Data Challenge serves as an engine of innovation for Canada’s public services
Canada’s Public Service Data Challenge neatly addresses the barriers to innovation in government – drawing great ideas out of the workforce, and pushing the best right through to implementation. In the run-up to its semi-final this week, participants, champions and judges explain the programme and give their tips for victory.
In April, Agriculture and Agri-Food Canada (AAFC) launched Agpal Chat, the government’s first public-facing generative AI chatbot. “It’s about as easy to interact with as you could imagine,” says Jay Conte, an AAFC policy analyst. “To find any kind of government support, rather than trying to navigate tricky clickbox searches or clunky search fields, you can literally just ask the question in your own language.”
Users can quiz the chatbot on agricultural programmes and services at the federal, provincial and territorial levels, using either English or French, and it responds immediately with well-written, comprehensive answers and helpful links. “If you’re used to traditional Government of Canada customer service, you’re going to be pretty floored by what it’s able to do,” says Conte.
Agpal Chat is proving just as helpful for the AAFC itself. “This has helped our department in raising its profile on artificial intelligence: it’s propelled us to a new level, where we’re seen as a leading department,” says Steve Rennie, AAFC director for data-driven technologies and acting director for data partnerships. He explains that both this new public service, and the department’s growing reputation, owe their existence to a cross-government innovation programme: the Public Service Data Challenge. “Had we not had the Data Challenge, I don’t think we’d have Agpal Chat up and running today,” he says. “Simple as that.”
Introducing the Challenge
Now in its second year – and approaching the crucial semi-finals, to be held this Wednesday – the Data Challenge has just won a Government of Canada Digital Community Award 2024: the Large Organization Award for Outstanding Digital Leadership. The Data Challenge programme is operated by Statistics Canada, Natural Resources Canada (NRCan) and Global Government Forum, and in design, it mirrors a similar UK programme. After all, the keys to successful innovation are similar in democracies around the world.
Read more: Four projects shortlisted for implementation in 2024 Civil Service Data Challenge
In large organisations, senior leaders tend to become distanced from the expertise and ideas of frontline staff, who themselves lack contact with people in other professions, roles and organisations. Management processes and organisational structures fragment the workforce’s incentives and create vested interests, fostering institutional inertia. Resources are tight and the demands of daily delivery onerous, so staff lack the time and space to pursue innovative ideas. And people get little credit for taking a risk, while a failed initiative can blot a once-promising career.
The Data Challenge addresses all these barriers to innovation and experimentation, beginning with an appeal to all federal public servants – who are asked both to send in their ideas for how government could make better use of data, and to volunteer as a team member. In its first year of operation, Conte was among those who answered the call.
The policy analyst had long been frustrated with the quality of departments’ search functions, and was convinced that generative AI technologies could help solve the problem. “I knew that the potential was there,” he recalls. “The questions were going to be: could we figure it out from a technical standpoint? And could we get enough support from management to overcome some of the risks associated with the tools?” ‘No harm in trying,’ he thought, and put in an entry.
Innovation in action
In all, around 100 ideas were submitted in that first year: “I was truly surprised by the response, and by the quality of the ideas coming our way,” comments Frank des Rosiers, assistant deputy minister for strategic policy and innovation at NRCan and one of the programme’s ‘champions’. Using the six judging criteria as a guide, a panel comprising departmental chief data officers picks out the eight most promising ideas, and the programme’s managers – drawing on those who’d volunteered – build a dedicated team around each.
Importantly, each team includes people from a range of disciplines and departments. “Everyone brought something to the table, and there was always someone who had the experience that we needed,” comments Nicole Johnson, a senior AAFC data analyst who – along with Rennie – joined the team tasked with developing Conte’s idea.
This approach gives the teams both the skills and connections required for success, and the diversity that itself supports creativity and problem-solving. “Having that cross-pollination across different areas of expertise is extremely rare,” says Stephen Burt, Canada’s chief data officer and another programme champion. “Opportunities like the Data Challenge show you the potential.”
Read more: Canadian government departments are balancing AI optimism with caution
The teams then enter a first research phase – developing their ideas, gathering evidence, and building relationships with relevant people inside and outside government. Here, they receive support both from the Data Challenge team, and from senior leaders: along with the three programme champions, a network of ‘advocates’ among senior data leaders stand ready to help the teams overcome any obstacles they face.
“We’ve got lots of people with good ideas, but sometimes in order to get their dream projects up and running, they need help – whether it’s computing, AI-type solutions or managing the data,” says Des Rosiers. When teams encounter a problem they’re struggling to solve – perhaps locating a particular dataset, or connecting with key people in another department – the champions and advocates are on hand to offer assistance. During this phase, says Burt, he works to ensure that departmental leaders “take it seriously, and think about how they can connect this great enthusiasm into their departmental structures to get things done”.
Personal and personnel development
This period of research culminates in the semi-final, where the teams pitch their ideas and field questions from the judging panel (see end box for advice for participants from previous teams and judges). Typically, four teams are then shortlisted: taking the judges’ feedback on board, they enter a second research phase – further honing their ideas and presentations in the run-up to the final, where they again face the judges.
Along the way, participants gain a huge range of new skills and experience. Most obviously, as Des Rosiers points out, the Challenge helps people to “develop their skills and stretch their abilities with the new tools that are being developed so rapidly” – such as artificial intelligence. “AI has become the hot topic,” says Francis Loughhead, an analytics specialist for the Canadian Forest Service and a member of the ‘ghost gear’ team – whose idea involved using AI to locate plastic waste at sea. “From a career perspective, the ability to say that I’ve worked on an AI project has been really, really good.”
The semi-final and final also provide experience of what is “a key skill in life, and one that we probably don’t spend enough time on in the public service”, comments Burt: the ability to “grab the attention of a senior leader who has limited time and 15 other things to do, and to get them excited about something”.
Along the way, participants build a network of contacts. “Since COVID and the shift to working from home, it’s been harder to meet people organically,” notes Conte. “This was a great opportunity to do that, and I’m really grateful for it.” Above all, though, he praises the opportunity to work in a “more entrepreneurial” way than is usual.
Freedoms and flexibilities
AAFC senior managers were supportive of the team’s work, he says, “because they saw the promise of potentially winning the Challenge, which was exciting for the department”. Meanwhile, the “timelines were so short that we didn’t have time to go through the normal, rigid approval processes”. The result was that the team had “the creative space to play, experiment, do things our own way, as long as we were reporting back regularly”.
“The timeline forced us to be creative in how we approached problems, and fostered this feeling of working at a start-up,” comments Rennie. “When you have a hard deadline and there’s a lot of expectation created around that, and then you bring in folks from many different disciplines, you can build something pretty unique and special.”
The Agpal Chat team’s experience is a common one. For the ‘ghost gear’ project too, the Challenge was “an accelerator; it gave rocket fuel to the whole initiative”, says Loughhead. “In the space of 12 months, it probably pushed us two years ahead of what would normally happen in government.”
Key to this compressed, highly creative process is the Challenge’s ability to carve out time for speculative research and experimentation in people’s diaries. “It’s hard to create the space and time to do something innovative,” comments Gabrielle FitzGerald, chief data and risk officer at the Canada Food Inspection Agency and a Challenge judge in this year’s programme. “The fact that this is a government-authorised effort gives people the opportunity and incentives to do so.”
An idea accelerator
At the final, Conte’s idea was selected as the winner; but the Challenge’s goal is that all the most promising projects ultimately get delivered. The ghost gear project is also moving forwards, explains Riham Elhabyan, a senior policy advisor at the Office of the Chief Data Steward at Fisheries and Oceans Canada, who originally put forward the idea.
Following the final, her team “did work with an AI service provider to develop a proof of concept, and presented the results to a couple of clean-up organisations”, she says. “We are discussing the funding possibilities to take this forward to the next steps.” Participation in the Challenge has, Elhabyan adds, been a “cornerstone of moving this forward: it provided the business case, and made us ready to move forward with the next stages”.
Ultimately, says Burt, the job of implementing these projects sits with departments, but “we can help, we can encourage, we can intervene, we can set conditions. I’m always happy to wield whatever small influence I might have.” His office can also help remove obstacles from the winner’s path, he adds: “If we discover a particular policy or regulation or standard that needs to be tweaked in order to make something happen, that would be a really useful finding.”
Key to the Challenge’s value, FitzGerald believes, is its ability to identify and develop projects that can be rolled out across government: “These are like pilots, proofs of concept that demonstrate the art of the possible,” she says. Rennie’s team have certainly got the message: “There’s not a week goes by that I’m not meeting with one or multiple departments to talk about the work we’re doing on AI,” he says. “We’ve been freely sharing the [Agpal Chat] code and all the documentation with other departments and agencies. It’s our hope that it can accelerate their own AI journeys.”
How to do government better
The Challenge model itself also presents wider lessons for public servants – pointing the way towards a more innovative, ambitious, non-hierarchical public service. The programme demonstrates the need to “plug into the grassroots, where the problems are more real and more engaging”, says Burt, “making sure that the enthusiasm and the frustrations of the folks on the frontline are connected to the desire to make change at senior levels”.
That’s certainly what occurred in the case of Jay Conte. “Sometimes you can get a little beat down working in government. There’s so much hierarchy, so much structure; you’ve put a lot of work into a project, and it runs into a barrier, and you start working under the impression that big things aren’t possible,” he says. “It’s been great to confirm that if you can get the stars aligned, you can do some pretty big things and overcome some of the big institutional barriers that you’d normally encounter. That injects some energy, and encourages you to try to take some other big swings going forwards into the future.”
Thanks to the Challenge, Canada’s agricultural communities and businesses now enjoy vastly improved access to support services and grants, while AAFC has developed a reputation as an AI pathfinder. “It was incredibly gratifying to be able to build something as innovative and creative as Agpal Chat in such a short time; to be able to do something like this on this kind of timeline is exceptional,” concludes Rennie. “I’m really proud of the team for pulling it together; and we’re very thankful to have had the Public Service Data Challenge as the driver behind it.”
The semi-final of the 2023-24 Public Service Data Challenge will be held on 15 May. The longlisted ideas are available on the Challenge website. Look out for our report on the judges’ decision as to which will move forward to the final.
How to win the Data Challenge
Speaking with judges, champions and the members of two teams that did well in the Challenge’s first year, we’ve pulled together essential advice for this year’s teams.
- Stay focused on the judging criteria
The judges measure every idea against six metrics, and teams must achieve a base level in each of them to secure a chance of victory. “Having a comprehensive set of judging criteria ensured that we were moving in the right direction in all the dimensions, whether it’s adherence to ethical standards or business value or the mandate of the organisation,” says Riham Elhabyan. Think carefully about how you’re meeting each criterion, and address them all within your presentation.
- Explain the benefits for the public
“Show how the solution will help Canadians: how it will impact people and organisations,” says Alexandra Cabedose, a policy and research analyst at Employment and Social Development Canada and a member of the ghost gear team. This is a particularly important criterion: the judges will be looking for ideas that will improve citizens’ lives – and they’ll want to see evidence to back up your claims. “It’s got to fix something in the real world,” says Stephen Burt.
- Demonstrate scalability
“Does it have application beyond a single area of expertise or departmental mandate? Does there appear to be potential to scale? Can we deliver it more broadly with relative ease?” asks Burt. Concepts can begin small, local and narrow – but the judges are looking for transferable ideas that can be adopted across government. Nicole Johnson recalls that at the semi-final, “the feedback from the judges was all about how we scale it and make it accessible to others”.
- Meet frequently
“We had organised, weekly meetings, where we discussed progress and got the different perspectives of the members,” says Elhabyan. “Each week, we would say: ‘Okay, this is what we achieved so far. This is what we want to achieve in our next steps.’ And then one of the team members would volunteer to do each of the next tasks.” Meeting frequently helped to ensure rapid progress, enabled the team to draw on all the group’s expertise and contacts, and strengthened the sense of common purpose.
- Build partnerships
“It’s critical to start as early as possible, reaching out to other groups involved in the space,” says Lee Croft, a policy advisor at Fisheries and Oceans Canada and a member of the ghost gear team. “Every time we had a meeting with another group or potential partner, that really helped us to advance how we were going to tackle it.” The judges will want to see that teams have connected with the people who’d be involved in delivery, gathering their views and winning their support.
- Get to grips with the data
“Sometimes we put a lot of emphasis on the bright and shiny dashboard – but if your underlying data isn’t good, it’s going to give you funny results,” says Wesley Yung, director of Statistics Canada’s International Cooperation and Methodology Center and a Data Challenge judge. “What data do you have? Where are you getting it? Is it really quality data? Can you trust it?” The judges will be looking for evidence that you’ve not only identified the datasets you’ll need, but also addressed the challenges involved in accessing and deploying it.
- Prioritise communications
“So much of the Challenge is about telling your story in a short period of time to judges, and you’ve only got 10 minutes,” says Jay Conte. “You need to start thinking about that from the very beginning.” Don’t go into too much technical detail, advises Elhabyan: in the short time available, it’s as important to demonstrate public value and strong partnerships as it is to outline the technological solution. And Johnson emphasises the value of practice: “We did lots of dry runs,” she recalls. “Practice, practice, practice!”
- If possible, demonstrate
“If you have something to show, that’s far better than telling the judges what you’re going to do,” says Steve Rennie. “We were fortunate, in that we actually had a tool that we could demonstrate; they could instantly understand what we’re doing.” If it’s not possible to develop some kind of a working model, then slides presenting the user journey and mockups of the tool in action can give judges a clear picture of what you’re trying to build.
- Show your working
Outline how you’ve conducted your research and development work, says Johnson. “What are your partnerships? Have you done hackathons, or surveys, or did you reach out to clients?” The judges will want to see both that you’ve done proper consultation and contact-building work, and that it has produced evidence that your idea is likely to prove valuable, cost-effective and deliverable.
- Master the details
After your 10-minute pitch, the judges have 20 minutes to ask questions – so “you need to have a mastery of the subject matter”, says Rennie. “You need to know everything possible about what it is you’re pitching, because you’ll get a curveball – and if you try to skate over that answer, they’ll call you on it.”
The judges are always friendly and supportive, but they are also senior data leaders who know what it takes to make a project work in government – so make sure that you’ve identified the weak points in your idea, and devise a way to address each of them.
About Matt Ross
Matt is Global Government Forum's Contributing Editor, providing direction and support on topics, products and audience interests across GGF’s editorial, events and research operations. He has been a journalist and editor since 1995, beginning in motoring and travel journalism – and combining the two in a 30-month, 30-country 4x4 expedition funded by magazine photo-journalism. Between 2002 and 2008 he was Features Editor of Haymarket news magazine Regeneration & Renewal, covering urban regeneration, economic growth and community development; and from 2008 to 2014 he was the Editor of UK magazine and website Civil Service World, then Editorial Director for Public Sector – both at political publishing house Dods. He has also worked as Director of Communications at think tank the Institute for Government.
Related Posts
