Customer service through email data parsing

by andrew pascoe on April 21st, 2009

Andrew Parker blogged last month about the rise of email data parsing, particularly that he was seeing a lot more startups using it as a method of input. (The point of his post was actually that because so many startups are doing it, there could be an opportunity for a collaborative open source project to cut out all the duplication of coding as well as processing.)

I came across an application of email data parsing the other day that I thought I’d post here, for two reasons:

it was an established organisation (not traditionally technically oriented at all), not a startup; and
it was for customer service rather than user input.

It was from Conde Nast (regarding a subcription to Wired UK).

Here’s how it worked:

I sent an email to them asking for a change of address. I did this because on their main subscriptions site, they haven’t yet added a link for Wired (because it’s a new title). I had actually replied to an earlier email from them - this had my subscriber number contained within it.
It was that email that they parsed. The results of the parsing were not to populate a website account (such as with TripIt submissions), but instead came back in an email to me. That email basically said “We’ve detected your account - and think you might this page (link) might help you”.
That page - in addition to the necessary links/info to resolve my query - had an easy way to show that this resolved my issue. This in turn closed the help ticket that was created when I first sent the email, and no doubt the email itself was taken out of the customer service pending queue, completing the circle and meaning the whole result was a) some simple email parsing reducing the need for any human cost at Conde Nast’s end, and b) a customer feeling happier because I didn’t have to wait for a human response the next day.