DHL, welcome to the club

It didn't take long after writing the entry below about Barnes & Noble to find a further example of a company failing to effectively make its data quality initiative(s) continuous.

DHL has a list of country codes and postal code formats here.

Apart from missing a number of countries which have recently (and not so recently) created postal code systems, the list still contains Serbia and Montenegro, a country that hasn't existed since 2006;
it has the pre-1999 postal code format for Argentina, and the pre 2006/2007 formats for Armenia and Malta.

Postal code and addressing systems change at a surprisingly speedy rate, and information resources such as this one from DHL decay over time if they are not maintained. I don't know whether DHL's systems expect the same formats as are listed in this document (which I acknowledge has been posted on the Singaporean DHL site and may simply have been overlooked), but it would be, to say the least, commercially damaging if they did.

Barnes, Noble and hoops ...

I recently came across this page on the Barnes & Noble online shop:

Entering international postal codes

explaining how they expected their customers to enter international postal codes so that their systems can validate the order.

One of the most obvious, but most overlooked, aspects of data quality is that it is a process and, importantly, that it is a continuous process. I'm all in favour of providing information, and lots of it, so that customers know what you expect, but posting a list like this and then forgetting about it demonstrates that the continuous aspect of data quality is being forgotten. The world has moved on, Barnes & Noble have not.

Like this, for example:



Yes, Argentina had a numeric, 4-digit postal code ... until 1999. Now their alphanumeric 7-digit code apparently is disallowed by Barnes & Noble. Or this:



Australia has never has a 7-digit postal code. I would give Barnes & Noble the benefit of the doubt an presume that they had mixed up the entries for Argentina and Australia, were it not for the insistence on a numerical code, whereas Argentinian codes contain 4 letters.

The other aspect of this page is the expectation that customers must jump through hoops to keep Barnes & Noble happy, to fit to B & N's technical requirements rather than B & N ensuring that their systems are correct. Take this example:



Grecian postal codes contain a space between the third and fouth digit. How long would a customer from Greece have to struggle to make their order? How much effort are they being put to to alter their (correct!) data to fit somebody else's system? Is this quality? Is this customer service.

No.

Personal names at Beijing: two steps forward, one step backwards

Holding the Olympics Games in East Asia has had one small effect on spreading an understanding of global differences in personal name patterns. As East Asians write their name in a different order to us Anglo-Saxons, participant names are being written, for some countries, in the correct order, and with the family name shown in upper case:

WU Minxia
Stephanie RICE

It seems however, that this recognition is fairly limited: not all countries have had their personal name patterns recognised. And, with all that money spent, not a sign of a diacritical mark anywhere. A wasted opportunity to be truly global.

Does your airline know where it's going?

I reckon that airlines have a very limited geographical understanding, limited to airports and without any understanding of the geographical context in which they are situated. I first began to suspect this when a data cleansing company (fortunately no longer extant) claimed American Airlines amongst its customers, and also claimed to be able to cleanse addresses for "more than 300 countries".

To save you having to run to an atlas to start counting, you'd be hard pressed, regardless of your definition of a country, to find many more than 240 on this planet.

My suspicions grew when I saw this SWISS map on the incomparable Strangemaps blog:


Now, the errors may not seem immediately obvious, but a great many cities have migrated significant distances away from their real locations. Pittsburgh is looking particularly moist, slap bang in the middle of Lake Eric, and Santiago in Chile has lost its name.

Then I was sure, after the same blog published a map from Aer Lingus:


I'm not sure how pleased the Americans would be to find so many of their cities placed in the north of Canada, but I'm pretty sure I know how Canadians would react.

I get the feeling that these errors have more to do with trying to get the maps looking nice and uncrowded than not knowing where places are; but this shows quite a lot of disrespect for the intelligence of their passengers. I'm looking forward to the next example, possibly showing London somewhere in the vicinity of Rome? I'm taking bets as to which airline it shall be!

Facebook

One of the rules of an effective data collection web form is that drop downs should only be used when the number of possible answers is limited, and that all of those answers are given in the drop down.

The Guardian reported on 7th July that Facebook has blundered in this respect as their marital status question gives plenty of options - even including "it's complicated", but not including (same sex) civil partnerships (as available in the UK) or registered partnerships (as they are known in some other countries, such as The Netherlands).

It may not be possible for Facebook to research and know all the possible civil status possibilities around the world, but at least they have now provided an option to clarify the answer given.

Not that this may be much of a salve for those involved.

How is your data?

Is everything working at the moment?

No weak links? No danger areas?

What happens when something causes a wave?

Will everything still be working? And how will your company's health be?

Credit card fraud

Processing addresses is a challenge for all of us, credit card companies included. It was, therefore, no surprise to me to learn how cursory the address check on credit cards for online purchases actually is.

For many countries there is no address check.

For the UK the check is only on building number and the numbers of a postal code, so the check for 10 Downing Street, SW1A 2AA would be 1012. As a very large number of addresses in the UK will share this same code, it's a check that is very easy to get around.

Considering the amount credit card companies lose to fraudsters, and the stress involved for the legitimate credit card holder, isn't it time that these companies put some time and resources into learning about addresses?