Posted in: analytics, Email, web20, xobni

Where Are My Email Analytics?

I use a lot of analytics products (Feedburner, Google Analytics, Google Reader Trends) and I get a lot of value out of them – it’s very helpful to be able to track and measure things. However, I’m been spending a lot of time thinking about the interface I use the most (email) and why there aren’t any good analytics products around my own email usage.

I’m not always the fastest on email (that title goes to my friends who work in venture capital), but I do try to be consistent in the time window for my responses. I tend to work under the theory that being consistent is the next best thing to being fast. My biggest concern is making sure that I do my best to get back to people in the time window that’s consistent with previous interactions.

Right now, the only two lenses through which I can interact with my mailbox are the traditional LIFV (last in first visible) approach or by the use of folders/labels. Neither one of those approaches gives me any sense as to the messages that really require my attention.

As email has become a more important communications tool, I’ve seen most of the big pain points get attacked. For me, spam is a non-issue; I still get spam but not so much that it can’t be managed. I also have tools that allow me to search through my email pretty effectively and retrieve messages where I’m looking for specific messages or keywords. The last big mountain to tackle is relevance. Relevance is hard to tackle because it means many different things to other people and unlike spam and search, the mark to hit is not a quantifiable goal (make all messages searchable, achieve a 99.99% spam capture rate, etc).

I’ve seen a lot of people talk about building more relevance into email. A lot of times it takes the flavor of “we’ll look at your inbox, your calendar, your task list, etc and magically divine relevance based on those factors.” That strikes me as a hard approach to pull of, partly because those are incomplete sources of data (for example there are people who I speak to on the phone a lot and when I get email from them it’s pretty important – you might not pick that up if you just look at my inbox and calendar) and for another reason I’ll get to later.

More than anything, I want an email analytics product that’s smart enough to do exception reporting and help me view my email through the lens of messages that are “out of bounds” in terms of the time it’s taken for me to reply.

To build a good analytics product, you need to have parameters that can be measured. For websites is page views, visitor activity via logs, etc. For feeds its distribution, readership, and clickbacks. You get the picture. Email has the same characteristics. You can see volume of communication, direction of communications, time between receipt and reply, time of day when I’m most active on email etc. Simply put, there are definitely enough parameters to build a statistical profile of my mail usage habits. If you focus just on power emailers, you’ll certainly have enough data to make things work better.
There are two things that cause wrenches in this whole plan. One is just simply the problem with out-of-bounds events. Say, for example, I got an email from Bill Gates. He and I haven’t exchanged emails before. But if he found fit to email me, I’d say that’s pretty important. Any system that relies on historical communications patterns has the danger of missing these types of events. The same can be said of emails from any new person, famous or not.

Second, I use webmail almost exclusively. If I used a desktop product, I’m sure you could deliver a product like this as a simple toolbar or plug-in. How will you make it work in a webmail environment (sorry, Greasemonkey scripts don’t count – that’s too high-end of a solution).

The only folks I’ve heard working on this problem recently are the guys at Xobni. I haven’t seen their product but I hope it moves the ball forward when it becomes available.

Back to Top