Widget Analytics – Measuring the widgets in the wild

Helping web analysts navigate the measurement and tracking of widgets.

Archive for October, 2007

Measuring interactivity within a widget

Posted by widgetgirl on October 29, 2007

Today marks an exciting product release from Clearspring! The team’s been working the past few months to roll out interaction analytics – specifically, the ability to measure how visitors are interacting with a widget. The metrics that we have put together break down as follows:

  • Clicks – The number of times a widget was clicked.
  • Clicked Views – The number of views where a widget was clicked.
  • Click Rate – Total clicks divided by total views.
  • CPCV (Clicks per clicked view) – Clicks divided by clicked views.
  • Time Spent – The average time spent per widget view as defined by the time that the page loads till the visitor navigates away from the page that the widget is on or closes their browser.
  • Interaction Time – The average time spent per widget view as defined by the aggregate amount of time that a user interacted with a widget.

Clicks – this metric is pretty straight forward, the number of times that a widget was clicked. The clicks that we are including are user initiated clicks on the widget. What we are not including are the clicks to share the widget. A visitor can click on the share services menu and expend a few clicks to share their widget into any one of the social networks (or even to just get the embed code). The clicks that we are including are those clicks where the visitor is interacting with the widget. A great example of this is the new Bee Movie (registration required for this widget). The widget allows you to “Bathe”, “Feed”, “Sleep” your bee. Each time you click on these buttons and interact with the widget, the visitor is generating a click.

Boo on WordPress for not allowing the insertion of widgets! So here is a screen shot of the Bee Movie widget – but you can click on the link above and grab the widget directly from Paramount’s site.

Bee Movie

Clicked Views – this metric is a derivative of clicks and views, the number of views where a widget was clicked. I think this is a great metric to measure if visitors are even drawn to your widget. It assumes that you want a visitor to click on your widget and/or that someone has tried to share your widget.

Click Rate – similar to clicked views, this metric is displayed as a percentage rate and gives the analyst a benchmark of the percentage of views where someone clicks on the widget.

CPCV (clicks per clicked view) – I love this metric. If you have a highly interactive widget where visitors can either play a game or interact heavily with your widget (like the Bee Movie for example), this metric will help you measure the engagement level of your widget. If your widget is merely a video or a vehicle for getting visitors back to your website, then clicked views or click rate will probably suffice.

Time Spent – measuring how long someone spends viewing your widget is not the easiest metric to tackle. If you think about it, page view duration in standard web analytics is really derived by the difference between the time stamp of two pages that have been viewed. In the case of a widget view, a visitor is not required to refresh the page….hence how does one calculate this. The team here had to weigh how “chatty” to make your widget in alerting that it is still alive on the page with two goals:

  1. Maintain user experience by not reaching out to the server every time a visitors touches the widget.
  2. Provide the most precise measurement of time spent available amongst the widget provider-sphere.

The end result, a default setting of 30 second “pings” back to our server that can decay incrementally over time if a visitor is not interacting. However, in the event that a visitor interacts with the widget, all clicks, events and an update on the time spent metric is sent within 5 seconds. We provide the Widget Creator the capability to turn this setting up or down depending on the type of widget that they have. There is more documentation on the Clearspring site as to how this works, but the point is that measuring the difference between two time stamps like traditional web analytics is not applicable here. One could easily overwhelm the amount of requests back to the server for analytics and degrade user performance.

Interaction Time – thank you web analytics village for helping me get my hands around this metric. The interaction metric measures the precise amount of time that a visitor interacted with the widget. So if they moused over the widget for 10 seconds at the beginning of their widget view and then moused over it again for 25 seconds five minutes later, the total interaction time would be 35 seconds.

Other cool stuff to check out this week:

Fox.com has launched a very cool 24 widget! I am a huge fan of the show was and was so psyched to see this widget on the Clearspring platform. What is interesting about their deployment is that Fox is using Clearspring’s new Launchpad platform. The new product we’ve deployed allows you to seed the widget sharing menu adjacent to your widget or just add a button to your page that invokes an expandable or pop-up menu to share your widget. Very cool stuff!

24 Trailer Widget

Advertisements

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

What is a placement?

Posted by widgetgirl on October 25, 2007

In my last post I discussed the concept of a Placement. A placement is the unique instance of a widget that has been placed on a web page. When a placement is created, a placement ID is assigned to identify that unique instance of a placement. When reporting on placements there are multiple ways to analyze them.

There is one metric I want to cover in this post – but three different dimensions to analyze that one metric over. This is a starting point and could easily spawn new metrics or additional interpretations of existing metrics.

Before we dive into definitions, let’s revisit how widgets can spread and what the nodes look like for spreading.

  1. A widget is seeded on mywebsite.com where visitors can come and grab the widget.
  2. When the widget is grabbed, mywebsite.com is given credit for sharing the widget.
  3. The visitor who grabbed the widget from mywebsite.com and puts it on theirwebsite.com has now created a new placement.

Each one of these scenarios presents three different ways to look at the website mywebsite.com.

  • Scenario 1:
    • Placement definition – The instance of a unique placement ID. A placement ID is assigned when a visitor invokes the sharing services on a widget and successfully shares the widget to a social network or grabs the embed code.
    • Placements metric definition – The count of unique Placement IDs.
    • In scenario number 1, mywebsite.com is a domain with a metric count of one placement.
  • Scenario 2:
    • Viral Hub (a categorization of domain) – A domain that has generated one or more placements.
    • Viral Hubs – The count of the placement IDs by domain that have generated at least one new placement.
    • In scenario number 2, mywebsite.com is a Viral Hub with a metric count of one placement.
  • Scenario 3:
    • Heritage (a categorization of domain) – The domain(s) from where a new placement originated.
    • In scenario number 3, theirwebsite.com has a metric count of one placement. And when we ask the question of “where did the placement on theirwebsite.com originate from”, we know that the heritage domain is mywebsite.com.

Confusing – a little bit. Where does this leave us?

mywebsite.com –> theirwebsite.com

theirwebsite.com <– mywebsite.com

Mywebsite.com generates a new widget placement to theirwebsite.com. Theirwebiste.com is now a new placement of a widget. Mywebsite.com is the domain where the placement on theirwebsite.com originated. Yes, it is one big vicious circle, but each scenario lends its own analytical value when analyzing how the widget is spreading. The next step is applying it to your own business.

Widget Optimization

Strategic Questions:

  • Where should I be seeding my widget so that it can become a viral hub and generate new placements?
  • Where has my widget been placed?
  • Which viral hubs are generating placements into specific sites (heritage)?
  • For non-seed placements , what new viral hubs are emerging?
  • How do I get future widgets out to non-seed placements that are strong viral hubs?
  • Should I build relationships or partnerships with non-seed placement viral hubs? What if they are individuals?

There are a lot of business questions to be asking when analyzing your widget strategy. There are three takeaways:

  1. Build your widget to be engaging and relevant to your audience.
  2. Seed your widget where it has the highest probability of being grabbed.
  3. Understand how your widget is spreading and focus on optimizing those distribution points.

Here is a link to my favorite widget of the week!

If you go to the bottom right side of the ivillage.com web site, you’ll see the “Daily Blabber” widget.  Click “grab it” and you can add it to your own site or profile page.

Catch you next week…..Jodi

Posted in widget analytics | Tagged: , , , | 4 Comments »

E-Metrics Summit – Marketing Optimization Notes

Posted by widgetgirl on October 16, 2007

It is day two at the E-Metrics Summit in Washington, DC. During my presentation yesterday there were a lot of great questions around how to wrap your widget and get it launched into the wild. Thank you everyone who came to listen. This is such a new area and the thought of releasing your content outside of your domain (and then trying to measure it) can be a little mind boggling.

Wrapping your widget and the foundation for how to widgetize your content was a harder concept for folks to grasp than how to measure your widget. Understanding how the technology works is a pre-requisite for learning how to analyze it. So how does this “wrapping your widget” thing work? Disclaimer – this is the laywomen’s definition, not a technical one 😉

A widget creator can take any piece of content and drop it into the Clearspring flash container through a multi-step process in our UI. The output of “wrapping your widget” is some embed code that you can place on your seed placement (see definitions below). The container allows for tracking of the widget (widget analytics) and the ability for visitors who view the widget to share or grab it for their own site. When a visitor clicks on “Grab It” or “Share It”, or whatever link or name you want to use as as widget creator, a service menu is invoked. The menu provides the capability to push or grab the content in the following ways:

  1. Push the widget to your favorite social network such as MySpace, Facebook and others
  2. Grab the embed code.
  3. Email a link to the widget
  4. Send a text message with a link to the a WAP hosted version of the widget

Warning – new definitions needed here for you web analysts.

Placement: A unique ID assigned to an instance of a widget on a particular page.

Seed Placement: A placement where a widget creator places their widget for others to view and grab.

Non-Seed Placement: A placement that has been generated from someone “grabbing” the widget.

More on Placements in my next post. Off to lunch here at E-Metrics!

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