Web Analytics India

- Contributed by Deep Moni Hazarika

It’s so much a ritual now to find out the depth of a page or website. How different is it to say rhyming with “death of a page or site”! Fear not. Depth or alternately called Page Views per Visit is says how deep inside your site did a visitor go in a regular visit. However do note that its Average Page Views per Visit that we are more likely to look at. This is a figure we more often than not don’t avoid calculating, to add that “extra piece of valuable information”. A greater depth of visit for a content generation site would mean a “Wow!” but the same for a support related site would mean “Oh no! I gotta bring it down”. So, reactions differ depending on the nature of the site.

But wasn’t I going to speak about some ‘misleading’? Well on a quick note, let me ask this – what does an Average Page Views per Visit of 0.87 mean? i.e. if there were say, 100 Visits, then there were 87 pages viewed. How do you explain that? Under ideal conditions, every visit has to have at least one page viewed. But the scenario looks confusing here. There are extra 13 visits and if every visit has to have at least one page view, then we are either missing some data or our calculation is wrong somewhere! Funny ehh! Technically, would it also mean that for those visits, a page did not get loaded in the browser completely or returned some error? Nope! Because, an incomplete page means a client to server status code was passed which was not 200 (the success status code). And as it goes, all web analytics tools reports only successful page requests when showing up those numbers. Then what did those extra visits actually do?

Looking at it another way, are we looking at the complete picture? Is anything wrong with that formula? How can that be? World over, the formula has been laid out plain and simple, again and again. Now let’s look at the same thing with an additional bit of information. There were 100 visits. There were 87 pages viewed. In addition to that, in those 100 visits, let’s say there were 13 pdf files downloaded. Going by the standard definition, the average page views per visit is still 87/100 = 0.87, but with a slight modification in the formula to include the pdf files downloaded, we get the same browsing behavior as 1 now, i.e. (pages viewed + downloads)/visit. Seems to make sense!

On dissecting this scenario, we can see that when we say pageviews per visit as the depth, we have been considering only the ‘pages’ viewed and not the downloadable contents. Considering the scenario where a visitor is not aware of your website but frantically looks around for some information through search engines and one of your websites pdf file gets listed amongst the top results. The visitor now willfully clicks on that pdf link and in effect has caused one visit and one download but no page view. This situation cites that accessing a single pdf also is an engagement with a depth of 1 for that visitor, considering the visitor didn’t access any other content in your website. Visitors of such kind coming from a search engine are one of the many possibilities where content is directly accessed without necessarily causing any page view. In effect to that, the depth of the website can be a number less than 1 if considered only for ‘pages’ and higher than 1 when considered for both ‘pages’ and ‘downloadable’ content.

This also shows how the simple formula can lead to confusing interpretations. So what is your take on the depth of visits to your site? Are you sure you are getting it correct?

My next post will elucidate on how the simple formula of page views per visit can itself be an all in itself when looking into some of the available analytics tools.

Reference links:

1. http://blog.webanalyticsdemystified.com/weblog/2005/07/average-page-views-per-visit.html
2. http://elementsinc.net/blog/2008/08/04/a-key-indicator-for-your-website-is-the-average-number-of-pages-viewed/

- Contributed by Mukesh Sharma

Numbers numbers numbers… where are the metrics??? Where are the reports??? Can i have the reports by noon? …demanded one of the managers.

It was a tense morning, not the first time though! It got even worse when we wanted to “export reports” from our trusted WebTrends. Now, so what if the report extractor is not working and throws up an error which analysts like us don’t understand?

Here’s a snap shot of the error.

Looking at the error file one can only say what the heck! How do I go ahead with this now?

Looks pretty messed up!! Isn’t it? Maybe there is a quick solution, but my focus was more on how to get the reports requested for and how to get back to the manager that his report might possibly get delayed. I terribly needed an alternative. Was there any? Well there’s always an alternative! And here the alternative was using the “Scheduled Jobs” feature in WebTrends. I had never used the scheduled job for extracting the reports before, but this turned out to be a perfect opportunity to learn and implement the feature. This feature does pretty much the same thing as the report extractor does, but the only difference being, one needs to schedule a report just like scheduling a profile. Now if you are thinking about the time, believe me it’s much better to use this feature instead of collecting the numbers manually.

Coming back to the reports which were to be delivered, the manager was pretty pleased to see the reports by noon!

Hey wait a minute, what about the “report extractor “? Well, that problem went on till a week or two, making calls to the WebTrends vendor, uninstalling and reinstalling the java plug-in several times. Add to this, a funny part in this tense situation – the vendor goes missing all of a sudden!

Finally the vendor returns. Ahh! Thank heavens. A call is set up with the WebTrends support team along with one of the much used remote support tool. But this wasn’t going to be a smooth ride. The next few hours saw me going through three different support guys with them again asking me to uninstall and reinstall the java plug-in, that in spite of me telling them that I already tried installing java plug-in afresh.

The problem wasn’t getting anywhere to a closure until I was put across the third guy. This guy too asked me to uninstall and reinstall java. Oh damn! This is too frustrating, I thought to myself, but I couldn’t really help. The call and the remote session went on for about an hour when suddenly the guy, asked me to open up IIS on the server where WebTrends was hosted and check the path where the exporter was installed.

Bingo! There was the problem. The home directory setting of the IIS setting had got changed – the setting for “The content for this resource should come from” had somehow changed to “A share located on another computer” instead of “A directory located on this computer”. Some magic!

Finally I checked if the exporter worked now. As a matter of fact it did and I thanked the guy for figuring out the problem. Was that the real problem? Should it have actually taken so many days to solve this problem? I thought, after the call ended that carried on for an hour and half on the final day, putting an end to over a week long problem.

Well I at least now know where to look for the problem if the exporter is throwing errors.

- Contributed by Deep Moni Hazarika

The presence of the internet as we see it now is more than a decade, though it took time to become a dependency in many matters – be it corporate presence, buying and selling online, news matters, hobby and what not. In this short span of time however, what happens to websites which are relatively grand dads or an elderly to any website which went live in the very recent present? First thing that came in my mind about them was whether they are able to generate the same level of interest to their visitors as before or not. In all probability, if they survived for long, then they sure have got something. Next thing then hits me is how much of visitor behavior data they will be having now. Think about sites of NASA, FBI, etc. which dates back as much as to 1995-’96.

Presence of enormous amount of data all around is so profound that at one point or another, we happen to hit at a time when we seem to be running out of space. Website data is not excluded. If you are living with web log files and your website gets even about 10,000 daily visitors, and also considering that the last time you did a live pages count in your site was somewhere even around 5,000 to 10,000 pages, you must be struggling at times to cope with the growing amount of data. Log files are growing; your visitors are creeping up. There is a goldmine of information with you but how do you extract out that gold.

As most would agree, more the data we have on a matter, more predictable is its future. In that view, considering those grand daddy websites had implemented some form of web analytics solution, are they able to decipher the code of predicting the future of their website data? I doubt! It’s just a strong feeling though and just a few reasons on why I think so. To support that, look at the long list of all kinds of web analytics products that had been around for a long time now – enterprise wide highly charged ones, some super jazzy ones, some with extra level of complicacies, and even to those which are smartly free to use. In my opinion, if we roughly speak about a plain vanilla web analytics tool (most of the free ones and some of those super jazzy and super cool ones fall in this category), none meets to the expectation of straightforwardly showing possibilities or outcomes based on your heap of past analyzed data. They just generate numbers and reports. Well that’s good but satiate a couple of your managers. But there can be more to the story. You are lost when it comes to the situation of “What – if’s”. There is a level of intelligence incorporated into some of the web analytics tools that takes into consideration non web data and also has the capability to churn out some of the “What – if’s” and there is more to be seen.

Companies now go for a mix n match of two or more web analytics tools, each of which would answer to certain specific requirements which are prevalent at that time. And there are people talking about actionable insights. Well, out of a couple of simple reports from your much used free web analytics tool, you can make out that a page doesn’t perform well, a landing page needs to be optimized better, set of keywords can be better targeted, fall-out of visitors from your best pages can be minimized and blah blah. Congrats that you are able to add insights. But more often than not, you forgot about the past data and what value that would have added. But alas, apart from using that dull calendar to calendar compare, there is not much to look into. The question here however would be – how do I…? what do I…? The sole matter is that there is very little guidance on what are the possibilities out of it. Next time when working on those commonly looking reports and numbers, think out that there could be more to it.