Exciting innovation
“The project started in early 2007,” says Erik Bolstad, Managing Editor at NRK. “The background is that the Norwegian Meteorological Institute freed all its data for access by the general public. At the same time, yr.no was launched as a collaborative project between the institute and NRK.’
As a result, they had copious amounts of data and an enormous challenge. Nobody had ever before drawn up such detailed weather forecasts for so many people, but they were convinced that it could be done.
“We saw early on that we needed to base it around aa geographic framework, so it was natural to gather everything around particular places,” says Erik. He goes on to explain that they accessed large databases of place names and began to compile them. “A so-called ‘yr panel’ was constructed, which comprised a few dozen volunteers from around Norway including managers at a few of the Norwegian Trekking Association’s cabins and camping sites, who all contributed valuable information."
This led to the idea of an hour-by-hour forecast. The Norwegian Meteorological Institute had created meteograms with weather forecasts for many years, but user tests showed that very few understood the contents of these graphs. So the project group started to simplify and expand on the existing graphics and put together hand-drawn sketches which were then used for simple research with users. The sketches were tested on a random sample of people at train stations, on the street and other places. The advantage of using hand-drawn sketches is that they were very quick to draw, and took far less time to adjust than digital renderings, or those uploaded to internet sites.
One solution for many
“We were always aware that we would have many different types of users," says Erik. “In order to come up with a design that would include the majority, without discriminating against anyone, we put in strong contrasts for the visually impaired, large and clear fonts, clear page divisions and a navigable menu.
Yr.no had a soft launch in June 2007 to allow the project team to debug and iron out the issues that could be expected in creating such a complex service. On the top of the page, there was a large red box informing that there were possible errors in the data and requesting feedback from users. As word of the new website spread, thousands of emails poured in, many containing suggestions for changes. These really helped to make improvements and bring the site to where it is today.
“We obtained unbelievably good insight by working in this way,” says Erik. “We read and responded to the emails ourselves, and pretty quickly we saw what were pointed out as problem areas. It is definitely time-consuming to respond to all the feedback yourself, but you get a far better feeling for what is working and what isn't, in this way,” he says.
Testing with people
After being on the internet for two years, the first professional user test was conducted in a lab setting, where a large number of people were closely observed navigating their way across the website.
“To optimally challenge the solution, we chose to primarily test it on people with no tertiary education or special computer skills,” says Erik. “We learned a great deal from this. Amongst other things, we saw that very few of these users looked at the graphs, but used the tables instead. We also saw that people with reading and writing difficulties naturally opted to navigate via the menu, instead of typing in the place names in the search field,” he says.
In 2009 yr.no introduced quality tagging, where all forecasts were tagged according to whether they were quite certain, uncertain or very uncertain. The tagging was done using red, green and yellow notification symbols. This gave rise to strong feedback from people who were colour-blind, around 5% of Norwegian men who could not see these notifications. This was resolved by introducing different graphic forms for the different categories.
“The most important thing I learnt is that you can conduct user tests far more often than is usual," says Erik. “But, you have to make it simple, with hand-drawn sketches. It takes an unreasonable amount of time if you always test on finished pages. When you test, you have to test understanding, simply ask the person concerned if he/she can explain what they see,” he says. “There are too many user tests that focus on what the users think of the actual solution instead of testing pure understanding of the content.”
Leader in Europe
Erik Bolstad maintains that yr.no offers Norwegians a far better weather forecast service than most others in Europe. After the launch of yr.no, a number of new Norwegian websites have followed suit aiming to provide good weather forecasts. This gives Norwegians more sources of weather information than most other countries and has made Norway a leader in Europe regarding web-based weather forecasts.
Yr.no is a huge success and is used by a growing number of people. Currently, around half the users are Norwegian, the other half is divided across users in Sweden and the rest of the world. Through a sophisticated hierarchy, the site can satisfy people requiring different levels of knowledge about the weather, from the casual checker to those needing detailed or specific information.
“Primary users go in to yr.no to find out about the weather for a specific place. But we wanted the site to also be a resource for groups such as school pupils, those interested in science and researchers. This is why we have a navigation structure that provides an increased level of expertise and information the deeper you go into the service. You can, for example, find out about the salinity and currents in the ocean, wind direction and how the weather has been day by day for the past year," says Erik.
Still developing
But innovation never stands still and Eric Bolstad smiles as he says,“There are currently four employees at NRK and three at the Norwegian Meteorological Institute that continuously work on running and updating the website. The site was redesigned in the spring, and now the system continuously registers how people click and navigate, so that we can detect user patterns more clearly and can at all times customise and update the site. The development of yr.no will never be entirely complete, and we hope that our users will continue to give us feedback and tell us what they want."
