Can We Trust Green Apps? Mapping out 14 Trustworthiness Indicators

Abstract

Green apps have emerged as ways users can engage with climate action, covering ventures that plant trees as users search for information (e.g., Ecosia) through to apps that facilitate behaviour change (e.g., United Nation’s AWorld). But how much can these apps be trusted to facilitate long-term engagement with climate action? Setting our research within the literature on trust, we combine expert interviews (n = 20) with the academic literature to outline 14 trustworthiness indicators. Each indicator provides a clear statement about what makes a green app more or less trustworthy. The indicators are grouped into six core categories: going beyond the app, meaningful collective action, designing the app, accessibility and inequality, data, and organisation. In doing so, our indicators speak to a range of research from multiple disciplines. At the same time, they provide a toolkit for users, practitioners, and academics to critically and productively engage with green apps.

1. Introduction

As of August 2021, Alipay’s Ant Forest had planted over 326 million trees in the arid regions of China. This tree-growing effort has been driven by the activities of 600 million users of Alipay, a popular third-party mobile payment platform in China. When users engage in green purchasing on the app, such as paying for public transport or certain products at the supermarket, they are rewarded with green energy points that can be accumulated and spent on planting trees or biodiversity projects [1]. Alipay’s approach to financing climate change projects through its app has gained worldwide attention, with Ant Forest winning the UN Champions of the Earth Award in 2019 [2].

Alipay’s Ant Forest is a leading example of green apps, defined as “mobile applications that aim to foster environmental sustainability” [3]. This broad definition casts the net wide, including different types of actions: apps can fund nature-based approaches to climate change by rewarding customers for purchasing environmentally friendly products and services (e.g., Alipay’s Ant Forest) or by using certain digital services (e.g., search engine Ecosia); some apps address market failures by allowing individuals or organisations to sell food that would have been wasted (e.g., Too Good To Go); and others focus on behaviour change by encouraging and advocating for people to act for the Sustainability Development Goals (e.g., the UN’s AWorld app) or financially rewarding people for environmentally friendly behaviours (e.g., BetterPoints). We consider any smartphone app where users engage in climate action a green app. Research from 2016 outlined 262 green apps on the Google Play store [3]. Given the emergence of green apps since 2016, we would expect the contemporary figure to be much higher.

Research examining green apps is often centred on specific themes, such as energy usage [4], broader behaviour change [5], and food waste [6]. Within each theme, research is generally set within specific disciplines—mostly Science and Technology Studies and Design. There is a distinct lack of cross- and inter-disciplinary research looking at green apps. The fruits of approaching green apps from these different disciplinary perspectives can be seen in work on Alipay’s Ant Forest, with work ranging from the effect of the app on pro-environmental behaviours [7,8], the motivations of those who continue to use the app [1,9,10,11], and the role of gamification [12] to the actual reforestation and rewilding efforts of the platform [13].

We need to extend this multi-disciplinary perspective to green apps. One paper may provide an excellent explanation of how users’ intrinsic motivation influences behaviour changes when using an app monitoring their energy usage, but what role does motivation play for apps where users are financially rewarded for engaging in active travel? Another paper could outline the merits of certain funding models for tree-growing organisations, but would this also work for carbon footprint apps? A recent paper from Tancredi et al. [14] attempted to answer some of these questions. They focused on mobile phone applications that promote environmental sustainability through pro-environmental behaviours. Rooted in work concerning behaviours and habits, they outlined a 10-point guideline for those designing these apps. Their guidelines are informative, especially concerning the role of nudges, gamification, and persuasive technologies. We look to build on this work by focusing on all green apps and looking beyond the design of the app, including issues of funding models, organisational structure, and integration into local services and infrastructure.

To provide this holistic, cross-disciplinary perspective, we focus on the concept of “trustworthiness” [15], asking how can we—as academics, professionals, and potential users—determine the trustworthiness of these green apps? We conducted 20 interviews with industry and academic experts in sustainability, technology, climate change, sociology, psychology, and geography. These discussions centred on how experts critically and productively engage with green apps, drawing on their specific expertise and experience. Using these discussions as our empirical basis, we then delved into the academic literature to contextualise the arguments and claims made by experts. This allowed us to identify the gaps in our current understanding of green apps, break down disciplinary silos, and locate the differences between our experts’ positions and established academic thought.

Our analysis outlines 14 trustworthiness indicators, grouped into six categories: going beyond the app, meaningful collective action, designing the app, accessibility and inequality, data, and organisation. Taken together, these allow academics, practitioners, and users to judge the trustworthiness of green apps.

2. Trust and Trustworthiness

To understand green apps, we focused on trust. The more a user trusts a technology, the more likely they are to use it—demonstrated in empirical work examining sharing economy platforms [16], chatbots for online shopping [17], smart homes [18], mobile banking apps [19], and dating apps [20]. Trust has not always been the focus of those looking at how people adopt and use technology. The work within the Technology Acceptance Model (TAM)—and the related, spin-off models, such as Sustainable Energy Technology Acceptance (SETA) and Responsible Technology Acceptance Model (RTAM)—traditionally focused on perceived usefulness and perceived ease of use. If the technology was perceived to be useful and easy to use, the adoption of the technology would be more likely. Trust was integrated within the TAM in the early 2000s in reaction to the enhanced levels of risk and uncertainty in web-based environments [21]. It is most often conceptualised as a relationship between the user and the technology provider [22].

Given that higher levels of trust are linked to higher levels of technology acceptance, attention turned to the factors underpinning trust itself. We can focus on the factors underpinning trust in apps: an app is more trusted when the user perceives the technology they are using to be secure (perceived technology security) [23]; there is a clear relationship between immediate benefits to the user—through it being useful or reliable—and the formation of trust in an app [16,24]. Other factors are external to the technology: the way people talk to each other about an app, including recommendations, affects trust [25,26]; when an app is an intermediary between a user and a set of companies providing products, such as the food waste app Too Good To Go, people’s trust in the company selling the goods is particularly important [27].

But trust formation is not static: people do not form their trust in technology at a single point in time [22]. People’s experience with technology is a key underpinning factor, with repeated usage increasing, decreasing, or maintaining trust [28]. For example, Oldeweme et al. [25] found that using a contact tracing app repeatedly reduced users’ perceived privacy and performance risks, meaning they trusted the app more.

Technology acceptance may outline what underpins people’s trust in technology, but it rarely asks a more fundamental question: is technology deserving of our trust? As Steedman, Kennedy, and Jones [15] explain, digital technology is often not trusted for good reason, due to discriminatory practices or poor handling of users’ data. In certain states in the USA where abortion has been criminalised, for example, women have reported deleting their period tracking apps due to fear of their data being used in a future prosecution against them [29]. And in 2019, the dating app Heyyo accidentally revealed 72,000 users’ data online [30]. Therefore, academics should shift “attention away from those doing the trusting to the trustworthiness” of the technology in question [15]. In the context of this paper, we must ask the following question: can we trust green apps to facilitate climate action?

Intervention studies provide evidence concerning the efficacy of such apps. It seems that there is a positive effect of smartphone apps on people’s attitudes toward saving energy [31], levels of energy consumption [4,32], and pro-environmental behaviours [5]. But these studies generally focus on the short term, often monitoring attitudes and behaviours immediately before and after the intervention period. When looking at game-based green apps, research has found no link between these types of green apps and long-term behaviour change [5,33]. More longitudinal studies are needed to determine whether green apps are used in the long term and whether they influence users’ opinions and behaviours.

When we consider green apps that facilitate the funding of nature-based approaches, certain apps report considerable impact: Ecosia, the search engine that funds the planting of trees through non-targeted advertisements to users, has planted 209 million trees; meanwhile Alipay’s Ant Forest had planted over 326 million trees as of August 2021. The claims of these enterprises have been rebutted by academics, however, who argue that planting trees is not the same as long-term tree-growing efforts that require funding, monitoring, stakeholder engagement, and considerable scientific expertise [34]. Other scholars are sceptical of any approach to tackling climate change involving people using their smartphones. For DeLuca [35], “smartphones are panmediation machines, the locus via which people mediate their worlds.” Being enmeshed within—and an active part of—the global surveillance structure means that using them to engage in climate action “would likely do little more than reinforce the world as it is” [35].

We do not outright reject green apps as media for effective climate action, nor wholly believe their promises. Instead, our paper outlines a set of trustworthiness indicators for green apps. These indicators are not quantitative ratings of individual apps (for an example see [36]). They provide a set of requirements– emphasising what is needed for an app to be considered trustworthy. The advantages of such an approach can be seen in other projects. The Trust Project [37] provides eight trust indicators to help the public know who and what is behind a news story. They frame each indicator around a series of questions, aimed at the reader. For example, Best Practices outlines four questions: Who funds the site? What is its mission? What standards and ethics guide the process of gathering news? What happens if a journalist has ties to the topic covered? These eight trust indicators are now presented on hundreds of news websites, from the BBC to the South China Morning Post. Our approach is similar: laying out a set of vantage points for academics, potential users, developers, climate activists, and practitioners to develop their trust with a green app from an informed and critical foundation.

3. Research Design

We conducted interviews with experts to explore the dimensions of trustworthiness in green apps (see similar approaches by [38,39,40]). We targeted two main areas of expertise: industry and academia. To better understand the climate actions presented to the user on the app, we aimed to speak to experts in Climate Change, Environment, Habitats, and Clean Energy. We focused on academic experts in policy and/or science who could provide technical knowledge of climate actions and/or set them within broader policy contexts, paying particular attention to those with knowledge of tree planting and behaviour changes (two key actions presented in green apps). To critically engage with green apps, we wanted to speak to experts in Use, Design, and Influences of Environmental-related Technology. We targeted academics from two disciplines: psychology to better understand how users’ motivations, intentions, and beliefs structure their use and influence of green apps; sociology to consider the social context within which people use green apps. In addition to these academic experts, we wanted to speak to people with professional expertise in Climate Change, Environment, Habitats, and Clean Energy and Use, Design, and Influences of Environmental-related Technology. Doing so was aimed at better understanding the context within which green apps were funded, designed, and rolled out to users, as well as the way climate actions played out within third sector and private sector organisations.

To recruit experts for our research, we took two approaches: we would either contact experts from our existing industry and academic networks or contact experts who authored key papers in one of our two areas of expertise. We recruited seven experts in Climate Change, Environment, Habitats, and Clean Energy and 13 experts in Use, Design, and Influences of Environmental-related Technology. The gender split of experts was roughly equal, with 9 female experts and 11 male experts (see

Table 1. Breakdown of experts’ characteristics.

Participant	Job Title	Sector	Academic (Ac) or Industry (In)	Male
Expert 1	Founder	Carbon	In	M
Expert 2	Researcher	Carbon	Ac	F
Expert 3	Researcher	Ecology	In	F
Expert 4	Engineer	Energy	In	F
Expert 5	Professor	Ecology and Policy	Ac	M
Expert 6	Consultant	Environment	In	F
Expert 7	Professor	Ecology	Ac	F
Expert 8	Professor	Sociology	Ac	F
Expert 9	Designer	Digital services	In	M
Expert 10	Designer	Digital services	In	M
Expert 11	Researcher	Psychology	Ac	M
Expert 12	Founder	Green tech	In	M
Expert 13	Project lead	Technology	In	M
Expert 14	Project lead	Green tech	In	M
Expert 15	Researcher	Green tech	Ac	F
Expert 16	Researcher	Green tech	Ac	F
Expert 17	Consultant	Behaviour	In	M
Expert 18	Founder	Green tech	In	M
Expert 19	Technology	Green tech	In	M
Expert 20	Researcher	Technology	Ac	F

). We conducted semi-structured interviews, which allowed for the conversation to address different aspects of green apps and trustworthiness and be specific to the skills and experiences of the experts. The interview was split into four sections. Each of the four sections will be outlined, with examples of questions that were asked: introduction to the research—a standardised section that outlined the scope of the research and the structure of the interview; expert background—a standardised section to understand how experts positioned themselves, e.g., “could you begin by introducing yourself and your work?” and “how long have you been doing work in this area?”; climate-related knowledge and expertise—a section specifically tailored to the expert, e.g., “you write about ‘social transformation process’, could you elaborate on this and why it is important for climate change?”; and green apps—a section asking the expert to relate their work to specific apps, e.g., “What scope do these platforms have to change behaviours?” or “Would you trust tree planting apps?” Here we used specific apps as starting points for discussions, including Ecosia, Treecard, and WWF My Footprint. During the discussion, new apps were introduced by experts, namely, SharingMi—an app that rewarded users for behaviour changes; Tred—a green finance app; and Too Good To Go—a food waste app.

The interviews lasted for 44 min on average (ranging between 30 and 59 min). All interviews were transcribed, pseudo-anonymised, and uploaded to QSR NVivo for analysis. Our analysis followed an emergent interpretive coding of the transcripts [41], guided by our focus on the trustworthiness of green apps. This process is not quantitative and therefore does not use inter-coder reliability. The robustness of codes was an emergent process that involved the project team taking turns coding the interviews in NVivo, with regular meetings to discuss the emergent codes. For a code to be included in the final analysis, all team members had to agree that it accurately reflected the qualitative data, linked directly to issues of trustworthiness, and was distinct enough from other existing codes. The project team used two metrics to guide the selection of codes for the Findings Section: the frequency of individual references to codes from the interviews and the frequency of experts referring to these codes (we set the lower threshold to three experts for a code to be included). We recruited one extra expert (Participant 20) after this analysis to address the lack of data concerning accessibility and inclusion. Only references to accessibility and inclusion are included in the Findings Section for this participant.

We report these codes in

Table 2. Findings table outlining code, definition, quote, frequency of references by participants and frequencies of references in extracts.

Code	Definition	Quote	#P	#E
Transparency of organisations’ data	Apps need to be transparent about their internal data, so the apps’ effectiveness can be assessed.	“Do they tell you what percentage of your donation goes to, you know, [do] they have annual reports that show that information? Do they clearly state that they have, [what] stakeholders are involved…and then are they actually monitoring whether the stakeholders get those benefits (…) Do they report data on their past projects? So interestingly like X, I happen to know the stats, I was reading it yesterday, only X% of organisations have tree-survival data on their past projects, and so there’s things like that.” (The actual number has been removed to protect the identity of the expert.) Expert 7 “So I think that kind of clear, accurate, transparent methods in the carbon accounting, alongside really well thought out restoration schemes is important, which is obviously a huge challenge.” Expert 2	16	45
Meaningful collective action	Apps need to go beyond individual action and provide people with a spectrum of political actions ranging from light touch to deeper collective engagement, connecting to existing policy and political discussions.	“[T]he danger of this [generalised appeals to the public to do something] is, it kind of puts too much emphasis on how much control we have as individuals” Expert 14 “Get some local people [to a local government planning meeting about wind farms] [through the app] to campaign.” Expert 11	10	27
Making data meaningful	Apps should make the data they share meaningful for the user by improving how it is communicated and educating the user.	“You can find all of that information, but you gotta scroll through a PDF to find it. It’s not in a digestible digital format. So that for us was pain point number 1, like, we need to find all of that information and then re-put it out to the world in a truly open source format, where academics, where small companies can hook into that data in a digital database format” Expert 1 “Either you can over simplify it or you can educate people to understand the complexities of it and we really lean into that educating people to understand.” Expert 19	10	21
Climate change specialist staff	Organisations that have more staff with expertise in climate action being facilitated will have more effective programmes with regard to climate impact	“[W]e actually did come up with an index and sort of looked at... what we found was, we did an analysis... one thing was the number of staff, the amount of scientific staff, their scientific qualification [was important to judge the quality of tree growing efforts].” Expert 7	9	20
Long-term behavioural mechanisms	Apps need to engage with long-term behaviour change mechanisms to be effective, because short-term mechanisms, such as financial incentives or gamification, are not effective at retaining long-term engagement.	“A key principle for long term engagement it means it needs to come from wi-, they need to feel like they own it, like they need to have ownership for it, accountability, responsibility, whatever -ility you want to attach to it, they need to feel invested in it, errm, gamification is good for that good quick short term fix or hit and great for pumping numbers but, as with many games, they suffer from long term engagement.” (Expert 18)	9	16
Easy to use	App providers need to ensure their app is easy to use compared to other similar apps—not just relying on users’ environment-related motivations.	“[E]very app that I ever see out there assumes it is the only app in your life. You know, it makes brilliant assumption that there aren’t 67 other things on your phone kind of pinging away at you, demanding your attention” Expert 9 “If they are adopting it and they’re not [for] super environmental issues, then like my question is: Is it pestering people too much? Isn’t turning them off so that then go back to the other app or not?” Expert 11	8	13
Not just an app	Apps are more effective at changing behaviours when they are part of a broader intervention that goes beyond just using the app.	“I’ll actually go as far as to argue, my opinion would be, unless it is tying into the infrastructure and feeding into the political kind of nature of those conversations, I would say it’s probably not going to work that well.” Expert 9 “They also need the infrastructure of kind of tapping into other services.” Expert 9	6	22
Connecting users to their locality	Apps should connect users to their local environment, infrastructure, and communities to facilitate meaningful engagement with the app.	“When I look at every tech solution now, the framing I use is ‘how can we make it local? How can we ensure that will get adopted locally?’” Expert 18	6	11
Stable, long-term funding	Apps are less effective at bringing about long-term change if funding is short-term or dependent on hockey stick growth.	“Because they are venture capital backed they tend to need to scale really, really quickly and just build a big tech platform without any kind of indicator as to who is actually serving.” Expert 18 “We had this great software, you know, and because it came with a grant (…) but then their grant ran out because, you know, it was it was two years later and that’s the way grants work.” Expert 13	5	10
Accessibility and equality	Apps need to be designed to be accessible for a diversity of users, intentionally addressing structural inequalities.	“The more ways you have to reach people, the better. So, if people could benefit from Ecosia not only by downloading the app, but by using a website or becoming a member (…) you can include other people, without necessarily expecting that they have a mobile phone that will support that app.” Expert 20 “I went to this wonderful conference called New Adventures in Web Design (...) one of the talks had this beautiful presentation from Lego, where it was all about, it was the video about their accessibility bricks where it gives you, like, spoken, spoken things on, it gives you basically accessible instructions on how to do stuff and. It had you suddenly realised that there’s this whole presentation about accessibility and there’s no accessibility in the video. There’s no subtitles, there’s no voice over (...) it’s like you’re pretending to do something for blind people, but not doing the voice over (laughs).” Expert 13 “So classes were competing against each other and actually it started to make its way into the curriculum (...) where you could then access subsidies to things like photovoltaic cells for your school roof. But again, you then start, it’s like it works to a point (...) You’ve like, just got like, all the things that then suddenly come with that in terms of like, social deprivation (...) And actually you’ve done is just go like, right. “Well, actually, the really posh school down the road won [(laughs)] and they get the cool stuff.” Expert 9	5	7
Net CO2e impact	Apps and their organisations need to account for the net environmental impact of their activities and actively reduce it.	“‘OK you’ve mined all these minerals, you’ve transported XYZ to create this device.’...I’m picturing a smart thermostat...the data centres running, all the data is collected and stored, and that how does that outweigh or outbalance the energy savings that could be caused” Expert 15	5	7
Asset-based framing	Apps should rely on positive framing that elicits hope and optimism rather than negative framing rooted in fear and despair.	“I don’t want it to seem like a compromise or a loss all the time, it’s more like, right ‘can we talk about the benefits of saving water and the benefits of saving electricity’” Expert 9	4	7
Transparency of user data	Apps need to be transparent about the user data they collect and how this data is used.	“And then we had one, one other topic because in the topic, ‘household energy’, yeah, household energy use and energy demand and so on. We did a lot of interviews (...) And we have just seen that people were so critical of them because they did. They couldn’t control what kind of data were actually collected and what, what was the real purpose behind that. You know, so they, they had a, a, a pretty high sense of, you know, scepticism. [Umm]. Always asking the question ‘no, but I know that this is promising something which sounds nice, but isn’t the real business case behind it something different and am I not giving away data that I actually don’t want to give away?’” Expert 8	4	4
Users are co-creators	Apps need to engage users as co-creators by actively engaging with them throughout the design and rollout process.	“So, again that’s the key difference, which is rather than what I call ‘developing behind closed doors’ or polish—get investment bias—and polish the first prototype to the nth degree, instead I would rather collaborate with communities from day dot.” Expert 18	3	4

, outlining the code name, the definition, key quote(s) that summarise the code, the total number of references to the code, and the total number of participants who referred to this code. For certain codes, we have provided nuance below the table.

4. Findings

In total, 14 app characteristics were identified in the interviews with experts. Experts discussed themes related to the app design, the types of actions incentivised, and the broader organisational structure, examples of which can be seen in the excerpts. Experts provided insights in line with their expertise, and therefore not all experts shared insights for all codes. On average, each code was referred to by seven experts (36.8%). The code with the fewest references involved three experts (15.8%) and the code with the highest number of references involved 16 experts (84.2%). The characteristics, their definitions, and example excerpts from the interview data are included in Table 2. The sub-sections that follow examine the characteristics that had intra-characteristic nuances.

The definition and quotations provided above describe the 14 individual codes. For certain codes, however, participants provided more in-depth justifications for why the code was important and outlined different proposed solutions. Below we outline the nuance of four codes: meaningful collective action, making data meaningful, long-term behavioural mechanisms, and stable, long-term funding.

5. Meaningful Collective Action (You Have to Understand Different Forms of Collective Action)

Experts were keen to emphasise that there was a spectrum of meaningful collective action, ranging from lower to higher levels of intensity or effort. Expert 16 outlines a working explanation of this spectrum:

It’s always just part of a [spectrum] from “recycling glass” right up to “marching outside of Parliament” [...] People do that a lot if they’re scrolling through Facebook. “Oh yeah, I sign this petition” but they’re not keen to often go out on a march. Expert 16.

This spectrum of meaningful collective action, however, can involve actions that become distractions for individuals. For example, Expert 5 argued that the push in the UK around reducing single-use plastics in the late 2010s detracted from other more impactful solutions, such as changing taxation systems.

5.1. Making Data Meaningful (Communicating Complexity Well)

Experts viewed simplification as misleading and outlined strategies to make data meaningful. In the rewilding and ecological space, Expert 19 argues that there is “too much push to really simplify things, to kind of give simple messages or hide a lot of the complexity”. In over-simplifying, Expert 19 argued that organisations can provide misleading information to the public. They referred to a recent analysis that showed if you added up “all the commitments by companies and countries in the world around their tree planting, there’s not enough space to plant all their trees.” This over-simplification of the solution to climate change was also highlighted by Expert 18, who argued that too many technology organisations put forward a “tech will save us” narrative.

Over-simplification was considered particularly harmful when it came to carbon schemes. Both Expert 1 and Expert 3 emphasised the delay between carbon being emitted in the present and a tree being planted to sequester the equivalent carbon in the future. Expert 7 lamented the lack of reference to the wide range of carbon sequestered in one hectare of woodland compared to another. Expert 5 and Expert 2 were highly suspicious of carbon credits generated via avoiding emissions—emphasising the over-crediting of these types of schemes. Experts also emphasised the lack of communication about the value of woodlands beyond carbon, including biodiversity (Expert 2 and Expert 5) and flooding (Expert 6).

For Expert 18 and Expert 19, it was more professional to lean into the complexity of the approach. To do so, experts outlined strategies to communicate this data. For trees, they explained that the data was relatable—most people know what a tree is (Expert 6 and Expert 14)—and it can be counted as discrete objects—“thousands of trees” (Expert 2). For other nature-based approaches or technology, more work is required to make data meaningful. When it comes to wetlands or peatland (Expert 2), for example, quantification of scale relies on measurement, such as hectares or cubic metres, and not counts. Communicating these types of nature-based approaches in a meaningful way can be harder. Some experts outlined how they provided equivalents to make this type of data meaningful to their audience:

Uh, making things sort of understandable by the public, we have to change our communication of sustainable urban drainage systems to a size of like a park. So, we had to change it to like you need 50 Hyde Parks by 2050. (Expert 6).

We put loads of energy into getting people to understand, like, what on earth a Kilowatt hour was. [mmm yeah] and like what that meant to them in real terms and kind of like sort of the leaving your hair dryer on or, like the oven on for X amount of time. And we just did quite a nice job of the comms around all of that. (Expert 9).

5.2. Long-Term Behavioural Mechanisms (Why Short Term Is Not Good)

Short-term behaviour mechanisms were not just considered ineffective but were also positioned as counterproductive. Gamification and financial incentives were key short-term mechanisms identified by experts. As Expert 18 explains, gamification is not an effective strategy: it is a “good short-term fix” of dopamine and is “great for pumping numbers” but “[gamification] suffers from long term engagement…because there are always other games, there is always somewhere else to get that initial dopamine rush.” (Not all references to green app games were critical: P15 referred to WildChain.io as a conservation game. They were descriptive, neither positive nor negative about it). Concurring with Expert 18, Expert 9 states that “[I]t’s so novel to download an app and go right, I’m going to cycle to the station and back each day...when it’s chucking it down with rain and it’s like February the 8th, like, yeah, I might drive actually that day.”

Expert 11 proposes that using financial incentives to encourage people to adopt an initial pro-environmental behaviour would make it less likely that there would be spillover pro-environmental behaviours. Financial incentives can also overcrowd other motivations. As Expert 17 posits, an external financial incentive to take part in an activity can be effective in the short term; however, once this incentive is removed—and the intrinsic motivation has been overridden—the person is less likely to keep engaging in that behaviour.

5.3. Stable, Long-Term Funding (Long-Term Funding and Financial Independence)

Experts were keen to outline solutions to short-term funding. Expert 14 discussed Ant Forest—emphasising that the app was “one of the most successful examples of how to engage citizens into tree planting.” This was largely credited to the close connection between AliPay and the Chinese state, who have a long-term commitment to “the green wall that they building in the Gobi desert.” Alternatively, Expert 19 pointed to their organisations’ membership-based funding model. This was a shift away from being funded by outside backers that could push them “in a particular direction” or having to grow at a high rate in the “short term”. It allowed the organisation to “focus on sustainable growth” instead. Adopting a monthly subscription model, where members would regularly donate money to rewilding projects, meant that the organisation could have a revenue stream with a “big impact” directed towards projects that the organisation and members want to fund. As Expert 19 emphasised, “you don’t need that many members to be putting like several million pounds plus a year into rewilding projects.”

6. Discussion

The previous section provided an overview of the different ways the experts critically engaged with the trustworthiness of green apps. This Discussion Section takes each code and examines how it corroborates, clashes with, or adds to existing academic knowledge. Through this process, we generated 14 well-evidenced trustworthiness indicators that provide clear demands of green apps, so people can critically assess their trustworthiness. These are grouped into six categories: going beyond the app, meaningful collective action, designing the app, accessibility and inequality, data, and organisation. Certain indicators could fit into multiple categories. In these cases, we have selected the most suitable category and provided additional information for our rationale (see #13, for example). These trustworthiness indicators may not apply to every green app.

6.1. Going Beyond the App

#1: Green apps need to link into services and infrastructure.

The experts argued that behaviour change interventions could not just rely on an app alone. The app needs to be part of a broader intervention that links into services and infrastructure. The SharingMi app—one of the apps discussed by Expert 9 and Expert 10—was not just an app. It rewarded users for behaviour changes in mobility, energy consumption, and community engagement. For example, users would be rewarded for hiring bikes to engage in active travel. An empirical analysis by Manca et al. [42] found that this approach—going beyond just providing an app—was successful in increasing active travel.

Work on carbon footprint calculators demonstrates the need for going beyond just providing an app. These footprint apps involve users completing a survey about their behaviours related to diet, transport, energy, and consumption—with this data underpinning the calculation of their annual CO2e (Carbon Dioxide Equivalent) footprint. Dreijerink and Paradies [43] argue that providing a carbon footprint app is not enough—these calculators should be integrated with broader interventions. They point to the effectiveness of Carbon Conversations—an approach where a volunteer coach guides a group of people through their impact and how to reduce it [43]—and the need for this non-app approach to be combined with carbon footprint apps.

#2: Green apps are more trustworthy when they actively build a community in the users’ locality.

For our experts, part of going beyond the app was connecting users to their local environment, infrastructure, and communities. The role of apps in general in fostering and developing local communities has been well documented, from app usage during COVID-19 [44] to personal messaging platforms [45]. Emerging work on green apps also emphasises locality in the design of apps [14]. Meshulam et al. [46] outlined how Olio—a type of green app that allows people to share food that would have been wasted with others—strengthened community ties through physical interactions during food collection. Experts argued that the more an app was linked to users’ location, the more likely they were to adopt it. There is little evidence in the literature of the importance of localism in user adoption when it comes to green apps, with some empirical evidence that community building has negligible long-term effects on retention [47].

More evidence outlines how locality influences people’s environmental-related behaviours and opinions. Localised and context-specific information on air pollution resonates with people [48]. In general, people perceive others’ biospheric values as low—this misperception is identified as one of the key barriers for pro-environmental behaviour adoption [49]. Building local communities can also foster pro-environmental social norms [50]. Social norms also play a role in fostering pro-environmental activism. Willis and Loy [50] argue that people are more likely to engage in environmental protests if they identify with others engaging in climate protection and perceive high levels of pro-environmental activism from peers.

6.2. Meaningful Collective Action

#3: Green apps should encourage participation in meaningful collective action at various levels of engagement and comfort.

Our experts emphasised the need for green apps to consider a spectrum of meaningful collective action. This chimes with the existing literature. Setting people only within the private (often consumerist) sphere is limiting, or, for some, counter-intuitive given the role of consumption in climate change [51,52]. As Steg [49] emphasises, humans can engage in consumption behaviours that actively reduce their impact on the planet, and this can also be complemented by “citizen organising”, such as protesting, voting for parties with strong climate policies, influencing organisations they are part of (e.g., workplaces), and supporting technology, policy, and system changes [53,54,55]. The effects of citizen organising are hard to measure, but there is country-level evidence that membership in international environmental NGOs leads to a moderate decoupling of carbon emissions and economic development [56].

Experts emphasised that it was important for this citizen organising—or public sphere action—to be set along a “continuum of political participation” [57], from signing petitions to attending local planning meetings. Understanding where the user sits along this continuum and providing lower-intensity activities for individuals to engage with climate change can provide a pathway to more and more involved forms of collective action (see [58]). This is especially important given the prevalence of self-reported low-intensity activities. In the UK, for example, 85% of people surveyed by The Department for Energy Security and Net Zero [59] reported that they recycled household waste.

6.3. Designing the App

#4: Green apps should be designed with good usability practices.

Users who are motivated to address climate change may decide to download a green app. But, as our experts were keen to stress, this intrinsic motivation alone will not keep them using the app in the long term. The green app needs to be well-designed and easy to use. For example, the search engine Ecosia may appeal to a user’s desire to address climate change, but can its core function—to search for information—compete with Google? Preliminary research suggests that it struggles to do so [60].

This links to a broader emphasis within the academic literature. As Vaghefi and Tulu outline in relation to health apps, continued use involves persistence of their health goals and users’ assessment of the app. Assessments involve judging interface design, navigation, notifications, and system rules [61]. This chimes with the technology acceptance research that emphasises the importance of ease of use when it comes to the use of technology [62] and the continued emphasis on the scales that assess usability [63].

#5: Green apps should focus on longer-term behavioural mechanisms.

Our experts consistently rejected short-term behavioural mechanisms, such as gamification and financial incentives, in favour of long-term approaches. Existing research largely supports this argument. Game-based smartphone apps do not have a proven link to long-term behaviour change [5,33], and the use of financial incentives to encourage people to adopt an initial pro-environmental behaviour would make it less likely that there would be spillover pro-environmental behaviours [64]. This is not the case for all contexts. The use of financial incentives for physical activity has been linked to long-term behaviour change [65,66], which can explain the continued emphasis on financial rewards for green apps [14].

In rejecting short-term gains involving gamification and financial incentives, experts pointed to longer-term behavioural mechanisms, including intrinsic motivation. As outlined above, biospheric values—the altruistic motivation to enhance the quality of nature and the environment—are closely linked to people engaging in pro-environmental behaviour [49]. For Tancredi et al. [14], gamification can be used in this context to strengthen existing intrinsic motivations. Often this motivation, however, can be overcrowded by financial incentives. An external financial incentive to take part in an activity can be effective in the short term. Once this incentive is removed—and the intrinsic motivation has been overridden—the person is less likely to keep engaging in that behaviour (see [67]). Other suggestions by experts are less well documented in the literature on climate change. There is some evidence that cultivating the “warm glow effect” for environmental action can lead to long-term behaviour change [68]. Whilst the evidence base for the identifiable victim effect is sparse, it has been applied to communication strategies around littering [69].

#6: Green apps should engage with both negative and positive framing that form part of a planned and intentional emotional flow of the user.

Part of these behavioural mechanisms involves communicating with the user. Experts emphasised that negative framing of climate change should be rejected in favour of positive frames. It has been shown that positive framing of messages on air pollution may “reassure and empower people that there are feasible actions to mitigate the effects of pollution” [48]. There is a large body of work, however, that emphasises the importance of negative frames in changing peoples’ opinions and behaviours.

Research documents that negative frames are often more effective than positive frames in improving green intentions and behaviours [70] and public engagement [71]. Climate anxiety, for example, positively predicts certain climate actions [72]. There does seem to be a place for negative framing in climate communication. Importantly, however, most of these studies often adopt a snapshot approach: exposing people to negative or positive framing and then measuring effects. It is more realistic, when considering green apps, that users are set within an “emotional flow” [73]. Nabi, Gustafson, and Jensen [74] found that those “exposed to the threat-based message [negative] reported greater hope after exposure to the solutions-oriented efficacy message [positive].”

We recommend that the design of green apps should be based on emotional flow. This involves appreciating how the user is exposed to different types of positive and negative messaging throughout their usage of a green app. Designers should consider using positive messaging followed by negative messaging—and attempt to capture the efficacy of such an emotional flow.

6.4. Accessibility and Inequality

#7: Green apps need to be inclusive by being widely accessible in ways that account for structural inequalities.

During our discussions with experts, they outlined how digital technology needs to be accessible to a wide diversity of users—and they provided examples of the way technology can exacerbate existing inequalities.

There has been considerable development in digital accessibility over the past decade, including the creation of accessibility standards, guidelines for assessing accessibility, and mobile platforms, such as iOS and Android, designing with physical and sensory accessibility in mind. But, as Bunyi et al. [75] argue, “accessibility in the digital world is still poor”, as demonstrated by the accessibility audit of Yan et al. [76] on mobile apps. In the world of digital health, co-designing or including those with lived experience in the development of an app is key. It can help identify and address specific accessibility issues for the intended audience—emphasising the need for inclusive language, personalisation of content and functionality, and customisation [77,78,79].

It is not enough for green apps to be accessible; they also need to address structural inequalities. People may have access to green apps but be excluded from actually using them due to a range of social, political, economic, and design issues [80].

As Zheng and Walsham [81] argue, we must understand how digital technology “is intertwined and implicated in producing and reproducing social orders and stratifications.” Bol et al. [82] found that those who used health apps tended to be younger, have a higher education, and be more tech-savvy than non-users—further marginalising those that need improved access to healthcare. Zhang et al. [83] outline a similar issue for smart city technology: smart mobility services are accessed less by women, middle- and old-aged people, Muslims, and manual workers—meaning these groups have less access to transport information. In not actively addressing inclusion, green apps will often reproduce systems of domination and oppression that exacerbate climate change.

6.5. Data

#8: Green apps should adhere to the ethical guidelines for user data privacy and clearly communicate how they do so.

Much of the academic literature around technology and sustainability often emphasise the importance of self-tracking data in changing behaviours [14]. When discussing data, our experts were keener to emphasise the importance of privacy and security of user data—an emphasis documented in the academic literature on smartphone apps [84,85]. These concerns over data privacy have resulted in a range of industry and government policy changes, including the EU’s General Data Protection Regulation (GDPR) and smartphone software’s “app permissions” that allow users more control over the data that smartphone apps can access [86]. For our experts, the ability to control smartphone apps’ access to phone data was not enough.

There was a call for transparency in the collection and use of data generated through green apps, with experts emphasising that the monetisation of data can often be hidden from the user. These hidden data processes are highlighted by Achilleos et al. [87] in their analysis of video conferencing apps, which documented the lack of transparency in what data is collected by the app, how permission is given for the data to be accessed, and how this data will be used by third parties. A lack of transparency can reduce users’ perception of data privacy and security, in turn leading to a lower chance they would accept technology that uses personal data [24,88].

This code was discussed less than others in our analysis. Our interview data did not provide a direct explanation for this. We assume that our experts see data security and privacy as important issues for smartphone apps, but do not consider them of specific relevance to green apps.

#9: Green apps need to be transparent about their internal organisations’ data by providing comprehensive and relevant impact data.

Experts consistently argued that green apps need to be transparent about their organisations’ data. Transparency has been linked to trust in the political sphere [89,90] and for companies [91]. Rahman and Nguyen-Viet [92] outline the importance of transparency for green brands to avoid accusations of greenwashing. They point to recent work showing that transparency positively influences consumers’ trust in green brands [93]. When it comes to mobile phone apps, Hovarth et al. [94] found that continued use of a contacting tracing app during COVID-19 was dependent on the transparency of evidence regarding the effectiveness of the app itself.

But, importantly, just providing information is not enough to develop trust. Tomlinson & Schnakenberg [95] outline how people place trust differently depending on how much information is shared with them and what is being revealed. Our experts distinguished between the types of data for the different climate actions offered by the app.

For nature-based approaches, such as tree planting, experts emphasised that data concerning the long-term viability of projects, stakeholder engagement, and monitoring needs to be presented. This reflects the contemporary academic work that evaluates tree-growing efforts [34]. For behaviour change interventions, experts emphasised the need for transparency in three areas: the need for data about long-term behaviour changes, especially important given the paucity of evidence for a link between app usage and long-term behaviour change [96,97,98]; the documentation of spillover pro-environmental behaviours, vital given that spillover behaviours are notoriously hard to cultivate [99,100]; and comparisons to other similar projects to judge the effectiveness of smartphone apps. As not all green apps produce this data, the emphasis on transparency can highlight where they need to improve their current data collection and analysis practices.

Too Good To Go (TGTG) offers an excellent case study on transparency. TGTG provides annual reports comprising comprehensive and relevant data about their activities, whilst also releasing technical reports that outline how metrics on the app (such as CO2e avoided) are calculated. All the reports are available to download from their website. We recommend other green apps follow this example.

#10: Green apps need to make their data meaningful for the user.

It is not enough for green apps to be transparent about their data. There needs to be a concerted effort to present information in a way that is actually meaningful to the user [101]. Fundamentally, people need to be able to easily digest and understand the data, referred to as “clarity” [102]. As our experts emphasised, there is a fundamental difference between releasing reams of datasets and presenting this data in a way that is comprehendible, useful, and meaningful to the user.

Carbon footprint calculators provide an important case study. Users will be asked to complete a short questionnaire so the app can estimate how much CO2e they produce annually. But 11 tonnes of CO2e per year—for example—could mean little to the user. A notable approach to make this data more meaningful is the “planet representation”, where people are shown the number of Earths that would be needed if everyone had the same consumption pattern as the user [103,104].

As these types of visualisations are key to making data meaningful, we argue that green apps adhere to the recommendations of Franconeri et al. [105], who provide a comprehensive explanation concerning what approaches “work”. Their nine-point summary can be used as a quick guide for those working within green apps.

6.6. Organisation

#11: The length of funding and the source of funding influence the trustworthiness of the green app.

There are a range of ways green apps can be funded, from large-scale venture capital backing (See Treecard) to funding provided by nation-states (see ActNow). Our experts were clear that apps were less effective at bringing about long-term change if funding was either short-term or predicated on hockey stick growth. For climate-related ventures, the problems of short-term funding have been documented. In their study of Swedish government funding for ecological restoration, Borgstrom et al. [106] found that short-term projects (1–2 years in length) were most common—working against the need for continuity. Short-term funding can also run counter to long-term, community-based behaviour change projects [107]. The length of funding, however, is only one factor influencing how funding affects the trustworthiness of an app. Expectations attached to the funding are another factor.

The demand for “hockey stick growth”—more often than not through “green venture capital” [108]—means that organisations prioritise scaling the app over the original purpose of addressing climate change. Whilst there is substantial literature on how to optimise green capital and green venture capitalism [109,110], there is less work aligning with experts’ position that this form of funding operates in opposition to an organisation providing a green app that actually tackles climate change. This can be somewhat captured in the arguments around “degrowth”, where there is an emphasis on reducing consumption and production to conserve natural resources and protect the planet [111].

#12: Green apps should provide a breakdown of their staffs’ expertise—and employ a greater number of specialists in the climate impact the app focuses on.

Some conversations with experts centred on green apps offering nature-based approaches, with experts emphasising the need for climate specialists being a considerable part of the staffing of green app organisations. A similar approach to understanding the staff within an organisation was taken by Schubert et al. [34]. They recorded the scientific expertise of staff by focusing on whether at least one staff member had either a bachelor’s or graduate degree (master’s or PhD) in a relevant scientific field, and whether they had a scientific advisory board with relevant expertise. Staff expertise predicted the best practices index. Looking beyond climate change to other sectors, the role of staff is mixed. Makudza et al. [112] found that workforce diversity was a significant predictor of employee productivity. However, Nyoman et al. [113] did not find a link between staff competence and employee performance. This literature urges caution when discussing the influence of staff on organisations’ practices.

#13: Green apps should be developed with intended users from the outset.

Experts emphasised the importance of involving users in the development process. This means embedding users within organisations from the outset rather than just bringing them in during the design stage (it is for this reason that we place it here rather than the designing the app category). Community involvement has been argued as a must for effective climate action [114]. The associated shift in power might prevent biased or hidden agendas, such as fossil fuel companies centring the conversation on carbon impact calculators [115].

#14: Green apps should report their environmental impact, outline measures to reduce it, and contextualise their impact within the broader system.

The digital technology sector is the only sector with rapidly growing environmental impacts, accounting for 3 to 4% of world CO2 emissions [116]. Organisations should abide by good practices for sustainable digital technology, notably through a sufficiency approach, data governance, and longevity in design [117].

7. Conclusions

This paper has put forward 14 trustworthiness indicators that critically and productively engage with green apps. First and foremost, these indicators have academic significance. They emerge from a dialectic between our empirical data (discussions with 20 academic and industry experts) and the contemporary academic literature. Taken together, they provide a holistic, multi-disciplinary approach to green apps by drawing on a range of academic thought, empirical research, and professional experience. We can observe the fruits of this labour most clearly in the six over-arching categories under which the indicators are organised: going beyond the app, meaningful collective action, designing the app, accessibility and inequality, data, and organisation. Whilst some of our indicators overlap with the work of Tancredi et al. [14], they also outline new considerations, such as organisational issues and transparency of impact data.

The relationship between the academic literature and the testimony of experts was not simple, with alignment in some areas and discord in others, as outlined in the discussion of each indicator. For example, #11: The length of funding and the source of funding influence the trustworthiness of the green app was built through the corroboration of experts’ emphasis on the problems of short-term growth predicated on hockey stick growth with the academic literature. And in the case of #6: Green apps should engage with both negative and positive framing that form part of a planned and intentional emotional flow of the user, this latter indicator was formed by thinking through the disagreement between experts and documented academic thoughts. Experts wanted a reduced reliance on messages of doom and hopelessness, whereas the academic literature emphasised the efficacy of negative framing. Here we used the concept of emotional flow to bring experts and the academic literature together.

Beyond academia, we consider these 14 indicators as an important resource for two groups of people. Given the proliferation of green apps on app stores, we argue that potential or existing users need to be better equipped to critically engage with the apps they come across. A user may find Ecosia—an app that plants trees in exchange for using their search engine—personally satisfying but be sceptical of the specific claims they are making. By thinking through #9: Green apps need to be transparent about their internal organisations’ data, by providing comprehensive and relevant impact data, they can be equipped to think through what data could be shared to back up the claims they come across on the app. This may result in them trusting the app more because there is extensive and relevant data or trusting the app less because they cannot access the data underpinning the claims being made.

Similarly, those working in the green app industry can use the 14 indicators to interrogate their own work. When we think about the category of meaningful collective action, their app might be excellent at building a community in a users’ locality (#2) but only provide the user with very low-intensity collective actions. Engaging with #3 thoughtfully would allow them to think through the different levels of collective action that could be facilitated—and carefully consider how individual users may sit along this spectrum.

Limitations and Future Work

Our sample of experts was relatively small and did not evenly cover the different aspects of green apps. This manifested in the amount of data underpinning each indicator: transparency of organisations’ data was covered by a lot of experts, whereas users as co-creators was discussed by relatively few. Future research should build on findings from this paper (and Tancredi et al. [14]) to explore remaining areas of interest for understanding the design and rollout of truly impactful and trustworthy green apps, alongside practical investigations of existing green apps. This future body of work could result in applicable frameworks for designing and operating impactful green apps. Such an endeavour is important when we consider the ever-expanding industry of green apps.