• Yehoshua Coren

    Nan –

    An enjoyable article — thanks!

    One area that I’d like to comment on is your description of “custom metrics.” You say that UA requires data to be sent with existing data types (this is true) and as a result the number of phone orders would be shown as “visits” (not really true). Something which you should be aware of is that UA is offering something known as “custom dimensions” which can be paired with custom metrics to represent almost any sort of data. Custom dimensions describe the data type sent to UA. So while it is true that the data itself needs to be pair with an existing data type (pageview, event, social, ecommerce etc), the actual reporting will very much be in real, easy to understand “human” language.

    For example, in this image (http://analytics-ninja.com/universal_phone_calls.png) we see the number of times users have called, how many total phone calls were made, and the number of times those users have visited the site. Much in the same way, “phone orders” can be set as a custom dimension that would be applied to either to an ecommerce transaction (better) or to an event (satisfactory).

    Bottom line, a custom report with Phone Orders (using order # as dimension) with Number of Phone Orders as metric is certainly achievable with UA. Most exciting, imho, is the ability to see ALL commerce activity, and then to segment the sales by “online” or “offline” revenue.


    p.s. The ability to do this with Google Analytics is not new, however, with the current GA it is quite difficult to implement. With Measurement Protocol, this type of data capture is native and much more accessible to GA users (though you correctly point out that professional assistance will still likely be required for most businesses).

  • ndawkins

    Thanks Yehoshua (especially for the image). I have seen examples of pulling in data sources without using custom dimensions and phone orders were shown as visits. Until your image, I had not seen any examples of using custom dimensions for those external data sources. I wonder, how many custom metrics are you pulling in? Is it more than phone orders? I’m curious as to how usable (or alternatively, how clunky) it gets in cases where there are many different data sources and metrics that are being pulled in. Dying to see some more examples….

  • Yehoshua Coren

    The key (pardon the pun) with Universal Analytics is that there must be a way to connect the dots between all of the disparate data sets.

    The in store sale example you have above is a good one. In order for a retailer to connect the dots between a visitor to their website and an in store purchase, they could provide the user with a unique coupon code on their website. That coupon code, when entered into the cash register, needs to be linked in a backend system to the ClientID that the user had when they were browsing the site. Then the cash register software sends information about the transaction to UA using the measurement protocol, and the transaction is applied to the visitor based upon their anonymous ID.

    In GA, one could then run an ecommerce report and segment by online vs. offline sales.

    The above example isn’t a simple to execute by any means, but it isn’t rocket science. So, when you say “clunky”, I’m not sure if you’re referring to how difficult solutions like this are to implement or if you mean how complicated one’s data set in GA becomes.

    In terms of implementation work, companies definitely will need resources for developers and analysts who know what they’re doing.

    In terms of data usability, as long as the data collection needs have been aligned in advance with a company’s business needs and objectives, my thoughts are that the data will be very useful. Mapping out a digital measurement strategy also requires qualified individuals, but for most businesses the value of data that answers businesses questions and drives decisions far outweighs the costs.

  • PaulusDeB

    I think with the whole NSA thing a fourth ‘Barrier To Adoption’ is in place: do companies really want to be associated with a company that hands over their data to the NSA?

  • http://www.TroyAClark.com/ Troy Clark

    Is this “Sick” or “666”?