Widget Analytics – Measuring the widgets in the wild

Helping web analysts navigate the measurement and tracking of widgets.

  • Share this blog

    Bookmark and Share
  • January 2008
    M T W T F S S
    « Dec   Feb »
  • Meta

  • WAA

    WAA Member

Archive for January, 2008

Widget Analytics – Social Network Stats Versus My Widget Platform

Posted by widgetgirl on January 29, 2008

When it comes to widget travels, not all destinations are alike. There are two camps if you will – those that allow widgets and create new placements in a clean manner and those that iframe in your desired widget from another web site and disrupt the data collection process. Did I lose you yet? One of the most difficult issues that I have seen so far in widget analytics is that the environment can really wreak havoc on your ability to analyze your widget (I like to call these “environmental threats”).

As a web analyst you have control over your site to some extent. If there is an issue with the way your data is being collected, you call a meeting and reign in the appropriate IT and Marketing folks to help you tag, filter or troubleshoot the issue. However, when your content is traveling into environments where you don’t have governance over how a particular widget metric is being generated, what is an analyst to do?

Social Network

The biggest challenge I have seen thus far is dealing with how widget placements are created from canvas pages and galleries within Facebook, Netvibes, iGoogle and Friendster. Each of these sites allow you to customize your start page or social networking profile page with widgets. The way that they host the widget within a gallery or canvas page is where the environmental hazards kick in. There are two ways that this can happen – let’s take Facebook as the first example.

There are two ways to add a widget to your profile page on Facebook:

  1. The “Canvas” page: One of the compelling Marketing tools on Facebook is to create a “Canvas” page to display your widget (or Facebook application). In FB, the Canvas page contains an image of the widget, not the actual widget itself. When a widget is “registered as a FB application, a page is created that takes a screen shot of the widget with a FB button entitled “Add Application”. When a widget is added via FB using the “Add Application” button, the widget is not being “grabbed” from a physical widget. Instead, FB reaches all the way back to the widget serving platform to “grab” a new placement of the widget. The end result – the new placement of the widget appears to be coming from the widget serving platform (Clearspring in this case) rather than Facebook. In the Clearspring analytics platform, Clearspring.com would get credit as the viral hub for the new placement instead of Facebook.com. If this doesn’t make sense, keep reading and you’ll see the difference.
  2. The second way that one can add a widget to their profile page is to “grab” it from another widget in a friend’s profile or as we say here internally (grab it from the container – the service menu inside the widget container that appears when a visitors clicks “Grab This” or “Get and Share”). When a widget is “grabbed” from another profile, the visitor has to interact with an actual widget and invoke the sharing menu within the widget. When this happens, the widget platform that is serving the widget has full control to assign a new placement ID to the new widget instance and connect it with it’s parent placement – the widget that it was grabbed from. When a widget is used natively to create a new placement within Facebook (or most sites), the Viral Hubs report in Clearspring analytics will credit the source domain for creating the new placement (Facebook.com).

In the case of bullet number one in the above set of examples, the widget serving platform has to know that FB canvas pages work this way. Once that has been identified, an override can be put in place by the analytics system of the widget serving platform to attribute the new placement source domain as Facebook.com and not Clearspring.com (I call this “analytics remediation”).

The second example that I am going to walk through is the iGoogle Gadget gallery. Widgets may be published to galleries from either the widget serving platform(in Clearspring’s case we have a UI for you to publish to many galleries simultaneously) or natively from going to the gallery’s web site and registering your widget manually. Regardless of how your widget gets into a gallery on Google, the way that it is shared from the gallery is where the environmental hazards start. Here is how Google works:

  1. The Widget Creator registers their widget as a Google gadget.
  2. The widget appears in the Google gadget gallery for a visitor to grab.
  3. Visitor selects “Add it now” for the desired widget to add it to their iGoogle start page.
  4. Google adds the widget to the visitor’s start page by iframing in a page with the widget on it.

There are two environmental hazards here. First, the iframing of the widget means that the widget flash container is not the first thing to load on the page. If you are a traditional web analyst, you know that you need that referrer field in the request to determine where a widget or object is being viewed from. When iframing, the URL of the page that is being iframed in (Clearspring.com in the case of Google) becomes the referrer (not Google.com). The remedy here is for the widget serving platform to recognize that Google is iframing and correct the referrer field so that all views of the widget and the registration of its placements are associated with Google.com.

The second environmental hazard is that because Google iframes, they do not always “force a new placement of the widget”. What this means is that Google may reuse the page that they are iframing and display the exact same widget/placement combination to multiple users. The result is that the placements metric that calculates how many times a widget was shared does not increment on the widget serving platform’s analytics system. The discrepancy here is that Google’s inline analytics on the gadget pages show (“this gadget has been shared x number of times) a number that is much higher than what your widget analytics platform will show. Override for this…yes, but it happens behind the scenes 🙂

The bottom line is that environmental hazards pose a constant risk to your widget analytics. As a platform, we have to constantly be scanning the environment for these hazards and entering into remediation whenever and wherever we can. Is it annoying, yes! But this is the world of Web 2.0 where anything can happen and you are not always in control of how others use your content or share your widget. The distributed web introduces a lot of new challenges for web analysts (and widget analysts). Understanding the landscape how to deal with the hazards poses the ultimate analytics challenge.

The Serenity Prayer applies to the Web 2.0 web analyst, “God, grant me the serenity to accept the things that I cannot change, the courage to change the things I can, and the wisdom to know the difference.”

Check out one of the newest widgets on the Clearspring platform – National Geographic’s U23D movie widget.  Super cool and definitely a topic for a future post.

U23D widget

Posted in widget analytics | Tagged: , , | Leave a Comment »

Introducing the Dilbert widget…on the Clearspring platform!

Posted by widgetgirl on January 24, 2008

Hello Dilbert fans! Clearspring had a monumental moment yesterday when we discovered that the official Dilbert comic widget is now using the Clearspring platform for viral distribution. How cool is that! For anyone who knows a little bit about our company, we are the anti-Dilberts of the world. The Clearspring crew is nimble, creative and so not into the office politics that poor Dilbert has to endure.

So let’s dissect this widget. The widget creator (WC) did a great job of building this widget. From a design perspective:

  1. Ability to pull comic strips for the past two years.
  2. Ability to view weeks at a glance within the widget to select the week you’d like to read (they show thumbnails within the widget that you can click on).
  3. Ability to rank the widget – and a display of how many times the widget has been ranked (bonus).
  4. Various sizes of the widget so that you the user can choose which size you would like to grab for yourself.

This is the type of widget that would be perfect to download to your desktop. And speaking of Desktop, we are on the cusp of releasing a new report in the Clearspring Analytics console that will help you to analyze your Desktop widgets. We currently support the ability for you to “grab” your widget and push it to Konfabulator, Vista Sidebar or Mac Leopard. A successful “grab” (new placement of the widget) to desktop is measured as an “Install”. More on this metric later once the report is deployed to production.

Desktop widgets have the exact same capabilities as web widgets, but introduce some interesting new data. For example, a web widget may have an average time spent metric that ranges in the minutes to up to an hour. For desktop widgets, most users power up their computer and then leave it on all day long. So an average time spent metric could range from 1 hour to 12 hours (depending on how much of a workaholic you are). For this reason, we are separating out the desktop analytics from web analytics so that the data points aren’t skewed.

Look for an upcoming post that defines all of the metrics available for your desktop widgets (some oldies and some new ones).

You can check out the widget at the official Dilbert Widget Site.


Posted in widget analytics | Tagged: , , , | 2 Comments »

Web Analytics Association Podcast – Widget Analytics

Posted by widgetgirl on January 18, 2008

The day before Thanksgiving (yes, Thanksgiving 2007), I was taped by the WAA Research Committee for an interview on widget analytics.  The various committees within the WAA are all volunteer and so after several weeks, the podcast has now been made into a transcript and may also be listened to on the WAA’s web site.  Thank you to Jennifer Day from Coremetrics for volunteering her time to tape the interview.

Check out the podcast (or read the transcript – my preference if you hate hearing your own voice 🙂 ).

TGIF – Jodi

Posted in widget analytics | Tagged: , , | Leave a Comment »