Circular Design to Scale Lasting Change

Stephen Labanyi at UXDX Community: Rapid Product, UX and Agile and Circular Design. Video: https://youtu.be/ksp9EN1Y3Dg

Readable transcript: edited from the recording's captions for readability (fillers and false starts removed, punctuation and section headings added). Wording is the speaker's own. Timestamps are positions in the video. Names marked [?] could not be verified against the audio.

Introducing ReefCloud

[00:00:00] Imagine if we could bring the best of emerging technology and human-centered design to drive collective action that can help to protect the future of one of our planet's greatest natural resources.

[00:00:17] Hi everyone. [?] I'm at The Dock, Accenture's global center of innovation based in Dublin, and today I'm going to speak about ReefCloud. It's a collaborative intelligent platform to improve coral reef monitoring and preservation efforts on a global scale. There are three key things that I'd like to cover. Firstly, it's the outcomes from this purpose-driven project. Secondly, the circular design process that we used, and thirdly, how we amplified the work to create impact. To begin, I'll play a short video.

[00:00:55] A picture says a thousand words, and it can tell us about our coral reefs. More than half a billion people depend on reefs for food, income, and protection. They're also critical to the survival of a quarter of the world's fish. Around the world, we analyze pictures to inform the way we manage and protect our reefs.

[00:01:18] But climate change is accelerating the impact of many of the threats our reefs face faster than we can monitor them. To bring together coral reef monitoring efforts from around the world, leading scientists in Australia and the Pacific worked together to develop ReefCloud. ReefCloud harnesses the power of human collaboration and artificial intelligence to equip us with the latest and best information about the state of our reefs.

[00:01:42] Building on automated data management, artificial intelligence, and statistical analyses, ReefCloud can estimate reef composition 700 times faster, meaning we have the latest information within days rather than months or years. ReefCloud integrates information from many users and provides an easy go-to platform for the latest insights. Our reefs are facing their greatest challenge. By working together, we can help give them a fighting chance. Find out more at reefcloud.ai.

How the project and the team came together

[00:02:26] Accenture first started working with the Australian Institute of Marine Science, or AIMS, in 2019, and the idea originated at a hackathon, which led to a six-week rapid innovation sprint at The Dock, during which time we discovered that AIMS was exploring a similar concept. AIMS had thousands of images from which they were struggling to derive benefit, and the findings from this phase were very promising. We continued to expand our work with AIMS in 2020, conducting systems research, creating new AI models, defining a value narrative and partner strategy, as well as conducting a series of rapid experiments. Then we transitioned the project to other parts of Accenture to scale ReefCloud.

[00:03:10] At The Dock we understood that this is a challenge which requires collective action, and we also found that the purpose-driven mission at the core was key in helping us to expand the team and grow a support network. While we were working within a constrained budget, many people offered to help as a plus-one activity, as they were excited by the opportunity. Our previous speaker, Jennifer, described really eloquently growing a culture of psychological safety and working nimbly to develop high-performing teams. These ways of working were core to the success of the ReefCloud project.

[00:03:46] We joined forces with other parts of Accenture: Applied Intelligence, the Perth innovation hub, Fjord Sydney, the advertising agency The Monkeys, and Accenture Development Partnerships. Their teams brought capabilities across data and analytics, design, strategy, and innovation. We also grew connections with non-governmental organizations working in the ocean space, for example, the World Wildlife Federation, the Ocean Conservancy, and the World Resources Institute. We then started engaging technology partners such as Amazon AWS and Mapbox.

Challenges along the journey

[00:04:21] But we faced a number of key challenges along the journey that stretched the team, and it forced us to find creative solutions. These include cultural change. There's a need for a mindset shift around purpose-driven initiatives to help move leaders in this space from intentions to action, and this is so important to increase the pace of progress. An initial mistake that we made was assuming that senior leadership would buy in and support. In reality, this took significant time, to help educate stakeholders and communicate how a broader set of value levers can deliver business and societal value.

[00:05:00] Secondly, around funding access. Initial difficulties in securing funding led us to explore non-traditional challenges, for example awards and bursaries, to prove enough value to unlock further funding. Thirdly, around remote co-creation. The project was run during COVID lockdowns and we didn't get to meet the client face to face, so we used custom-designed working sessions to align and co-create frequently.

[00:05:26] The fourth key challenge was around time zones. We had a client base in Australia, and we distributed team members across Ireland and the US. On that basis we took it in turns on what we described as the graveyard shift, so that no one location was always inconvenienced. We spread the load among the team.

A circular design approach

[00:05:52] Then in terms of the approach that we took to solving these challenges. The nature of this specific challenge, and other challenges that clients are bringing to The Dock, are becoming increasingly complex, and as innovators we need to evolve our skill sets to help clients navigate complexity and drive lasting, scalable change.

[00:06:13] Recently a European Commission report estimated that 80% of a product or service's environmental impact is determined during the design stage. I'm really passionate about the role and the opportunity for design in creating a brighter future. For this project, I adapted a framework created by the Ellen MacArthur Foundation and identified key methods to help to explore, define, develop, and deliver from a circular perspective. I'll explain how this process was key to the success of ReefCloud.

Explore: systems thinking research

[00:06:44] The first stage, explore, focused on defining an agile systems thinking research approach, and the starting point was to gain understanding of the interdependencies in the marine ecosystem, to help maximize the impact of ReefCloud and to focus efforts for coral reef preservation. We took a collaborative approach to foster engagement, designing interactive working sessions with marine biologists and divers.

[00:07:07] We then conducted a series of experiments to help synthesize. To gain understanding of the elements that impact coral health and how they're connected, we created a model known as a concept map. This helped us identify what we knew, what we'd like to learn, and how to articulate AIMS's vision. In this map, the elements are portrayed by the circular nodes, the connections are portrayed by the lines, the current state is represented by the green nodes, and the intended future state of ReefCloud is represented by the purple nodes.

[00:07:41] We created an interactive version of this map to allow us to explore clusters of nodes and analyze the connections. It was really valuable, as it helped us identify central themes for the research, for example environmental change, collecting data, describing data, synthesizing data, and influencing. We investigated these themes further in later experiments.

[00:08:05] Then, based on the understanding of the connections between the elements and some core themes, we next investigated cause and effect by creating a causal loop diagram. Our research on the current state of the marine ecosystem identified a reinforcing cycle of social and environmental change, with over 500 million livelihoods at risk as a result of coral bleaching. That is leading to action, for example over-tourism, pollution, urbanization, that increases coral reef degradation, where 75% of the world's reefs are critically endangered. That in turn accelerates change in the marine ecosystem, where 25% of marine life is in danger.

[00:08:48] Current attempts to rebalance the system create a second reinforcing cycle, where the rate of coral reef degradation leads to pressure on the speed of data collection, which is slow and inefficient. Divers collect vast amounts of images for monitoring activities, but they have no efficient way to tag or upload the images. Marine biologists have to manually annotate hundreds of pixels per image, and they struggle to derive insights or trends due to lack of connected data sources. They then manually create reports based on scarce insights, restricting the ability to generate awareness or influence decision makers, and these factors indirectly contribute to overall coral degradation.

[00:09:33] ReefCloud will attempt to balance this reinforcing cycle that's detrimental to the health of marine life. It introduces foundational aspects such as standardized data organization and enhanced annotation to enable faster data collection and generation of richer insights and trends. Based on this, machine learning models can be trained to predict outcomes, and this ability can help to influence economic and environmental policy, inform action, and rebalance the rate of societal and environmental change. Key opportunities for Accenture were identified around data organization and labeling, faster, more accurate annotation, generation of trends, and the ability to predict outcomes.

Define: vision, value narrative and partners

[00:10:23] In the define stage, we leveraged our research findings to define the North Star vision, value narrative, and partner engagement approaches. We also focused on auditing the user experience, prioritizing future state features, and co-creating concepts.

[00:10:42] How we shaped our understanding around cause and effect helped us to define the future vision for ReefCloud. The vision aims to outpace climate change in two ways. Firstly, to support the global integration of monitoring efforts by defining best practice methods, growing awareness and collecting a really robust data set. Secondly, to influence policy by creating that single source of truth, democratizing data access and using modeling to predict future outcomes.

[00:11:18] We also created a narrative that considered the value of reefs from environmental, societal, and economic perspectives. Coral reefs are a vital part of the Earth's ecosystem. The death of coral reefs represents a huge loss, as much as $375 billion annually for the local economies that they support globally, and the loss of habitat for a vast number of creatures that depend on the reefs. But the economic value of healthy coral reefs is staggering. A recent World Wildlife Federation report estimated potential net benefit streams of 30 billion US dollars per year if reefs were well-managed and intact.

[00:12:00] To address these challenges and to take advantage of these opportunities, building strategic partnerships is vital. We identified more than 50 partners from a market scan and we grouped them into partner archetypes. We identified seven metrics to rank the strength of synergies and engagement potential. We also conducted portfolio deep dives into 17 high priority partners and 17 medium priority partners, and created an outreach plan for pre and post product launch. The initial partner engagement strategy focused on funding and technology partners to enhance the pilot, to add additional detection capabilities, provide offline and mobile functionality, and to integrate external data sets.

[00:12:47] We also synthesized overall research findings to surface key pain points and root causes. We then co-created with the AIMS team on future state features to address root causes and augment the user experience, and we asked the AIMS team to rank the features based on perceived value. A key learning here was that the most desirable features broadly covered a flexible annotation process; a method for identifying algae, which was described as the holy grail by the client due to the difficulty of distinguishing algae from coral; and then the ability to intuitively analyze and report on data insights.

Experiments and concepts

[00:13:26] We ran a series of experiments based on these prioritized features, and we focused on developing that partner engagement strategy to secure the long-term viability of this initiative. We focused on testing high priority opportunities for desirability and feasibility, creating low-fidelity mock-ups to iterate concepts at pace, and also in-depth scenario tests to bring the concepts to life.

[00:13:54] This included a UI concept to bring the features to life, and we firstly focused on analysis and reporting features. We conducted scenario tests and surveys on low to medium fidelity designs, and we found this to be a highly effective way of rapidly iterating concepts.

[00:14:25] The design and analytics teams also co-created a future state annotation experience that was designed to ease the burden on marine experts. This approach improves the speed, accuracy, and efficiency of the annotation process, reducing cognitive load with guided workflows. The key features here include flexibility to select manual or automated annotation; the option to set confidence thresholds based on the extent to which the data set has been trained by AI models; the ability to quickly and easily paint similar segments rather than manually annotating hundreds of pixels per image; and also contextual prompts on coral species based on machine learning.

[00:15:07] This new flow enables tasks to be completed in just a few clicks. A key learning here is that the combination of automatic methods and human insight offers a smoother annotation process with a more thorough result.

[00:15:23] We also ran a series of technology-focused experiments: for example, image segmentation to define coral health; the ability to distinguish algae from dead coral; optimizing the cloud architecture to reduce overall cost, which was previously their biggest monthly outgoing; and then enhancing the infrastructure pipeline to improve security.

Develop: testing with users and the brand experience

[00:15:46] The work led by The Dock in the explore and define stages was then transitioned to Accenture Applied Intelligence for the develop phase. This phase focused on advancing concepts with rapid user testing and iteration, refining the brand experience, creating high-fidelity designs, and supporting UI specifications.

[00:16:10] During the beta phase, we rapidly tested and iterated designs with over 200 users across 24 countries. This includes scientists and managers from Australia, Fiji, Palau, Vanuatu and the Maldives, who've been closely collaborating with NGOs, universities and government organizations to ensure that individual needs are considered: from local stakeholders, for example indigenous communities, marine parks, state managers, to national and international bodies, for example national ministries and international science advisory networks.

[00:16:47] In terms of the look and feel and the personality of the product, a brand style was developed using color to emphasize key information and KPIs on coral health and change over time, for example, and highlighting controls that allow users to quickly navigate to different spatial levels or save and export selections. ReefCloud also includes subtle animations and parallax scrolling to create a rich user experience.

Scale: launch, what's next and amplification

[00:17:18] Finally, the scale stage focused on launching ReefCloud publicly, gathering client feedback, and amplifying the work to maximize impact and gain wider support. ReefCloud will launch publicly at the Our Ocean Conference in Palau in April 2022.

[00:17:43] What's next? From a circularity perspective, we've identified three key dimensions to focus on to ensure the long-term viability of ReefCloud. Firstly, it's about strengthening the core platform, including UX enhancements, mobile functionality, data interoperability, and additional applications of machine learning, for example new species and new diseases in other geographical locations.

[00:18:09] Horizon 2 is looking at building tangential applications, for example application of AI algorithms in coral nurseries, augmenting computer vision to identify other indicators of ecosystem vitality, and integration with public fishery and monitoring systems. The third horizon is around platform monetization. This includes integration with carbon market mechanisms, custom-made tourism applications, for example diving or hotel experiences, and then utility for supply chain ESG monitoring.

[00:18:47] Our client and thought partner on this initiative, Dr. Manuel Gonzalez Rivero, has described the work as the biggest breakthrough he's had to date. We finally got to meet him. After a significant time of working remotely, on many working sessions and conference calls, he visited The Dock in July of last year, and I think it really helped to further deepen the relationship between Accenture and AIMS. We're working with him on the future vision with this now live platform: how can we ensure the long-term viability? How can we look at sharing best practices? How can we continue to grow that partner ecosystem?

[00:19:44] On this work, it's not just the scientific community that's taken notice. An Accenture global media release and coverage from 15 international media outlets has raised the profile internally and externally. Additionally, industry experts such as Richard Fevers [?] and Vincent Kenneil [?] have become strong advocates for the program. If I was to call out any one factor that was key to the success of the initiative, it would be the approach that we took to amplification: growing awareness, getting buy-in, proving value. It had a ripple effect. It really started to generate momentum for the project, and those are key learnings that I will take to future wicked challenges going forward.

[00:20:33] Beyond the scientific community, ReefCloud has won sustainability awards at the Asia-Pacific Spatial Excellence Awards and the Irish Design Institute awards. It also won an award at the Accenture Eco Innovation Challenge in 2021, which helped to generate funding to scale the platform and provide access to mentoring support from Accenture subject matter experts. More recently, ReefCloud was recognized at the Fast Company World Changing Ideas awards in 2022.

[00:21:04] Coral reefs are home to one quarter of the planet's marine life and are facing their greatest challenge, and with ReefCloud, we aim to give them a fighting chance. I look forward to evolving the circular design principles I learned from this project and applying them to address other wicked societal challenges. Thank you so much. With that, I'll open to any questions.

Q&A

[00:21:28] Host: Great. Thank you, Stephen. That was really interesting. It's one of those problem spaces that I never even knew existed until I heard you talking about it, but it makes sense, and it's good to see that some progress is being made in that area. If anybody does have any questions, please do write them in. Otherwise I'll get the ball rolling. One thing: when you were going through your phases, research, define, deliver, you were showing in the define phase the demo and things like that. What is your distinction between define and deliver? Two things jumped into my head. The first was that it was all just a prototype and smoke and mirrors, or you're actually building through the define phase.

[00:22:25] Stephen: It's a super question. When we work on projects, we commonly define short, sharp phases to prove enough value: scope the work, experiment, test, prove value, refine. In the define stage that you were speaking to there, we were building on our research understanding. We were looking to take our hypotheses, test them, bring some of the key opportunity areas or features to life, and at that early stage test with different groups of users. Whether it was divers, whether it was policy makers or external stakeholders, bring them to that wider group of stakeholders, get the reactions, refine.

[00:23:19] Those early stage scenario tests would have been created by a combination of us co-creating with the client, with design teams, analytics teams, software teams working really closely together to help bring those features to life. Because what we're testing there for is desirability, feasibility, and viability, and also, in this sustainability space, what we might call integrity or responsibility as well. Those were four key vectors that we included as part of our process for testing. We were looking to make sure that we had just enough robustness as part of our approach to make sure we could stand over the work, bring it to those different groups, test, get the reactions, refine, and move on.

[00:24:10] To your question there in terms of the level to which it was smoke and mirrors or fully developed, the way I would frame it is we started lo-fi, we rapidly iterated and tested, and as we gathered feedback we gradually iterated that to medium fidelity, to building out scenario tests. Then once we validated with wider groups, that's where we started to create those full flows, to start full development around it. Because there's a huge amount of research from a technical perspective that went into this, for example in terms of looking at mapping at different spatial levels, how to navigate users through that, and how to then develop custom reports off the back of that. For me, I think it was the level of close collaboration across the team that really helped us to move at pace.

[00:25:05] Host: So it's almost a blurring. There isn't a hard and fast define-deliver boundary.

[00:25:14] Stephen: I think that's a really fair point, in that the delivery is in the defining; the delivery is incremental. You may be looking, for example, to prove value based on a specific hypothesis. Can we distinguish between algae and dead coral was one focus area for us. We were looking at that from a technology perspective, and looking at it from a user perspective. What are the jobs to be done here? What are the key barriers in this space? Where are we looking to create additional value? Mapping out flows based on that. Then looking to make that real in a way that we could walk users through an as-is and a to-be flow, but always with that longer term view of how we would then scale this. So to your question there, it almost merges a little bit, or it's blurry: we're very much in the define stage, but with a view to the deliver stage.

[00:26:20] Host: Great. Well, thanks for that. There's one question came in, or actually we're getting flooded with questions. We won't have time to answer them all, but as I said, we'll put those questions on the Slack channel and I'll invite Stephen to answer them there. Let me just read a couple of them. I might jump in with the team size question. Alana has asked specifically how many designers were involved, but can you just elaborate on the full team who were involved?

[00:26:50] Stephen: Of course. Of all the projects I've worked on recently, the breadth of the skills that we had on this project was really comprehensive. We had a system designer and an interaction designer. I'm from a design background, but worked as a project lead on this particular initiative. We then had a commercial expert who we worked with, and a strategy expert who helped us to build the partner engagement approach. The commercial expert helped us to make connections with some of the potential partners, in terms of what's the value offering for them and what's the right kind of transaction there.

[00:27:46] We also worked with at least three analytics experts on this project, who looked at the image segmentation: that emerging technology, how best to make use of it, and how that can add value not just on its own but when you've got the AI and the human working together, how that can deliver the optimum experience. As well as that, we had a software expert looking at the technical feasibility of delivering on this, if we're going to create a product with all of those specifications that are required.

[00:28:25] That was our core team starting out, but then as we grew, we saw the need. We were working on this project for a given length of time, but we also needed to have that view three years into the future, five years into the future. Where do they need to be? Who's going to be working with them to help them continue progress? That's why we transitioned the project at a certain stage. We brought in Accenture Development Partnerships, who are experts at working on that longer term strategy, and also Accenture Applied Intelligence, who, once we had brought the product to having all of the user journeys defined and the concepts mapped out, took it to that next stage of full scaling.

[00:29:16] From there the team developed. There was a larger software team, there was a branding team from Fjord Sydney that got involved, and then, as I called out in terms of the awareness piece, that's where the advertising agency The Monkeys got involved. A broad range of skills, but for me, how closely we were knitted together was key to the success of the project.

[00:29:40] Host: Great. It is nice to be able to lean on that many different specialities.

[00:29:47] Stephen: Absolutely. It was really rewarding, and for me a key learning from that was, as designers, if we can stretch to have that level of understanding, or at least appreciation, for other people's roles, what they do and the value that can bring, I think that's where the magic can happen. That's where the real sparks can come, when you're co-creating on opportunities.

[00:30:14] Host: Excellent. As I said, we do have some other questions, so I will post them into the Slack channel after this. But thank you very much, Stephen. Honestly, people were interested, because questions started flooding in there at the end. Thank you very much for sharing your story.

[00:30:30] Stephen: Great. Thank you so much.