Most people don’t even know that an online census is designed they just think you take the paper form and put it online. Unfortunately it is simply not that easy, the interactive environments, activities and tools are fundamentally different between an offline experience and an online one. Also peoples expectations of a digital experience is higher than a real world experience and that goes back to the notion that “technology is here to help us” which we all know is not really true. Additionally there is a whole raft of legislation pertaining to human rights, ease of access and use for people with disabilities which must be considered before starting (which have always been impossible to add in a retrospective fashion, enterprise systems architects take note).
1.0 Designing an Online National Census – Understand the domain and the Target Audience
So when we started to design the National Census there was just two experience consultants, a technical BA, with a content author (or wordsmith) to join later, we also added someone to manage and conduct usability testing and a very savvy UI developer who actually understood Accessibility W3C and could build it in from the start. There were also a couple of PM’s who did administration, a different time maybe, but we hardly saw them.
Understanding the Domain
Well never having done one before we asked the experts in Canada and Australia and had a number of insightful calls, were sent a load of useful document and we scavenged a great deal more from online sources. We then went through the huge requirements document and found the three pages that related to who would be using the Census by Country, Language, Household Type, and the then three different forms they would be getting due to localised questions. Based upon this we created twenty eight persona groups. We then had to decide who was the Target Audience from the persona groups and found it was not clear who should have piority in the design. So we went back to the client (you’d have thought that the strategy and vision for doing the project would be in the requirements, it was not) and asked the killer question
“Why are you doing this project?”
They looked a bit confused at being asked it but answered after we got to the head of the NGO. The answer was simple they needed more data and more high quality data (quantity and quality) than they were getting from the paper form in order to support public policy and planning of services and resources. We had discovered the problem, but we needed to categorise it better than that to find a solution. So we asked for completion data from the last two census to understand a pattern of behaviours, a nice fun bit of data mining. I was able to dig out my OLAP skills and use Business Objects XI to find the issue. From that we found a group of severely under represented users with a downward trend.
Out Target Audience became men 18 to 24 who don’t fill out the census.
The National Census experience was then designed for the least represented group of users in the completion data from the previous census. And what about everyone else, I hear you say. Well people who fill out the census, fill out the census and if we can make the experience better for the people least likely to do the census we enable everyone else at the same time.
2.0 Designing an Online National Census – Test the Design with the Target Audience
Testing Early and Often
Before the prototype was even complete we tested the login experience with the target audience, the results were a disaster. The problem was limitations set by the planned architecture and SLA’s, a ten minute login session, a one second page load with 60,000 users. Men 18 to 24 were not happy to be so limited, so we tested again, without the planned architecture and SLA’s and got some really useful result that enabled the design team to change the architecture SLA’s.
First Findings and How they changed the Architecture
So what did we find, first we tested in context, your meant to be able to do this in your home, with all the freedom of movement that allows.
When presented with the login, men 18 to 24 tended to shout abuse at the screen, they might get through it but then find they needed documents they did to have or were overwhelmed with the time commitment they perceived it would take to complete it. A common response was to head to the kitchen and make a cup of tea and a piece of toast. We timed this reaction “Login to Tea & Toast back to Census” was around eighteen minutes, from this we proposed a login timeout of twenty four minutes.
We found that men 18 to 24 would spend a max time of 30 minutes doing the form, regardless of government threats to fine them if they did not complete it. So the final form was designed to be completed in 35 minutes for a single person in their own home. The sense of accomplishment and beating the government did inspire five more minutes engagement.
Finally as part of this user testing, finding out that men 18 to 24 would not retry to login unless incentivised by having clear instructions, enabled the design team to get graceful deferral added to the architecture of the project.
3.0 Designing an Online National Census – Usability is a guideline not a set of rules
The actual design broke several usability guidelines but enabled a user to rapidly complete the forms. They could jump questions and comeback later, provided that the dynamic component of their form was already built. In this way we gave people a sense of accomplishment rather than usual government experience of feeling foolish by failing to do things right when nothing has been explained.
For a Government IT project it was astounding, there was no negative press response, they system went live as planned, worked and was easy to use. I understand the NGO was overwhelmed by the amount and quality of the final data set.