Are you New Here ? You might want to:

                    1. Take a look at some of the most popular posts 
                2. Subscribe to the RSS Feed
                3. about StartupDunia
                4. Leave me a tip by email

The voodoo of web metrics

December 19, 2007 · Posted in Tech tidbit 

The emergence of Web 2.0 and Internet startups has given rise to a whole new industry in itself - web metrics. If you follow the Web 2.0 world, it is unlikely that you havent heard of at least one of the following companies - Comscore, Hitwise, Compete, Alexa, Quantcast. All these companies provide various web metrics such as number of unique visitors, pageviews, hits etc. for websites.

The metrics released by these companies often become the subject of interesting conversations within the tech blogosphere. While a lot of the top tech bloggers frequently refer to Comscore and other metrics, very little is discussed about the process and methodology used by these companies to compile this data.

Ever wonder how exactly these companies compute their metrics ? What exactly is the voodoo behind these numbers ?

In today’s date, when web metrics play such a huge role in website valuations, I thought it would be a good idea to get an overview of the various companies in this segment and the methods they use for data collection.

In the first installment of this multi-part post, I’ll overview some of the data collection methods used within the industry. In a follow up post, I’ll review the companies and what specific methods they employ for data collection and metrics calculation.

There are three primary methods of data collection while computing web metrics. Each method relies on a different physical collection point where the data is tapped.

1) End User based data collection:

In this method of data collection, end users are generally required to install some piece of software on their computers, which will track their Internet usage. The software will in turn report the results to the metrics company, which can then compile the results. Two variations of the user end software are available:

(a) browser based toolbar - Users install a toolbar in their browser, which will collect the usage data. Alexa employs this method of data collection. The drawback of this approach is that non-browser related usage will not get reported. Moreover, if the end user is using a different browser, the Internet usage will not be effectively reported.

(b) Panel based approach - Users install an application (non-browser-based), which tracks both browser and non browser based Internet usage. The users agreeing to this method of data collection form a sample ‘panel’ and so this method is called panel based data collection. Often times, end users are provided with fringe benefits for participating in the “user panel”. Once the results are obtained from the user panel, the data is sampled and normalized to compute the final metrics. Comscore employs this method for data collection.

2) ISP based data collection:

The idea is to intercept and track the data at the network level, which would obviously be the ISP level. Data is gathered from the ISP and then sampled and analyzed for computing the final results. Note that in this particular case, a lot depends on the ISP chosen to track the data. And based on the subset of ISPs being used for data collection, the data can get skewed.

3) Website based data collection:

This involves tracking the metrics at the website itself. The website generally adds a tracking code into their webpage and the metrics get collected and reported to the company. A limitation of this approach is that the website owners have to agree to using the tracking code in their website. And a lot of website owners are wary of getting 3rd party metrics code into their website.

As we’ll see in a follow up post, the web metrics companies employ one or a combination of the above mentioned methods for their data collection.

Internet is a medium that prides itself on being measurable, quantifiable and chides TV and radio for the very same reasons. Quite ironical then — isnt it ? Web metrics and stats related to web page views and unique visitors etc. is still at best a ‘guesstimate’.


1 Star2 Stars3 Stars4 Stars5 Stars (No Ratings Yet)
Loading ... Loading ...

Similar Posts

Comments

5 Responses to “The voodoo of web metrics”

  1. Avinash on December 20th, 2007 12:00 pm

    Hey Pranav,

    Welcome back…nice post…could u also give examples for point 2 & 3. Any idea how Compete reports its figures?

    Rgds
    Avinash

  2. pranav on December 20th, 2007 12:33 pm

    avinash,

    HitWise uses (2) and the best example I can quote for (3) is Google Analytics…

    I’ll review each company (including compete) in a follow up post soon.

  3. Swaroop on December 20th, 2007 12:59 pm

    Quite right. Who is the best of the lot right now?

  4. pranav on December 21st, 2007 1:36 am

    Swaroop,

    I am not sure who can be called ‘the best’ (and as part of the follow up post, I’ll mention why)…but majority of the top bloggers do lean towards comscore and compete.

    -p

  5. Srini on December 27th, 2007 11:16 am

    How do we check comscore data for a given website. On alexa, and compete there are search boxes. How to go about comscore..??