How tackling a nation-wide challenge was made possible through deep immersion, machine intelligence and designing with people.
A century-old challenge.
There have been some attempts — with the first national initiative dating back to 1926 — to stimulate land registry across the country. Big plans of action, large investments. Some advancements below Tejo river, but little results on the north of the country, precisely where the biggest risk of fires exists.
So, when we were invited by the State Secretary of Justice to join a project team that would tackle the century-old challenge of nation-wide rural land registry, an issue that gained priority on the governmental agenda following the aftermath of the biggest and deadliest fire to ever occur on national territory and something that has eluded governments and private enterprises for ages, we kept our cool, professional poses (this is the government, after all) and quickly accepted the challenge.
What got us interested wasn’t just how challenging the project was but also the potential to go past land, and look at the people who owned land. Understanding some of their needs (what landowners value in their lands, the sense of possession vs. the sense of belonging, future generation’s relationship with the land, etc.) and the ways in which land registry could be a helping tool to address them. It’s not just about knowing where their lands are, but knowing what their lands are.
Our expectations were indeed high, starting this project, but the best was yet to come. We would only realize the full scale of possibilities BUPi could unveil some months later. But first things first: we needed to set things up.
Field research. Literally.
So, what is BUPi? Well, the acronym stands for Balcão Único do Prédio which translates to something like Centralised Property Desk. A government-created simpler, digitised approach that would allow land-owning citizens to identify and register their own properties (almost) autonomously, replacing the traditional and slow method of going from property to property with GPS devices.
To fully test this approach, and understand where it could “break”, a one-year pilot program — targeting 10 of the municipalities that were the most affected by fires — was set in motion. Although BUPi was conceptualised to be used fully autonomously by landowners in the future, it was decided that, for the purposes of this pilot run, service counters, complete with official technicians, would be implemented.
Together with other service providers, we were hired in order to get the pilot program up and running. Our brief was pretty straight forward:
How can we mobilise local populations towards the service counters?
We could make at least two a priori deductions: landowners, not being obliged by law to properly register properties in BUPi, would only do so if they had something to gain from it (i.e., if they wanted to sell their lands); and most of the target population was above the 60 years old mark. This challenge deserved not only a team with a wide set of skills but also a less linear approach to a common project. That’s why a project lead, a senior researcher, a communication strategist, a multidisciplinary designer and a strategist packed their bags with clothes, books and post-its and moved to the secluded municipality of Penela, for the first month of the pilot program.
We were there to conduct research and decided to do so through highly active research methods, with a participatory attitude. It was experimental season: we sat at service counters; created, distributed and tested different versions of flyers; trialled additional analogue and digital methods to locate and define the shape of properties; experimented with unusual — but very pertinent — locations for service counters (like the local market, where we spent many early mornings); and even tried to buy a terrain, posing as investors. Backed up by more traditional but always useful methods like interviews, this intensive research phase allowed us to arrive at four main challenges that needed to be tackled in order to mobilise the necessary amount of landowners:
Unstructured and complex communication between all stakeholders. There were different interpretations of the law, different goals being communicated and an unhealthy dose of myths and misinformation being spread around the local populations;
Unrealistic goals. Although highly simplified, from a legacy standpoint, the new method had points of friction that delayed its speed, thus decreasing the chance of mapping the 10 municipalities in one single year;
The main goal was registration, but it wasn’t the most decisive data point. Registration started out as the main focus of the first month of the pilot program, but unfortunately, this wasn’t the most useful piece of data when it came to being able to piece together an image of rural properties.
Technicians were scarce and limiting the scalability of the project. Their shifts were the only interval of time when the project moved forward. A digital platform that was only useful from 9:00 to 18:00, 5 days a week, a less than ideal scenario when we were trying to map so much land.
As you should already be guessing by now, mobilisation means more than just coming up with a communication campaign to us; a transformation of the service (both online and offline) also needed to be considered. Our path to the solution was designed around three axis:
Decentralisation of the service counters so that they could be where people actually lived or spent their days (e.g., local cafés, markets); these new counters also helped tackle the lack of official technicians: here, a new, simpler “drafting” step was created, so we could have less qualified staff helping the landowners, before they went to the actual official service counters.
Newly targeted communication strategy taking into account not only small-time property owners but also local (and national) stakeholders that could help us raise awareness and create mobilisation towards action. These experimental tactics included engaging with local events (e.g., weekly mass) and multiple key messages, tailored to the specific relationship (ranging from purely functional to highly emotional) different profiles of landowners had established with their properties.
Property location and shape as the two new main goals. This was reflected not only on alternative communication messages but also on a new flow for the service counters: finding a property location and drawing its outlines felt like less boring steps, in the middle of a very arduous property registration process.
Our strategic recommendations were accepted, tested and were starting to deliver satisfactory results. We just had to implement all of these changes and watch the numbers skyrocket, right?
Obviously, it wouldn’t be that easy: the goal for this pilot was to get as close as possible to 100% of the pilot areas registered, located and outlined, in one year. And every day that we couldn’t achieve the ideal daily rate, just meant that the next daily rate would be even harder to achieve.
Our recommendations improved the rate of completion, but would never be applied instantly across the whole BUPi system. We needed a new approach that could take us into beyond-human gear. Enter machine intelligence.
Welcome to the machine… intelligence.
During our field research we had encountered several sources of structured and unstructured data related to rural properties. Some were digital datasets, sometimes de information was stored up in more traditional media. The point is that data was up for grabs. And grab it we did: playing around with different databases, cross referencing data, we got to a point where we had learnt how to find and map rural properties, by ourselves.
Near the end of our stay in Penela, we stopped to ask ourselves the simple question that shifted the whole project: if we can do this by hand, why wouldn’t a computer be able to to it? We quickly outlined the concept for a prediction algorithm, capable of inferring the location of a given property and, using satellite images from different years, predict and suggest a shape for it. This would make the whole registry process a lot easier: landowners would then only have to confirm or correct the suggested shapes.
It would take some effort and a little bit of luck getting the required datasets. But all we needed was the “ok” from the client. Which we did get: an approval to conduct a proof of concept. Together with big data and geographic information system partners, we started testing assumptions, data types and approaches. After nearly 3 months of testing and iterating, we got to positive results.
The first half of our concept was proven right: it was indeed possible to predict where a rural property was located.
As of today, anyone using BUPi can get the predicted location of a rural property by inputting just a number on the platform. That simple. We also got to promising results, regarding the shape outlining of properties, but are still nailing a final version that can provide us with wide-reach, public-ready levels of certainty.
In the end, all this work produced the kind of results we like the most: those of the tangible variety.
Firstly, our recommendations contributed greatly to augment the registration rate. In the first two months of the project, about 2% of the pilot properties was registered. After our recommendations started being implemented, wewent from 2% of completion in January to over 48% in October.
Secondly, all the changes to the communication strategy, coupled with a better working service — that increased trust and word of mouth awareness — generated so much demand that 45 000 individual bookings could not be fulfilled on time. We consider this to be a “good problem” since it validated the assumptions behind the new mobilisation strategy.
Lastly, through the refined algorithm, we were able to predict the location of all rural properties in the 10 pilot program municipalities.
From land registry to land knowledge.
All these results show amazing progress, a real openness for innovation and willingness to embrace new methods, shown by local political and civic structures. But this is only the start of this journey. In 28 June 2019 the Portuguese government approved a country-wide expansion for BUPi(news article in Portuguese) and so, a new stage begins. One in which assumptions will again be tested, this time at a national scale.
It’s all about reframing the information that we have access to. One could look at BUPi like a directory for rural land. But we prefer to extend our perspective and look at the immense potential for relevant, impactful information that this project could unveil: be it geomorphological data that could influence environmental decision-making or socio-economical data that could better define national and local policies, we think of our country as a high-unlimited database and envision BUPi as a platform that could allow us to extract all of this information and turn it into transformative knowledge.
Before you go, let us share some of our principles that supported this endeavour:
Design from people
Yes, we ended up using lots of tech, but that was only possible because we spent several afternoons talking with real people, like Mr Júlio, reviewing all the data he had about his rural properties. Our new solution was also a response to the difficulties many people had in locating properties for which the records had been lost more than a generation ago.
Work the middle
Like other project partners, we started this journey on the fringe of the project, with a very specific task. Through honesty and constant sharing of information and results, we learnt how to work with everyone, in order to achieve the best possible results.
Scale your ambitions
At certain times during the project, this looked like an almost impossible challenge to tackle, but by reframing what was possible, we were able to turn what at first seemed like pointless results (after the first two months) into an opportunity to re-contextualise the problem. A new approach that unlocked new possibilities going forward. We hope to be able to share more on this last topic, sometime around next year.
If you have any questions or insights about this project, feel free to hit the responses section or to contact us at firstname.lastname@example.org👇