Last night, I was challenged to explain (and defend) content strategy. Three sheets to the wind after a pub quiz, this is no simple task, but I hope I acquitted myself passably. I say "hope" because there was a really interesting question I couldn't answer to my own satisfaction. I wonder if any of you folks out there in the ethereal internet hive-mind can help me out?

A friend - a rather concrete thinker who mathematically models complex biological systems for a living - pointed out that my examples were largely routed in business-to-business web sales and support. He challenged me with:

Say you've got a political website, so your goal is to have somebody read it and vote for you - how do you measure the effectiveness of that content?

Well, you would... umm... Oh dear.

I guess what we're talking about here, to yank it back to my present comfort zone, is a sales process where your point of conversion is off the site. The political example is perhaps a little below the belt, since what you can and can't do, and what data you can and can't collect is so restricted. You can't throw up a "How did you hear about this election?" questionnaire in the polling booth. Exit polls don't pull in your browsing history and site session information. Not everyone fatuously tweets and geo-tags each moment of their lives. Oh, and folks lie.

The business example might be easier to attack. You could have, say, a site for a farm shop that only did over the counter sales. Either way, it's tricky.

I fell back on some of the work I've done usability testing and benchmarking documentation, and suggested similar, quick and dirty, small sample qualitative UX trials. I'm not wholly sure that was right.

Any thoughts? How might we measure and curate for this kind of discontinuous conversion?