Section: Alumni Databases
What do you think about using Address cleaning services like National Change of Address and Electoral Role validation?
I have really big misgivings about some of these data cleaning exercises, particularly when carried out against the Electoral Roll (ER) and National Change of Address (NCOA).
I think the companies like Ebiquita, UK Changes etc etc are trying to market their services, particularly Electoral Roll and National Change of Address checking, as the best way of keeping our databases up to date. I remain very sceptical.
Many of our alumni, especially but not exclusively young ones, are highly mobile. They represent probably the most mobile section of the population apart from Travellers! Thus very significant numbers of them use parental or some other address as a permanent mailing address, which will differ from where they are registered to vote, where they are in the phone book this week etc etc.
- The ER database is always, by definition, out of date. It is collected in October each year, once a year, and takes 3 months or so to publish. Thus if you query the ER in January 2004 you are looking at information that dates from September or October 2002. Given that 5% or so of the population moves each year (and our alumni are more mobile than that of course) then it is arguable that at least this percentage of the "non matches" that come back from ER checking are not matched because our data was already better than the ER.
- The NCOA database is worse. I think it's great as a last resort when you have lost someone and there is no other way of finding them. But as a way of updating existing addresses it's really very error prone. We did a test of our entire database against NCOA, and enquired of those for whom NCOA had given us a change whether this was right. In some cases it was, but in others NCOA was completely wrong. For example I had someone who NCOA had told me had moved from Aberdeenshire to Singapore. Had I accepted this I would have lost this alum, since the move was a job-related one he had done for three months. He had, quite reasonably, put a forwarding order on his Aberdeenshire address so he got his mail while he was in Singapore. But then he came back and cancelled the forwarding order. But NCOA still reports that he lives in that rented flat in Singapore, since the dataset does not capture where the person is when a forwarding order is cancelled.
So I think we need to use these databases intelligently. ER checking is great to verify that someone was there last time they did the ER data collection. But I am really not sure that tells us very much, since there is no guarantee they are still there. I am stil getting ER derived junk mail for people who left my house in the summer of 2000.
NCOA is genuinely good for finding people as a last resort, where an existing address has failed.
So what do we do instead?
- We are obssessive about return addresses on envelopes. (Please forward, or if new address unknown then return to....)
- We send out a form with FREEPOST address every time we mail our University magazine
- We have web forms which people can fill in to tell us their changes.
- Our interactive web site tells people the postcode we think they have, which prompts them into changing it where it's wrong.
- We publish lists of missing alumni and encourage them to "shop" themselves and others.
- We do a questionnaire every 3 to 4 years, but in the meantime our magazine carrier sheet contains a dump of basic data including home and work addresses, employment information etc
- From time to time we go through our missing people and write to the addresses they used to be at, asking the current occupier to let us know where the person now is.
- We do an e-mail newsletter which encourages people to keep in touch
- We offer an e-mail forwarding service, which we ourselves can use to e-mail people where the snail-mail address has failed.
In summary, then, if you can afford it, I would not stop communicating with people until you really know positively that they are not there, since I think a lot of the data cleansing procedures that are carried out give wrong negatives.