Are online retailers being tripped up by postcode entry?

How very true. It never ceases to frustrate and amaze me how the small details cause big problems when it comes to data capture.

We test dozens of forms like this every year, and we have been seeing the same faults for more than a decade. Whilst some are obvious like these, others are less easily noticed.

For instance on one website I noticed that for any given postcode you either got all the houses with numbers or all the houses with names, but never a mixture. The first premises in the postcode determined whether you got numbers or names.

Furthermore, properties were not listed at all if they had both numbers and letters in the first line.

We also see issues with capitalisation (sometimes you must capitalise, sometimes you must not - why???) and whitespace (again you sometimes must or must not have whitespace in the middle). Sometimes leading or trailing whitespace causes errors - while no one will type a space, you often pick them up when copying and pasting.

Why is there not a design pattern that everyone can use (maybe there is?).

According to the Website Optimisation Company 2012, 11% of visitors abandon their cart due to the complexity of the checkout process, this is a massive number! And postcodes really are just one of the many reasons why customers abandon their carts, like the absence of a guest checkout option, or the lack of shipping information! http://z6.co.uk/d284c4ad

Albie you make a good point. Most of the problems caused by postcodes not being returned are largely due to companies using address data that is not being updated daily.
With almost 3,000 address changes taking place every 24 hours to the 29 million records in Royal Mail’s Postcode Address File (PAF®), the database is not a static file and is constantly changing to reflect new builds, updates and deletions. Using a postcode validation service that includes daily updates as standard will significantly improve the accuracy of customer data.

I have a gripe about another short-sighted thing that some websites do, which is very similar to the postcode issue at the checkout, as described above.

This is where a store finder only allows you to find your local branch by entering the postcode.

This works fine if you happen to be looking for the nearest store in your local area, as virtually everyone knows their own postcode.

But what if you're away on holiday somewhere and are looking for the nearest outlet, say, of your favourite coffee or restaurant chain?

Chances are that you won't have a clue what the postcode is for where you're staying.

Hmm, would it be possible to have a Google style - 'did you mean...' then suggest a postcode based on other entries?

With a large database of users that may be feasible, or even if it just suggested a correct combination of words/letters it may highlight the mistake for the user?

@ L Dickson - The businesses mentioned in this article that are 'doing this right' all use Experian QAS...

Undeclared limitations on what the customer can input are always going to cause a problem. And it is easy to miss these things if you always test the site with someone doing it "properly".

A related grouse of mine is sites that can't accept some characters in passwords (e.g. don't do case-sensitive, or only allow numerals and letters, or impose silly restrictions on length). Much worse of course if the customer is not warned - so you enter the very secure password you just thought of and are told it is an error ...

There's a larger problem at play and it harkens back to software testing methodology: Every validation program is judged by how many wrong entries it rejects (higher the better) rather than how failsafe it is from the customer perspective. We always advise our customers to avoid causing shopping cart abandonment by over-validating customer inputs in crucial steps like checkout. It really doesn't cost so much - compared to the potential loss of revenue caused by overzealous autovalidators - to pick up the phone and ask the customer what exactly is the data they wanted to enter versus what they actually entered. Some of them take our advice, others don't. I guess you can't win 'em all.

PS: Note to Econsultancy Web Master: While on the subject of how to make a process fail safe and reduce abandonments, you might want to find a substitute for reCAPTCHAs to authenticate comments on this blog - the way the reCAPTCHA program operates, it becomes more and more difficult to crack them. According to anecdotal evidence, almost 25% of website visitors fail to get them right the first time. No one knows how many of them bother to re-try with the second one.

The spaces and capitalisation on a postcode can be fixed with a simple regex, so no excuses there!

Does any one have numbers from the real world of what percentage of errors are down to the '1' vs 'i' and 'o' vs zero? Maybe our users are a really careful bunch, but we hardly see any of those.

@L Dickson - in most countries the idea of postcode lookup or any rapid addressing is a foreign concept. It might confuse the users and be counter productive. I think in these countries some kind of post-entry validation might be a better route. I.e. only interfere with the checkout flow once you have the cash and think the address might be bad.

Ketharaman makes a valid point. You reach a point of diminishing returns where the cost of trying to cater for every potential user input error can't be justified based on the likelihood of these errors occurring. A line has to be drawn somewhere or we'd all be fussing about 5s mistaken for S's, 8s for Bs etc etc

Play were always number 1 in my book for usability, they managed to tame their programmers right from the start back at the end of the 90s. The failure to get programmers to live in the real world costs businesses big time in abandoned carts.
Ps agree dump your captcha, it's a shocker...