classified it as the top 10 search keywords. The top 10 have been split into two groups:

  • top keywords by visits and
  • top keywords by average time on site

The reason for this classification was to gain a better perspective over the performance of the search terms. Just because more people land on my site by entering a particular keyword does not tell me much. But, patterns of key words where people entered a particular search key - and stayed on my site for a long time: these are the type of audience that I need to chase. Others, I need to improve.

Search Keyword analysis

The results for the search keywords are here in the excel sheet.

Analysis for period Nov 1st to Dec 15th

From the keyword analysis for the first period, (Nov 1st to Dec 15th), two things stand out:

  1. The term chankya neeti has performed very well - it has brought in 4 visits. On an average, these visitors have been on the blog for about 5 minutes. This implies they liked the search result and read the blog.
  2. By the 4 term, the average time on site drops to 0. This implies people were not happy with the results and they left almost immediately after hitting the site. Most of these search has been for the quotes on the book By the river Piedra, I sat down and wept. Somehow, the blog is not displaying the expected result.

Analysis for period Dec 16th to Jan 15th

Comparing this with the next period, there is a drastic change:

  1. The total no of search keys have increased from 43 to 143. Almost 30 of the key terms have given me an audience who have spent at least a minute on the blog - a drastic improvement!
  2. The most successful key (high visits + longer time) have now changed. Now they are for the quotes from the book The Zahir , etc.
  3. Comparing keys sending most visits v/s keys that lead to longer time on site, it is revealing that top quality key-terms have been one-off. Rather than the blog appearing consistently for a particular key, people are discovering the blog by chance. This is a good way to identify keys for Search Engine Optimization.

Some of the improvements that I could think of were these:

  1. People search 'alchemist' were disappointed. They did see the quotes from the book - but it is not organized. Hence, they left the blog.
  2. Top quality results have been for those blogs where the title and the URL are meaningful. For example, the book review blog for Social Intelligence has the title Book Review: Social Intelligence and the URL as http://rakshay.blogspot.com/2007/12/book-review-social-intelligence.html. By being consistent it has got a good quality audience

The above metrics then implied that my blog had started to get a better attention from Google - since searches were more often throwing up my blog. Yet, some of the keywords were not leafing audience who liked the site. This implied either that the content was confusing google or it was not optimized for the right kind of keywords. So, now, I have to read through the blogs, check the title and URLs and optimize the blog so that it starts to gain a better performance.

Conclusion

The pattern analysis for two time-period for just a personal blog threw up a huge list of questions and some startling results. Imagine how it could unlock the areas for improvement if this had been a commercial grade high-visibility web site!

Just before I sign off - I found that there were 9 visits from a same person in Tech Mahindra reading all the blogs regarding matrimonial.. hmm. i wonder who this was? And - one person from Millington, NJ searching for my name on google had landed and remained for almost 10 seconds on my blog! Any clues who these people were?? :-)

Glossary:

Referring Sites: The web site where a link to the current blog was placed. Users clicked on this hyper link and arrived at my blog

Organic Search: Process of entering a search keyword and arriving at a site by clicking on one of the search results. This is used to compare against the paid search.

Reference site for analytics

Life Unraveled - http://rakshay.blogspot.com

Web Analytics results(excel sheet) - http://rakshay.googlepages.com/webanalytics.xls

Article: Lessons on corporate blogging

-Akshay Ranganath

In his article, Lessons from Corporate Blogging , Nicholas Carr, the author of the controversial paper, 'Does IT Matter ' puts in his thoughts over corporate blogging. Here's what he has to says:

  1. Set some sensible and hygienic ground rules - don't be too stifling but, at the same time, don't allow anything and everything on the corporate blog
  2. Remember that a blogger can create a huge audience - and the blog is associated with the blogger. Not the corporate. So when the blogger moves, the blog moves. The corporate could have some potential repercussions.
  3. Align the blog with the business strategy. Don't blog for blogging sake. Dell being a 'direct-to-customer' leverages listenting to its customer through the blogs.
  4. When getting started, try with a narrowly focused blog - and then open the flood-gates. Apple successful did this.

Check out the complete article at BusinessWeek.

Reference:
Lessons in Corporate Blogging . What your company can learn about keeping an online journal from the likes of Dell, Microsoft, and Apple

What my guts tells me about future of media?

-Akshay Ranganath

There was a wonderful article on how where the Media as a space is headed. This is an article by a VC about how he sees his kids use the new media and what he can derive from their usage.

In short, he feels:

  1. DVDs will be big mainly for the TV shows - it delivers more bang for bucks than movies.
  2. Movies are cool in theatres - still
  3. Books are still read - the old way. Reading online is still not a comfortable idea. Surprisingly magazines are a huge hit and they should be growing more..
  4. Radio and Newspapers are not so important anymore. Sure people use them a bit - but, they are not the primary source for entertainment and news respectively.
  5. Social Networking is big, games on all kinds of devices are big and teens have no problems to seamless switch between devices.

The tile of my post is a paraphrase of the article.From the Indian experience, here's what I feel will be happening in India:

  1. DVDs are now getting big - mainly due to Laptops, DVD players and most importantly, Moser Baer's chape DVDs. Considering tickets in Bangalore now cost anywhere from Rs 200 to Rs 500, its a big burden on the family. Maybe it's affoardable for a Double Income No Kids (DINK) family, but, for others its a no-brainer decision to go for DVD. Considering that we are now getting DVDs of latest movies (I got "Jab We Met" DVD just 2-3 months after the release) it is going to be a big business for movie makers.
  2. Books are still read - if in doubt just check the pirated books available on the foot paths!
  3. Games are in and so is FM radio. Yes - we are a bit late in catching up with the personal entertainment field - but, lately, Mobile devices have turned into personal music systems. Just see the people plugged into music in any of the BMTC buses.. iPod though is still rare. Unless you are a Cognizant employee, you'll probably not be sporting one :-)
  4. Adoption of email and internet is growing - but driven by different factors as compared to West. From the way my parents use internet for surfing matrimony sites and contacting prospects, I beilive the older generation is adopting technology more for fulfilling the traditional filial responsibilities rather than for simply staying in touch. Similarly, they are still not too comfortable with FM radio - preferring the "K" soaps on Telly, Breaking News on NDTV, et. al and newspaper for all opinions. Considering the profliage of news channels and coupled with never ending analysis, they are simply digging their own grave by driving away people who want to get a general feel of happenings - fast.
  5. Orkut is the best Social Media site. Facebook is a fab. LinkedIn is gaining ground. Recruiters are now targeting prospects via Orkut. This could act as a big turn off if the trend continues.
  6. Mobile phone usage which was a big fab is now turning into more of a utility - and hence, the Average Revnue Per Subscriber (ARPU) could reduce. So, Mobile phone companies will push features like collaboration gaming, song downloads, etc more.

Well, these are my thoughts at the beginning of the new year. Do let me hear your thoughts.

Reference:
What My Kids Tell Me About The Future of Media - http://avc.blogs.com/a_vc/2008/01/what-my-kids-te.html