According to media reports 670,000 Australians didn’t hit the submit button on their e-census forms, rendering their information useless. In the days following the census, Fairfax, News Limited and even Google were recruited to remind us all to check we clicked the all-important submit button!

So why did so many “forget” to submit their form?

We’ve heard that there were very few technical issues reported. From a technical perspective the Australian Bureau of Statistics (ABS) deserves a huge pat on the back for doing such a great job with the technical infrastructure. I know I wouldn’t have been happy to spend all that time logging my information only to have the system crash at a crucial moment.

So if technical issues weren’t to blame for so many failed forms, what was?

We think it came down to a simple question of user interface design.

This project, like many Government ICT projects, focused very heavily on the technical implementation. We suspect that not nearly enough focus was given to the user experience of the e-census to ensure that ordinary people could easily and intuitively complete the form.

If the ABS had focused more strongly on user experience research they would have spotted a myriad of issues that users were likely to encounter when filling out the census online.

Let’s have a look at a few. We’ll start with the obvious… the final page where we all “forgot” to click submit.

Final page

The issue here is actually pretty obvious. It is a problem often seen in e-commerce: users get to a point in a process that they perceive is the end of the process and they stop.

You might remember buying something online and encountering a page that looks like an order confirmation but it’s actually a page that asks you to review your order and it includes a great big flashing sign reminding you that “Your order has not been submitted – please review and click submit”.

There are better ways to design e-commerce checkouts that avoid luring users into the “forgot to submit” trap. But funnily enough, in the case of e-census, where there was no final account of all the answers provided, and the information contained on the page is not necessary for completing the census. So why even bother with the page itself? It is altogether unnecessary.

The final submit button could have been positioned on the previous page at the end of the questions.

Such a design would have not only avoided the need for users to click through to an unnecessary page but most importantly, would have all but guaranteed form submission. If the final submit button was located in the same location on the final page as all the previous “next” buttons from previous pages – most users would have clicked it out of habit.

How much longer?

However, this assumes people actually make as far as the last page. It is likely a significant number of people will give up some of the way through. It’s not that the census is particularly long (depending on the number of people) but there are several factors which make the experience seem long and a little tiring.

Firstly, there is no progress indicator informing users exactly how many steps they have to complete (there is a kind of progress bar-come-navigation down the left-hand side, but it does not include every step). Plus there is no indication of how much time it may take from any given step in the process to completion.

Informing users where they are and how long they have before they complete a multi-step process is a lesson from usability 101. So it is surprising a progress indicator was overlooked. What’s less surprising, however, is that this lack of information would cause significant numbers of people to give up.

Make the computer do the hard work

One of the big advantages of online forms is that they can make the user’s life much easier. There were several opportunities which would have made completing the e-census much quicker and, had they been considered, would have helped avoid abandonment.

Consider the multiple address fields, one of which is pictured above. When completing addresses for “Person Two” and beyond, why isn’t there a button labelled “All addresses same as Person One”. For someone who has to enter three different addresses for their current address, their address of one year ago and their address of five years ago, this would be a handy time-saver. Imagine if that same person had to then complete the same three addresses for their partner and their children. All of a sudden, a handy time-saver becomes a massive time saver.

Instead, there is every chance that by the time that person got to the third and fourth person, they might just be a little tired of e-census and ready to give up.

Thankfully there are options to auto-complete each address field based on the addresses previously completed for any single person, but strangely these options rely on the user remembering the question number from a previous page so forcing the user to recall information that they are unlikely to even notice. Rather than reading “Same as question 8″ (or similar) these options should actively help the user by reading “Same as current address” or “Same as last year’s address”. In any case, these options are no help if all three addresses are different.

Less is more

E-census did provide help for users in case they did not understand questions. But there was soooooo much text to read. It’s likely most users would have skipped over as much text as possible; it’s a well known fact that most people scan rather than read online.

Better design could significantly reduce the need for much of this text. An obvious example is seen in the address questions. E-census informed users to “Enter abbreviated name for state/territory, example: VIC, QLD, SA, NSW”. But why weren’t the possible options for state/territory provided in a dropdown menu? This would have avoided the need for help text at all because the dropdown would have provided all the information a user needed to understand what they needed to do. Plus a pre-defined set of answers would also avoid human error.

There are also several examples of unnecessary help text:

• “For example, for arrival in 1974, enter 1974″. Umm… did I really need that instruction?
• “For more information see the Census Guide or press the information button for this question”. This “helpful” tip was provided for e.v.e.r.y s.i.n.g.l.e question and located beside the information button. Surely users understand to click the more information button if they’re after more information…

All in all, the Stamford guys spent about 40 minutes looking through e-census and we discovered at least 25 more time-saving opportunities. So we could go on and on, but we won’t. Suffice to say, the biggest issue is the fact that just over 30% of users abandoned e-census. If such an abandonment rate was experienced by an e-commerce company, it would have the potential to send it bankrupt.

Conducting user experience research on the e-census before launch would have helped to avoid such a high abandonment rate and would have saved the ABS the considerable money it will be costing in attempting to encourage those 670,000 Australians to finish and submit their form.