I've signed up some interesting contributors for my workshop on DarkSocial at SpotOn London 2013 but I still feel I would like to try to generate some data so that my contribution is not purely anecdotal. I have analytics stats I can show of course, but l last week I was fretting about how I could demonstrate that it's not all about numbers and that dark social channels lead to high levels of engagement. If I ask students to complete an online survey via email I get low response rates, but survey responses continue to nose dive, so that's not surprising. Around half the email conversations I get from students on courses I am teaching originate from mass emails such as my weekly newsletters, but that's a fairly small scale experiment. I'd really like to step this up a gear and draw some wider conclusions.
After investigation, it's clear that MailChimp, ReadNotify, etc, can't actually deliver what I need in terms of verifying engagement. It's not their fault, modern email systems are designed to maintain privacy, which is a major reason why people like them so much. So my only alternative seems to be trying to measure what people do in response to email rather than going down the route of simple numbers. Which is where the problems start. To their credit, MailChimp and ReadNotify make it very clear that this is murky territory and that the worse thing you can do is cause loss of trust by intruding too far. So I need to find ethical ways of measuring these relationships, and here's my framework for doing that:
- Any task I ask email correspondents to perform in order to measure response rates has to be authentic, i.e. something that that might normally do.
- Any task-based measurement has to offer the recipients a potential benefit for participating, for example, links to extra information or service that they want.
Lou Woodley has just pointed out (via a dark social channel, of course :-) that this may be one reason that dark social usually sits within the domain of the Marketing department rather than rather than traditional analytics. And that's a pretty good reason for wrestling it back off them :-)