We’ve talked a lot recently about the power of decentralized commerce and use of widgets to enable this new paradigm. Of paramount importance to the success of decentralized commerce, however, are the analytics we’re able to track. Today, a look at the general categories of analytics we will need to capture when using widgets.
I’ll use the term “external metrics” to mean analytics about a widget’s surrounding area. “Internal metrics” are gathered within the widget itself. “Multichannel metrics” refers to the various flavors of a specific widget, and on what platform it is operating. “Viral metrics” refers to metrics we gather when the widget (or content within the widget) is sent to someone else.
Internal metrics borrow their elements from both the e-commerce world and the content delivery world. On the e-commerce end, we track typical e-commerce metrics, including:
- Product views
- Category views
- Cart abandonment
- Checkout and sales
- New customer registrations
- Existing customer sign-ins
Taking a page from content delivery folks, we’ll track events such as movie clip views (and movement within a clip), and what other buttons were pushed, forms filled out, etc. The specific metrics here will depend on the key performance indicators (KPIs) the company uses to gauge the success of these types of marketing campaigns.
External metrics for widgets are very similar to ad campaign metrics. They include the placement of the widget on a site (such as run-of-site, homepage, MySpace profile, etc.), and the placement of the widget with a specific page (above the fold, below the fold, top left, etc.). They also include metrics traditionally used in affiliate marketing. This includes the type of site hosting the widget, and some unique identifier for that site.
These metrics enable us to put internal metrics into context. Internal metrics must be sliceable by external metrics. In other words, I may want to understand a widget’s effectiveness on people’s pages in social networks, as opposed to their effectiveness on personal Web pages or custom storefronts.
With the explosion of platforms on which widgets can run comes a more complicated view of widget metrics. There are at least three different types of widgets in use today:
1. Widgets in some encapsulated form (such as a flash module) that can be embedded on any Web site.
2. Widgets built on an extensible platform, and integrate deeply with their host environment (such as Facebook applications)
3. Widgets built specifically to run within a non-PC environment (such as installed programs on a PDA or phone).
It can become a complicated task to create widgets that work over multiple platforms and deeply integrate with each hosting environment. Luckily, various companies are rising to the occasion and filling that need. Clearspring Technologies, for instance, has a “widget platform” on which you can build extensible widgets. Because they abstract host-specific information (much like a class library in C++ for those technically minded), one can create a widget once, then migrate it to other platforms (such as Facebook) or other operating environments (such as mobile phones).
This type of platform makes multi-channel analytics a lot easier because each variation of the widget is built upon the same reporting platform. If you create each version of your widgets yourself, this can become a more complicated (though not insurmountable) issue.
Multichannel metrics we need include:
- What variation of the widget is used (stand-alone, platform integrated)?
- What device is being used (PC, phone, kiosk, etc.)?
- Any other environmental information we can gather about the host platform
We’ll need to slice the internal and external metrics by the multichannel metrics. We must understand the value of our “in phone” widgets vs. “on PC” widgets. We should understand the relative value of widgets deeply integrated into their environment (such as on Facebook) versus completely encapsulated widgets that are unaware of their hosts.
Finally, we must look at viral metrics for these widgets. Not all widgets will have a viral component, but many do. Taking a page from viral marketing, we will need to consider:
- How many times the application as a whole is recommended to someone else
- Adoption rate of the new application to the recipients
- How many times content within the widgets (such as product information, store locations, videos) are sent to other people
- Content use, and future use of the application by recipient
I’m much more interested in widgets’ e-commerce aspects. Years ago, I wrote about the metrics we created when designing the wishlist for BN.com. I was excited to see Blue Nile has recently created a widget that allows customers to place their wishlist in their Facebook profiles. Some metrics I hope they’re tracking include:
- How many customer syndicate their wishlist
- What sales result from the wishlist syndicated on Facebook (versus the wishlist on Bluenile.com)
- How many new customers join Bluenile to use this functionality (i.e.: how effective is the widget as an acquisition tool?)
Lots of Metrics — What Do They All Mean?
It seems widgets borrow metrics from almost every type of online marketing. Companies like Coremetrics and Omniture are actively working to create environments in which all these metrics will make sense and are actionable. Already, with intermediaries like Clearspring Technologies, we’re seeing widget-specific metrics that can be fed into these larger metric engines. Eventually, we’ll be able to get a very clear picture as to how each widget variant is doing, and what environments provide the best host.
Comments, thoughts, questions? Let me know!
Until next time…
Nominate your choice of technologies, companies, and campaigns that made a positive difference in the online marketing industry in the last decade. Nominations end August 3 at 5:00 pm (EDT).