No Shortcuts to Data Quality in National Addresses

Graham Rhind


Last month I started looking at the variety of worldwide address infrastructures. Address systems vary between countries, but also very much within countries.

A good example of this is building numbers. A block of flats may give each apartment its own number, while another has one number for the whole block and apartments are shown by sub-building indicators – “flat n” or “suite y”.

One door may lead to many apartments with the same number but no sub-building indicator (common in Germany, and causing additional problems when house-holding routines are attempted with those addresses, as people with the same surnames but not related to each other may live at what appears to be the same location).

Going postal

In other cases, the same building number can have multiple entrances. My own address has three separate entrances for five residences – but with the same building number, which means we often have to manually re-direct mail which has been incorrectly delivered.

In the USA another phenomenon exists: a fraction in the building number – 174 ½ or 952.5 – where a new residence is built between two existing ones!

Why this variation? Well, it’s quite simple, really. Though we often think of addresses as being something postal, and while postal authorities have usually been seen as the managers and owners of national postal data files (though less so now), the structure of addresses (the thoroughfares, their names, and the way buildings are numbered) is in almost all cases the responsibility of local government, and their priorities are not postal delivery.

In a right state…

A block of flats which replaces a single residence is likely to have a single number to avoid having to re-number every building on that street, while new flats on new thoroughfares are more likely to be numbered separately.

This is also the case with the relationship between postal codes and administrative areas. My customers often search for solutions to infer, for example, a province name from postal code information, and are usually perplexed when they find out in how few countries’ postal code and administrative areas coincide.

Indeed, why would they? They don’t have the same purpose. There are countries which appear to have neat correspondence between the two – a ZIP code of 1 in the United States, for example, should indicate Delaware, New York or Pennsylvania. It usually does. But not always; ZIP codes are actually routes and not areas, and they do bleed across state boundaries.

No shortcuts

In other cases, where postal codes were created to coincide with province boundaries, such as in Italy, administrative changes have blurred those overlaps considerably.

I’ve been working with addresses for a couple of decades now, and before that I was in market research, so I had to learn how to create questionnaires. One thing I’ve learnt in both cases: the amount of diversity and human creativity should never be underestimated, and there are not always shortcuts to be had.

You can’t always define a gender from a form of address (salutation/honorific), and you can’t always define a state name from a postal code. Embrace the diversity and understand the systems for the best and highest data quality results.

About The Author

Graham Rhind is an acknowledged expert in the field of data quality. He runs his own consultancy company, GRC Database Information, based in The Netherlands, where he researches postal code and addressing systems, collates international data, runs a busy postal link website and writes data management software. Graham speaks regularly on the subject and is the author of four books on the topic of international data management. You can find him on Twitter via @grahamrhind.


Paul Smith says:

I live in Penzance, England and there are certainly a few oddities here. My house number is 34 whilst my neighbour’s is 79a. Unusually for England, there is also a “1/2″ numbered house in (I think) Causewayhead, Penzance

Posted on: 13th June 2012 - 1:12 pm UTC Reply Natalie says:

Thanks for your comment Paul. I feel sorry for any new postal workers in your area!

Posted on: 13th June 2012 - 3:59 pm UTC

Practical International Data Management Online.  A free resource from GRC Data Intelligence. For comments, questions or feedback: