A mixed-method approach to study the impacts of energy micro-generation combined with appliance-level feedback on everyday practices

In alignment with EU climate action plans, member states have adopted national strategies to reach the target of net zero emissions by 2050. In Ireland, the Climate Action Plan 2023 (Rialtas na hÉireann, 2023) sets the target of 80% renewables in energy assets by 2030,^{Footnote 1} and a 40% reduction in residential sector emissions compared to 2018. The latter requires active participation from households. The incorporation of renewable sources in the energy system is perceived as essential, e.g. to decrease costs for households, promote the use of self-produced energy and reduce the volume of consumption in the grid, especially at peak times.

Energy is an invisible, abstract and untouchable force (Brandon & Lewis, 1999; Burgess & Nye, 2008; Fischer, 2008; Froehlich, 2009; Hargreaves et al., 2010; Nilsson et al., 2018), making it difficult for people to change their energy-consuming practices. For this reason, many studies and policy strategies have investigated the potential of energy feedback as a lever to reduce energy consumption, by making energy visible to users (Abrahamse et al., 2007; Darby, 2001; Fischer, 2008; Grønhøj & Thøgersen, 2011; Zangheri et al., 2019). However, other relevant literature questions the implicit assumption of households as rational agents, with the capacity (or will) to change their everyday habits when provided with detailed energy feedback (Buchanan et al., 2015; Nilsson et al., 2018; Shove, 2010; Strengers, 2013). These studies are based mainly on a practice theoretical approach, viewing energy as a means to perform a range of different practices constituting people’s daily lives. These often normalised practices can be difficult to change, as they are the result of interconnected elements, e.g. materialities and meanings that support taken-for-granted habits. In this view, the provision of information in a standardised way is not enough to change stabilised practices (Hargreaves et al., 2010; Strengers, 2013). In addition, other studies show how having some form of micro-generation, like rooftop solar PVs, is more likely to facilitate practice changes and time-shifting of daily activities by drawing people closer to the energy they use (Buchanan et al., 2015; Ellsworth-Krebs & Reid, 2016; Gram-Hanssen et al., 2020; Keirstead, 2007; Nilsson et al., 2018; Shove, 2010; Strengers, 2013).

In this paper, we will use the results of fieldwork conducted on two islands located off the west coast of Ireland over a period of four months, in autumn 2022 and spring 2023. The main research question is: How does the combination of PVs for electricity micro-generation and appliance-level feedback affect the performance of everyday practices? To answer this, we will present the results from eight households that have PVs and have received appliance-level feedback through a webapp displaying information from smart plugs connected to different home appliances. We apply a mixed-method approach, analysing results from interviews conducted with the households and energy data from the smart plugs. We will consider people’s experiences at two points in time: the first dates back to autumn 2022, when a first round of interviews and house tours with the households was carried out. During that time and after this first interview, smart plugs were installed and feedback was delivered. The second refers to spring 2023, when a second round of interviews took place, centred on the feedback provided. This enabled the researchers to study people’s everyday life, practices, usage and understandings of solar PVs before and after the provision of appliance-level feedback. This study aims to be explorative, looking at the changes people have brought to their practices when using the feedback provided in combination with the existing PV system. The webapp did not provide any practical suggestion, as its aim was to be informative, looking at how people responded to the information received.

We apply a practice-theoretical perspective, taking social practices as the main unit of analysis. Since the practice turn in social theory in the early 2000s, practice theories have been widely used to study consumption. Practices are defined as routinised behaviours that constitute daily life, and as collectively shared entities made up of elements like materials, competences and meanings (Shove et al., 2012). In the energy context, materials refer to the objects, technical devices and appliances used for energy consumption, production and monitoring; competences refer to the understandings and knowledge people have about these devices and in general about energy use; meanings refer to the personal interests and beliefs associated with energy use, and that make people become involved in certain practices more than others – e.g. whether they value environmental concerns, energy costs, energy security, etc. Using this practice lens, behaviours are defined as the “individual performances of social practices” (Spurling et al., 2013), so studying energy consumption means focusing on those everyday practices that consume energy, such as cooking, doing laundry, washing dishes, etc. Our study focuses on households’ energy-consuming practices and how they are affected first by having PVs, and then by adding appliance-level feedback. Focusing on individual households will be central in utilising this approach as it will show the situatedness of practices. Practices, in fact, vary among households and are performed differently, and this explains how homes with very similar characteristics and technologies present different consumption patterns (Gram-Hanssen, 2010), and respond differently to similar systems and to the same kind of feedback, as our case study shows.

Conducting this study on islands provided unique insights due to the singular socio-geographical setting and related challenges, e.g. in the provision of material from the mainland and in the instability of energy infrastructures (Tellarini & Gram-Hanssen, 2024). Furthermore, the selection of a tightly knit community enabled a deep engagement with the households involved and the field site. The focus of this paper will be mainly on the small-scale effects of the technologies installed on everyday practices of households with PVs. Without the pretension of making our results generalisable, the focus on practices can make the findings usable also for other settings.

The paper will be structured as follows: Sect. "Theoretical background" provides an overview of the literature about the impact of energy feedback and micro-generation on practice changes; Sect. "Methods" describes our case study and the methods employed for the data collection and analysis; Sect. "Empirical data analysis" presents our analysis of the qualitative and quantitative data; Sect. "Discussion" discusses the findings and, finally, Sect. "Conclusion" draws the main conclusions.

How does energy feedback influence households’ everyday practices?

Energy in general, and electricity in particular, differs from most other consumable goods by being invisible, abstract and untouchable (Brandon & Lewis, 1999; Burgess & Nye, 2008; Fischer, 2008; Froehlich, 2009; Hargreaves et al., 2010; Nilsson et al., 2018), and by operating at the “level of the sub-conscious within the home” (Dobbyn & Thomas, 2005: 6). According to Burgess & Nye (2008), energy is doubly invisible: firstly, for how it is silently and almost invisibly delivered to the home, through wires and infrastructures; secondly, because energy serves to perform daily routines that are not easily related to the energy that serves to perform them. This double invisibility complicates attempts to promote energy savings and changes in behaviours and practices, spurring interest among policymakers, energy utilities and researchers in energy feedback as a strategy for making the invisible visible. Indeed, studies especially from the 2000s, argued that energy feedback could increase households’ energy awareness and lead to significant reductions in consumption (Darby, 2001, 2006; Fischer, 2008; Grønhøj & Thøgersen, 2011; Ueno et al., 2005, 2006).

Researchers (Abrahamse et al., 2007; Fischer, 2008; Froehlich, 2009; Grønhøj & Thøgersen, 2011; Karjalainen, 2011) have outlined a series of design characteristics that can impact the effectiveness of feedback: for instance, the frequency with which it is delivered, the measurement unit and granularity utilised, the location and visibility of the system in the home (Darby, 2010; Ehrnberger et al., 2013; Ellsworth-Krebs & Reid, 2016; Hargreaves et al., 2010; Keirstead, 2007), its visual design, whether it provides indications on recommended actions and whether it provides comparisons (seasonal, temporal, or social). For example, Fischer (2008) emphasises the need for clear, frequent, long-term and appliance-level feedback. Nilsson et al. (2018) similarly suggest that disaggregated feedback may help to identify the most energy-intensive activities. Berry and colleagues (2017) find that, ideally, feedback broken down at the appliance level should be combined with energy generation feedback (e.g. in households with PVs) for optimal system management.

However, other researchers have pointed to the limited impacts of feedback highlighting, as one of the main obstacles, the disinterested and demotivated consumers who struggle to make sense of the information received and may be unaware of how to actuate change (Buchanan et al., 2015). Some (Buchanan et al., 2015; Jakobi & Schwartz, 2012) also point at the limits of “one-size-fits-all”, standardised approaches that conceptualise feedback as “universally useful for every consumer in the same way” (Jakobi & Schwartz, 2012: 1) and that do not take into account the high situatedness and embeddedness of energy consumption in specific socio-technical contexts, for which technologies need to be tailored to the complexity of the home and to how people live (Strengers & Nicholls, 2017). Authors such as Abrahamse et al. (2007) emphasise the importance of tailoring feedback to specific goals to effectively engage households in reducing energy consumption. Similarly, Castelli and colleagues (2017), in their 18-month-long qualitative study, developed a web interface that enabled users to personalise their visualisation dashboard, recognising that households vary in knowledge, experiences, attitudes and interests, all of which affect the actual effectiveness of feedback.

Studies emphasising the importance of energy feedback in reducing energy consumption often (implicitly) assume that consumers act as rational agents and understand energy consumption as the result of a “rational and individual decision-making process” (Strengers, 2011: 319). As already mentioned, this approach has been criticised by researchers from the practice theory field. Strengers (2008) points out how the elements of practices are often overlooked when designing energy management systems, stating that more attention should be dedicated to comfort and cleanliness expectations, which are deeply embedded in people’s daily lives and play a key role in shaping their routines related to energy use (Öhrlund et al., 2019; Shove et al., 2012). These routines are also affected by the relational dynamics within the household, which contribute to forming interlinked practices, performed by each individual, all aspects that may undermine those feedback systems designed with in mind mere individual rational consumers (Gram-Hanssen, 2011; Nicholls & Strengers, 2015; Strengers, 2013). Feedback systems, then, need to be designed while keeping in mind these relational dynamics, which affect how such systems are understood and used. Martin & Strengers (2024), analysing the impacts of “non-energy” feedback, highlight all the tacit, sensory, embodied and social experiences that people can learn from when using energy. Also, Gyberg & Palm (2009) point out that knowledge, like energy feedback, is a situated activity, rather than a package to be delivered irrespective of the receiver and independent of time and space, with the expectation that people will act on it. For this reason, it becomes relevant to learn about practices in order to make technology suitable for local contexts and specific households (Juntunen, 2014). Additionally, technologies cannot simply act by themselves: on this matter, Wright (2019) states that people need training in using such technologies to get actual benefits from them. Understanding the ways in which people use these devices becomes fundamental: even the learning process itself may be an obstacle to people by being time-consuming and demanding, as stated by Hargreaves and colleagues (2018). In our results, it is evident how the provision of information is necessary, but, at the same time, is not enough to initiate long-term changes, as practices are affected by a number of contingent factors, not by information alone. Many other authors have questioned the potential for long-term effects of information feedback systems on household practices, especially once the feedback is removed (Grønhøj & Thøgersen, 2011; Hargreaves et al., 2013; Strengers, 2011; Van Dam et al., 2010; Van Houwelingen & Van Raaij, 1989).

Another problem related to feedback is that it can legitimise existing practices and “overlook those considered non-negotiable” (Strengers, 2011: 319). Buchanan and colleagues (2015: 89) refer to this as “the potential for unintended consequences”. For instance, households might learn that they are consuming less than they thought, which can legitimate higher energy usage (named as boomerang effect in Nilsson et al., 2018). This may be the case especially for disaggregated or appliance-level energy feedback. Furthermore, Nilsson et al. (2018) find examples of rebound effects, where energy savings in one sphere can justify higher usage in another.

In conclusion, even if the literature shows mixed results regarding the impact of energy feedback, most recent studies indicate rather limited reductions, which are sometimes even outweighed by rebound or boomerang effects. This is not surprising seen from a practice theoretical perspective, since people’s energy-consuming routines are only to a limited degree governed by rational decision-making, but instead determined by the complex of elements that constitute everyday practices. In this paper, the focus on information delivery serves to see the situatedness of feedback and its effects on practices, its possibilities and limitations, and how it is used by households.

How does micro-generation affect households’ everyday practices?

Whereas the impact of energy feedback on households’ practices appears limited, studies on micro-generation indicate a higher level of engagement among households in changing and adjusting their everyday practices according to the energy produced (Gram-Hanssen et al., 2020). Studies find a tendency among households with PVs, to time-shift energy use to daylight hours in order to be able to consume as much of their own generation as possible (Olkkonen et al., 2017). Dobbyn & Thomas (2005) show that the interaction with micro-generation can foster energy management strategies. Interestingly, Goulden and colleagues (2014) mention that micro-generation seems to foster the development and embodiment of new skills among English households with PVs, including “checking the weather forecast and setting the washing machine to run when the sun was out” (ibid.: 26). Micro-generation may promote a closer and more engaged relationship with energy, by making people become an active part of its production (Ellsworth-Krebs & Reid, 2016; Gram-Hanssen et al., 2020; Keirstead, 2007; Mela et al., 2018). In more general terms, prosumption (Ellsworth-Krebs & Reid, 2016) can be seen as what Strengers (2011: 135) called “energy-making practices”, which involve specific combinations of materials and competences that produce energy (Tellarini & Gram-Hanssen, 2024).

However, studies highlight diverse levels of engagement among households with PVs. For example, some show a lack of behavioural change as the perceived benefits of time-shifting daily activities were considered too small (Palm et al., 2018). Others provide examples of households that even increased their consumption due to the “free” electricity they get (Abi-Ghanem & Haggett, 2011; Baborska-Narozny et al., 2016; Palm et al., 2018; Strengers, 2013). Furthermore, the already established temporalities and the social relations within the household contribute to making the time-shifting of everyday practices (e.g. laundry) more complex and socially bound (Friis & Christensen, 2016; Khalid et al., 2019).

The literature identifies several factors influencing households’ engagement in prosumption. First of all, the type of account settlement scheme seems crucial, with households not compensated for excess electricity being more inclined to shift consumption (Christensen et al., 2017; Gram-Hanssen et al., 2020; Olkkonen et al., 2017). Second, active involvement in the decision to acquire PVs seems to enhance engagement (Dobbyn & Thomas, 2005). Third, people’s experiences with PVs differ according to their experience with the installation process and the availability and the type of information they have access to (if any) (Baborska-Narozny et al., 2016; Dobbyn & Thomas, 2005; Keirstead, 2007). Keirstead (2007) notes that households with little initial energy awareness were among those benefitting most from the information received during the installation. On the other hand, lack of knowledge and support might result in improper usage and inability to maintain the systems (Karakaya & Sriwannawit, 2015). Fourth, the reasons for which people decide to buy solar PVs may differ substantially, affecting the engagement with the system and the use they make of it (Palm, 2018; Schelly, 2014). Finally, the capacity and willingness to time-shift practices also depend on the “rhythm of everyday life and the way in which this is organised” (Ellsworth-Krebs & Reid, 2016), which might not always fit the PVs electricity generation rhythms (Gram-Hanssen et al., 2020).

Energy feedback only rarely figures in studies on the impact of micro-generation on households’ energy-consuming practices. Some studies mention apps providing PV electricity generation information, but households typically rely on weather reports or direct observations to decide whether to time-shift activities (Christensen et al., 2017; Gram-Hanssen et al., 2020), or they might check the display of the PV inverter to see the current energy generation (Gram-Hanssen et al., 2020).

To the best of our knowledge, the impacts on practices of adding appliance-level feedback to households with existing PVs have not been investigated, except for studies focusing on living lab contexts. For example, Paetz and colleagues (2011) provided real-time feedback on household load and appliance power, as well as on PV output, in a living lab setting where both PVs and smart devices were introduced simultaneously, in the first phase of the study. In this paper, we aim to provide insights regarding the influence of first having PVs, and then adding the element of appliance-level feedback.

Fieldwork context

Islands, being among the most vulnerable places when it comes to climate change and its effects, should play a central role in climate mitigation strategies. The Aran Islands, actively engaged in the energy transition, were selected as an interesting field study. The Aran Islands consist of Inis Mór (Árainn), Inis Meáin and Inis Oírr, and are located off the west coast of Ireland, with about 1300 permanent residents (Central Statistics Office, 2022). With no locally sourced fuel, they meet their energy needs by importing fuels from the mainland, to which they are connected via a 3 MW undersea cable that delivers electricity to the islands (Tellarini & Gram-Hanssen, 2024). There is no local microgrid, but the communities are working to become autonomous in their electricity generation through renewable sources (Comharchumann Fuinnimh Oileáin Árann, 2022). Significant progress has been made in upgrading buildings through insulation measures, heat pump installations as well as solar thermals and PV fitting. All the households with PVs are still connected to the grid. There is a mix of electricity pricing schemes on the islands, depending on the supplier, with some people having a flat tariff, charging the same price at all times of the day, and others having, for example, the night saver rate, with cheaper electricity at night. Of the eight households selected for this specific study, only one (H4) had the night saver rate and got PVs at a later stage, in this way having two different systems working in parallel.^{Footnote 2} Five of the households part of the study have batteries (see Appendix A). Through the Microgeneration Support Scheme, introduced in 2022, households can sell back the excess energy produced by their PVs (Department of the Environment, Climate and Communications, 2023). At the time of the field study, this was still quite new, and most households did not yet have the opportunity to benefit from the scheme. For this reason, households with batteries preferred to use directly the energy produced by the PVs and eventually stored in the battery, rather than selling it back.

Each island has a Development Co-op, whose main role is that of dealing with daily matters on the islands, including education, waste management, road maintenance, etc. On Inis Mór, there is also the Comharchumann Fuinnimh Oileáin Árann (CFOAT, or Aran Islands Energy Co-op), a community-owned organisation established in 2012 with the primary purpose of decarbonising the islands and fostering their energy autonomy. The CFOAT and the Development Co-ops together play a central role in supporting households in their process of retrofitting homes and securing heat pumps, solar thermals and PVs.

Data collection and analysis

This study is based on a four-month-long fieldwork (August-November 2022 and March–April 2023) during which the researcher (first author of this paper) resided on Inis Mór and made frequent visits to Inis Oírr. Due to the relatively higher number of solar PV installations on Inis Mór and Inis Oírr, and the collaborations established with the CFOAT on Inis Mór and the Development Co-op on Inis Oírr that facilitated the recruitment process, only two out of the three islands were involved in the study. The CFOAT is also a partner in the project that funded this research.

This paper builds on data collected during the two main phases of the fieldwork (for details on the fieldwork timeline about these two phases, see Table 2 Appendix B). The first one lasted three months between August and November 2022. In this phase, 15 households with PVs were recruited and the first round of interviews and house tours took place. In this first round of interviews, various topics were explored in light of PV usage, such as: organisation of everyday life and activities; experiences with the installation of PVs; knowledge about the PV system; practice changes and time-shifting of practices following the PV installation. Then, eight of the 15 households got smart plugs installed on home appliances (Fig. 1) and feedback on the appliances’ energy consumption was provided through a webapp. The researcher (first author of this paper) took care of the installation of the smart plugs, which in one case was made on the day of the first interview, and in the others, a few days or weeks after the conduction of the first interview (see Appendix B for detailed information on the timeline of each household). Overall, between October and November 2022 all plugs were installed and energy data started to be collected. All the data collected through the smart plugs were transmitted via MQTT to an open-source backend (open.WARE), where they were consolidated and stored. The feedback was ready to be delivered to the households between November and December 2022. The second phase of the fieldwork took place in March and April 2023, where a second round of interviews on the feedback provided was conducted with the eight households. These eight households were selected following three main criteria: First, they needed Wi-Fi in order to have the plugs installed; second, households that were part of other projects, already having some equipment installed to common appliances (e.g. washing machine), were excluded to avoid overloading them with further monitoring systems; finally, households were free to choose whether they wanted to receive the smart plugs and continue with a second round of interviews. Out of 15 households, two did not have Wi-Fi; two already had smart plugs connected to the main appliances; three decided not to have them installed. In Appendix A, more details can be found about the eight households selected, their main energy sources and the appliances connected to the smart plugs. Each household got five smart plugs, which allowed the monitoring of 40 appliances in total. The appliances to be connected to the smart plugs were chosen following two main criteria: firstly, at least two of the appliances had to be washing machine, dryer or dishwasher; secondly, households could choose to monitor two or three appliances according to their personal interest.

Example of the type of smart plugs installed

Full size image

To provide the feedback, a comprehensive analysis of electrical consumption using Smart Plugs (Power and Consumption of single Devices) (Fig. 1) was conducted, complemented by additional data obtained through an API from the REACT project.^{Footnote 3} This supplementary data included information such as the power generated by the photovoltaic (PV) system, the status of the battery storage and overall energy consumption. The feedback on the REACT data could be provided only to the five households who were part of that project: some of these households received only partial feedback according to the availability and quality of the data. The researcher gave the households instructions on how to access the webapp and where they could find the most relevant information. For some households, this happened in person; for others, a video was made and sent to them.

For data analysis and visualisation, we used the open-source visualisation framework open.DASH (Castelli et al., 2017). This platform facilitated the creation of customised data views, empowering users, including researchers, to analyse data without the need for programming skills. The entire process adhered to the principles of End-User Data Work, ensuring that users could seamlessly interact with their consumption data (Castelli, 2020). The open.DASH framework assumed an integral role in visualising the data. Participants received individual logins that provided access to a standard dashboard displaying their real-time device performance and consumption metrics (updated every ten seconds) (Figs. 2 and 3). This dashboard presented key indicators such as current device power and energy consumption over the last four weeks. The information concerning the appliances connected to the plugs included details on the power consumption measured in kWh, the cost of the energy consumed by the appliances (in Euros) and the contribution to CO2 emissions. The energy consumption from the different appliances was collected with 10-s granularity. Along with the real-time information, users were free to change the time period for the display of consumption data according to whether they wanted it over one or more days, weeks or months. Between March and April 2023, the second round of interviews carried out served to discuss the feedback received and practice changes. The households are free to access the platform until the end of 2024, when the project that funds this research will finish.

Screenshot of an example of web interface, displaying four appliances in their real-time usage (on the left) and usage over a week (on the right). Read from top to bottom: dishwasher, freezer, fridge, kettle, washer (i.e. washing machine)

Full size image

Screenshot of another example of web interface, displaying five appliances in their real-time usage (on the top), usage over five days displaying the appliances in the same graph, specifically: dishwasher, dryer, freezer, kettle, washer (i.e. washing machine). On the top right there is information about the appliances’ contribution to CO2 emissions, and in the bottom line the money spent per appliance

Full size image

As previously stated, some of the households (H3, H4, H9, H12 and H14) were also part of the REACT project that installed their PV systems. H4, H9, H12 and H14, in the same period in which they received the webapp for this specific project, also got an app from REACT. In this app, they were provided with information mainly about their PV system, and could see how much energy they were producing, how much was stored in the battery, and how much they were taking from the grid. H9, H12 and H14 thus did not check the appliance feedback webapp much, but instead relied more on the REACT app.

The qualitative data were transcribed and analysed using NVivo software. The researcher used an inductive coding method, starting with the notes and photos taken during the interviews and house visits, and continuing more systematically on NVivo, where codes were created to organise the data into categories and subcategories.

Data collection, including recordings, text, pictures and energy data, followed ethical and legal rules set out by Aalborg university according to General Data Protection Rules (GDPR) for personal data protection, consent and anonymity. Study participants accorded informed consent by filling and signing a consent form.

Quantitative data analysis

A wide range analysis of data collected from the households was conducted to see whether the experiences they reported in the interviews were aligned with their actual consumption. To visualise different scenarios, use cases and feedback alignments, “Matplotlib”, a python library, has been employed. However, after the collection of energy consumption data from households’ appliances, we applied data preprocessing to make the data useful for further analysis. Due to some technical and other errors (e.g. blackout or Wi-Fi issues), a few missing values were observed in the data. To address this problem, the missing values were filled out with the median values. The energy consumption from the different appliances was collected with 10-s granularity. The dataset was collected with such granularity because the focus of this study was to collect and explore detailed energy consumption practices. This study also uses weather data to analyse day and night energy consumption for different appliances and applies sunrise and sunset times to calculate day and night hours.

To detect possible practice changes related to the combination of having PVs and additional appliance-level feedback, we structured the empirical data from the interviews considering two periods in time: the first before the installation of the smart plugs and the provision of feedback (first round of interviews in autumn 2022); the second after the plugs’ installation and the feedback delivery (second round of interviews in spring 2023). See Table 1 in Appendix A for further details on the households, their main energy sources used for daily practices (especially cooking and heating) and the appliances connected to the smart plugs.

Our analysis and discussion are mainly based on the qualitative data collected through the interviews. The quantitative data presented are used to look at the actual practices of people, as valid support for the qualitative data. We have selected five households, useful to illustrate the diversity of responses in our study: these five households were chosen because they represent a wide diversity in practices’ performance, in response to the PVs and appliance-level feedback, showing different ways of engagement with the two systems. For instance, H8 had a high engagement with both PVs and appliance-level feedback webapp, while H3 had a low engagement with both systems. H7 instead showed a higher engagement with the PV system than with the webapp provided; H5 and H9 were analysed together as they initially had a low understanding of their PV system, which then changed due to the different use they made of two apps and to their different electricity schemes. For each household selected, we will start by looking at the first interview, to see households’ knowledge, engagement and experience with their PV system prior to the web app and feedback provision. Then, we will investigate whether and how the feedback provided affected their engagement with the PV system, their energy use in general and their everyday practices. The selected households analysed in this section are: Household 8 (H8); Household 7 (H7); Household 3 (H3); Household 4 (H4); Household 9 (H9). The other households were included in Appendix A for transparency.

H8: High engagement with PVs and appliance-level feedback

Embodied knowledge and habits with PVs

Jack and Ellen^{Footnote 4} had been living in their house for around three years at the time of the first interview (October 2022), and the PVs were already there when they moved. They stated that they both take care of house chores, such as laundry and dishwashing. Ellen is the one cooking most of the time, while Jack takes care more frequently of waste sorting. They both work during the day, so the evening time is when they are usually home. Generally, they showed a high engagement with their PV system and a good knowledge of how it works, but it is Jack the one most interested in it, due also to his job that involves electrical systems, while Ellen considers that she got the chance, with time, to be more involved in it, following Jack’s interest and suggestions. The PV inverter is located in the kitchen, and Jack checks it quite frequently, to see how many kW they are producing. During the first interview, Jack stated that they tend to use most of their appliances during the day, to make the best use of the PVs. Because of their working hours and especially during the “dark months” in winter, with scarce daylight, they tend to use appliances such as washing machine and dryer on weekends, when they are home also during the day. When asked about time-shifting activities, Jack said:

“[…] it's not that hard, it's just kind of… remembering that it’s there [the PV system] really more than trying to figure out, because you're getting free electricity, so… well use it! […] It's just kind of… nearly… muscle memory I guess, if you want to call it that. […] It takes a while to get used… to just remember, but when you do eventually, you remember. It's just a habit, really, yeah.”

In the first interview, Jack said that, on sunny days, they had the habit of putting more than one machine on at once (e.g. washing machine and dryer), because their PVs would be producing the most. When it is cloudy, instead, they would only use one machine at a time. In this regard, many households—including H8—are aware that their PVs produce something also if it is cloudy and dull – although not as much as when it is sunny.

In this case, the materiality of the home, e.g. the location and easy accessibility of the inverter, may have contributed to this household’s high engagement with their PVs and to the routinisation of new practices related to their use. Furthermore, the competences that Jack gained through his job, as well as his interests in the energy field, certainly played a role in the formation and consolidation of these practices.

Practice changes due to the combined knowledge of PVs and appliance-level feedback

This section shows the changes in H8 practices resulting from the integration of appliance-level feedback with the PV system. In the first interview, Jack said that they do not have any means to monitor the energy consumption in their home, but the bill. Thus, they immediately showed interest in the possibility of having smart plugs installed to monitor the consumption of specific appliances. Jack assumed that the appliance consuming the most would be the fridge, being old and, in their opinion, not efficient. As Fig. 4 shows, the fridge turned out to be the most energy-consuming appliance for H8.

Overall monthly consumption (in kWh) of different appliances between 14 November 2022 and 15 May 2023 (in H8). The X-axis represents the total energy consumption for corresponding appliances and Y-axis represents different months

Full size image

During the second interview, both Jack and Ellen participated: Ellen stated that it would be Jack mainly checking the webapp, while she would ask him about it, out of her curiosity to know the consumption of their appliances. In this interview, they clearly showed an increased knowledge of their appliances’ consumption, and even a higher awareness about the functioning of their PVs. This combined knowledge about PVs and appliances’ consumption led to changes in their daily lives. In their opinion, the appliance-level feedback made them realise how much, in terms of kW, their appliances consume, and this made them more careful in their usage, especially in consideration of the fact that they have two kW of PVs.

One of the changes they implemented is that of turning the dishwasher on during the day: in fact, it was the only appliance they were using at night because they believed, being small and quite new, it consumed little electricity.

Jack: “We used to kind of turn on [the dishwasher] at night and stuff… Because let's say […] we thought it was easier on power than it actually is. So now we kind of save that for during the day…”

Interviewer: “Why? What brought you to change this?”

Jack: “Well I’ve seen it was running at nearly two kilowatts as well. So seeing it was nearly like two kilowatt to run… so, I thought it'd only be 500 Watts or something small. But then, seeing it was two kilowatts, I said, okay, we'll leave that now for during the day, instead of at night.”

Figure 5 illustrates what H8 saw in the feedback app, while Fig. 6 shows how they related to the information received: it is possible to see, in fact, that they used the dishwasher mostly during the day. This relates to the app bringing new information and making them realise the actual power usage of their appliances in the home, especially related to their already well-established knowledge on how much they can produce through the PVs. To see the day/night use of washing machine and dryer, see Figs. 23 and 24 in Appendix C.

Screenshot of the feedback app, displaying what H8 saw about their dishwasher

Full size image

Day/night dishwasher usage in the quarter March–May 2023

Full size image

Another interesting example from this household refers to their use of multiple machines simultaneously, when the weather is sunny – as they stated in the first interview:

“Like in a sunny day I turn on the dishwasher and washing [machine] together because you’re making enough. But on a… dull day I’d turn on one at the time to let the panels a bit of power in the one.”

Now, knowing that the PV peak is around two kW and having the new information that each machine consumes around two kW, discourages them from using more than one machine at once. This shows a skilled combination of the knowledge about the PVs, through the inverter, and the newly acquired knowledge about each appliance’s usage, through the web-app, as shown in the quote below taken from the second interview:

“If it’s… a lovely sunny morning, you know, great, we can turn, as much as we can on, so, you turn on the dishwasher first, and when that's finished, go with your washing machine and…You kind of turn everything on one by one then at that stage. […] Because you know when everything's kind of… like we’ve only two kilowatt solar. So you know, let's say, the dryer at its peak is going to use, what's it say, 2400 [W]? So you try and keep the 2000 for the dryer and let the ESB^{Footnote 5} do those 400, which isn't much.”

In Fig. 7, we have represented the consumption of dishwasher, washing machine, and dryer to see whether they run simultaneously or not. The figure clearly shows that they run more than one appliance at the same time only in very few cases. For example, on 17th March 2023, the dishwasher and dryer were running together for some time and the dryer started running just a while before the dishwasher was turned on. So, from the analysis of consumption in the selected period March–May 2023, we can conclude that their engagement with the web app is very high and the alignment between what they reported during the interview and the actual consumption is also conclusive.

Use of washing machine, dishwasher and dryer in the quarter March–May 2023 (H8). The X-axis represents the day and the Y-axis represents the time of the use of the appliances

Full size image

Regarding the frequency with which they checked the webapp, Jack stated:

“[...] I was on it a lot in the beginning let’s say. The first two weeks I was kind of checking a lot, but then, I might check it now every two weeks because [...] I know the power kind in my head and what everything is doing. […] I know the usage in my head, so I check it maybe once a week to make sure they're all the same power and nothing's broken.”

Also when using the app, H8 incorporated new practices that became normalised – at least in the short-term^{Footnote 6} – as they did not find it necessary to constantly monitor it once the information was absorbed and new practices became routinised. This recalls something mentioned earlier, when H8 incorporated new habits following PV usage. In their case, having once learnt about appliances’ usage, frequent checks were not deemed necessary. Instead, they wished they could shift the plugs to some other devices to see their usage and get an idea of their efficiency.

In this example, new materials such as the plugs and the webapp enriched the already well-established competences and knowledge about PV production by providing information on the specific usage of each appliance, leading to some practice changes.

H7: Good knowledge and engagement with PVs, lower engagement with appliance-level feedback

Making best use of PVs during the day

In H7, Bryanna takes care of most of the chores, e.g. cooking, doing laundry, etc. This is mainly due to her working hours which allow her to be home during the day and at lunchtime. In the household, Bryanna considers herself to be the one being more attentive to the PV system, and with the greatest interest in it. They installed PVs in 2018, and she states that in her opinion “the longer you have it [the PV system], the more conscious you become of it”. Because of her working hours, Bryanna uses most of her appliances during the day. On sunny days, she tries to turn as much as possible on, “two washing machines, two dryers, whatever gets done by 16 o’clock”. For her, using energy as much as she can during the daytime became a natural habit: “There’s no point to me having the solar panels and putting everything at night” and even if it is cloudy, “you’ll [still] have something, it’ll be cheaper than pitch black”. The meter, located at the entrance, has a flickering light that helps them be aware of when the PVs are producing or drawing electricity from the grid. They do not consciously check it, but Bryanna says: “You notice if you go past, if it’s flashing”.

Compared to H8, it seems that H7 has a higher tendency to use multiple appliances at the same time (Fig. 8). Figure 8 also shows their preference for using most of the appliances during the daytime, with few exceptions – especially in relation to the dishwasher (see also Fig. 9).

Use of washing machine, dishwasher and dryer in the quarter March–May 2023 (H7)

Full size image

Day/night dishwasher usage in the quarter March–May 2023 (H7)

Full size image

Both Figs. 8 and 9 show that the dishwasher is used quite frequently at night. This may be due to several factors: Bryanna states that, despite using most appliances when there is daylight, sometimes the dishwasher is put on at night for a short wash. She says that she usually tries to put it on during the day when she comes home after work, but her husband sometimes puts it on at night, so that he can empty it the following morning before going to work. She believes that this is okay and she does not mind her husband doing that: first of all, she thinks that, being a short wash, it is not expensive – also from checking the webapp. Then, she states that sometimes “you just have to put it on”.

Figures 10 and 11 display a more consistent use of the washing machine and dryer during the day, similar to what Bryanna stated.

Day/night washing machine usage in the quarter March–May 2023 (H7)

Full size image

Day/night dryer usage in the quarter March–May 2023 (H7)

Full size image

See Fig. 25 in Appendix C for the overview on the consumption of the measured appliances in H7 for the period 11 November 2022–27 September 2023.

Increased awareness, lack of change: effects of the webapp and appliance-level feedback

Bryanna has not checked the webapp much. She says that she would have been more involved with it if there were notifications reminding her to check it. The only change she reports regards the usage of the TV, as her mum used to have it on most of the day, as a radio. Once, they checked the app and read how much they had spent until that day on the TV, and not knowing exactly what period in time that number referred to, they got worried that it could be the daily or weekly spending: “I was looking at the thing one day and I said, we have used 84 euros on the TV” (see Fig. 26 in Appendix C for information on TV usage).

In general, Bryanna thinks that they cannot change much in their daily consumption, as they are already careful about using most of the machines during the day. She refers to the TV and the kettle as those that she would be more interested in checking, as she feels that she cannot do more for washing machine and dryer: “I wasn't really looking at the washing machine and the dryer because it's on during the day. So, I feel as if they’re okay!”. During both interviews, Bryanna laughed about their kettle, which she thinks is on all the time. She tried to cut her usage of the kettle, for example by avoiding the habit of turning it on and walking away to do some other tasks while the water cools down again, meaning that she needs to reheat it. However, she was surprised to see in the webapp how little the kettle consumes:

“And the kettle, I was surprised how little… [...] I was like, no, you put it on once, don't walk away! But then, yeah, put it on, reheat it, reheat it again, back to my old ways!”.

To see how often Bryanna turns on the kettle, we computed its turn-on time frequency considering its power (Fig. 12). As shown in Fig. 12, the frequency varied widely across the days. Reflecting her statements, Fig. 12 shows that she turned on the kettle every day and, for some days, more than 20 times.

Kettle turn-on frequency per day for H7 in the period March–May 2023

Full size image

Her usage of the kettle is also affected by the social relations within the household, and she feels that she has no control over when and how often her family members turn it on. Besides the family members living in the house, she states that her door is always open, so it is very common for neighbours and relatives to walk in for a chat and tea, turning on the kettle themselves. The use of the dishwasher displayed a similar effect of household dynamics on practices, highlighting that the individual is often not alone and independent in their performance.

It is interesting to see the variations between informants in their practices’ performance and energy consumption, due to different ways of reading the information provided. Looking at the interconnected elements that form practices, it is possible to see how, for instance, competences and meanings affect how people perform their practices and relate to the information received and to the materiality of their home (which includes PVs). For H7, the dishwasher consumes a similar amount to the one in H8, around two kW, and they have the same PV capacity (two kW). Still, H7 mostly focused on the webapp information provided about the money spent, while H8 was more interested in the appliance power consumption (kW). These different interpretations of the webapp, as well as a different engagement with everyday practices, led to different responses from the households in their appliances’ usage.

H3: Low engagement with both PVs and webapp

Disinterest and discouragement in using PVs

In H3, both interviews were carried out with Claire. Claire’s husband handles the PV system, but he is out for work during the day, while Claire stays home dealing with most of the chores. She believes that summer is their highest consumption period due to hosting tourists (see Fig. 27 in Appendix C for details on overall appliance consumption). For instance, the washing machine is on most days during summer, but primarily on Saturday nights during winter. Generally, this household showed a lack of engagement with their PV system. They never check the inverter, as they were discouraged by installers and technicians who warned them not to touch anything. Furthermore, the inverter is placed in the attic, which makes it quite inaccessible. Claire stated that, to her, the inverter and the information displayed there do not mean anything, and she would need to have it explained by someone. Since having PVs, Claire has maintained her routines, finding no reason to change them. When thinking of the usage of her appliances, she reflects on the non-negotiability of most of her practices, for which “you have to put on your washing machine”, “the kettle has to be boiled”. Time-shifting her daily habits to her means disrupting her day: for example, if she washed her clothes on Sunday morning instead of Saturday night, “you’re kind of a day behind if you do that”. She dries clothes out but also, similarly to other interviewees, gives them a quick blast in the dryer. Because of this, like all the other interviewees, she feels very much dependent on the weather, not in relation to the PVs, but more to the possibility of drying clothes out.

According to Figs. 13 and 14, both washing machine and dryer are used mostly during the day (with few exceptions), which may be due to the household’s preference for hanging clothes out to dry them before giving them a final blast in the dryer.

Day/night washing machine usage in the quarter March–May 2023 (H3)

Full size image

Day/night dryer usage in the quarter March–May 2023 (H3)

Full size image

Meaninglessness of the webapp and appliance-level feedback

Similarly to Bryanna (H7), Claire believes that their kettle is “always warm”, as relatives and neighbours have the habit of entering the house and turning on the kettle to make tea: “Nobody looks to see, is there water in the kettle?! Only me! [laughing] […] They just take for granted and they press it, yeah. […] So it’s a habit.” Because of this habit, she feels that she needs to keep the kettle always filled with water, to avoid the risk that it gets switched on with no water in it. Due to their high usage of the kettle, the only time Claire got to check the webapp was to monitor the kettle, but after looking at the graphs, she said that she could not make sense of that information. She also did not talk about it with the other family members, underlining her lack of interest in it. In general, she did not understand the use of the webapp: “I have to boil the kettle, it doesn’t matter what I use. I have to wash the clothes… […] The bill is coming, whether I have the app or not”. As she made a similar statement about the non-negotiability of her daily practices in the first interview before the webapp provision, it can be noted that the webapp did not promote any change.

Similarly to H7, Fig. 15 shows a high usage of the kettle, with changes occurring throughout different days. Though on average the turn-on frequency for H3 is lower than for H7, it varies widely across the days and for some days the frequency is above 20.

Kettle turn-on frequency per day for H3 in the period March–May 2023

Full size image

H4 and H9: Limited use of PVs, despite high willingness to learn more about it

In this paragraph, we will analyse the cases of two households. They were also part of the EU project (REACT) that installed PVs and batteries in the homes of their participants. In the first interview, both H4 and H9 showed interest in the PV system but complained about the lack of information and feedback on how it works. About the same period when they got our webapp for appliance-level feedback, they then got a feedback app from the REACT project, where they could check their PV production and battery status. One significant difference between the households is how they related to the two apps and the impact these had on their practices.

Interest in the PV system, but lack of knowledge on how to use it

In H4, Anne and her husband share responsibilities for cooking, dishwashing and doing laundry. They both work, but Anne is home more often, especially in the afternoon, due to her more flexible work schedule that allows her to take care of the children after school. Already from the first interview, H4 showed great interest and willingness to learn about how their PVs work. However, Anne found that they did not have any information on it yet and their battery did not work properly. She felt “desperate for some information” about her PVs and battery. For example, she did not know what the inverter is and where it is located in the house. Furthermore, they are on a night saver tariff, which makes electricity use cheaper at night: because of this, even after getting the PVs, they stuck to the old habit of setting timers on the dishwasher and washing machine for them to start in the night or in the early morning. This is also convenient for them because they can empty the dishwasher and hang clothes out in the morning before going to work. According to Anne, the PVs had no impact on their energy usage at the time of the first interview:

“I think once I actually see that [how the system works], it will make an impact, but because we haven't seen, like, I have no idea how much it's producing or not producing, so when it is producing or not, so it's not making, um, an impact on my habits. [...].”

When asked if she ever refers to the weather to put on certain machines, she said that even if the sun is shining, she would not go and put extra washes on. Figure 16 shows that, in most cases, appliances are not used simultaneously. Compared to the other households, H4 uses the appliances, especially dishwasher and washing machine, in the early morning (ca 5 to 9 AM), reflecting their usage of the night saver mode.

Use of washing machine, dishwasher and dryer in the quarter March–May 2023 (H4)

Full size image

In H9, it is Máirtín being home most often, as he is retired while his wife still works. He states that he is the one taking care of maintenance jobs, and he believes that, since they got the PVs, he is more attentive in putting on the laundry when it is sunny, so that he can also hang it out for drying. Sometimes, his wife puts on a wash in the morning before leaving for work, and he hangs the clothes out when it finishes. Máirtín showed a strong interest in the PV system but, similarly to H4, he felt that they lacked information about it, stating to be “in desperate need of information”. He looked forward to having some feedback, with a specific focus on the battery. He never goes to read the inverter to check how much energy they are producing because “it’s difficult to get to it”, being in the attic. Unlike H4, H9 does not have a night saver for electricity, so even if Máirtín did not have detailed information on the system, he tries to adapt his washing habits to when the sun is shining – although laundry was already very much connected to having dry days, in order to put clothes out on the line.

H4: Increased awareness through the app, but higher resistance to change

Anne said that the biggest change brought by the app was a heightened awareness, especially about the kettle. In fact, she believes that seeing the plug and especially reading the information on the webapp, made her avoid switching it on continuously:

“[...] I had a bad habit of, I put on the kettle, and then I'd get distracted by something else, and then I'd have to put it on again, later and, so I'm not doing that now.”

For example, if they have the stove on during the evening time, she uses an old kettle, putting it on the stove and waiting for it to boil, rather than turning on the electric kettle and having tea instantly. In this case, her environmental values and her attention to energy costs, which she considers to be quite relevant in determining her life choices, play a central role in forming the meanings at the base of her everyday practices, contributing to the development of new practices. The kettle was also the appliance she was most interested in checking on the webapp (see Fig. 17), as she defined it as a “luxury item”, something that she can have more control on – opening the discussion on the negotiability of everyday practices:

“I suppose I have control over the kettle, like it's a, it's a luxury item, I only use it to, you know, mostly make teas and coffees and things. We don't have to do that, whereas washing clothes we have to do it, washing dishes we have to do it. Um, the kettle is something that, I can cut down on, um, if needed, do you know? So, that's probably why it's more of a… of interest, it's more in my control.”

Anne showing how she used the app, especially in focusing on the kettle and comparing its usage to the one of the other appliances

Full size image

Figure 18 shows that the number of times H4 switches on the kettle is much lower compared to H3 and H7. In this case, it seems that the webapp made the household inhabitants aware of their kettle’s energy usage, making them change some of their practices.

Kettle turn-on frequency per day for H4 in the period March–May 2023

Full size image

Regarding other appliances, Anne kept using them during the night or very early morning, making use of the night saver. Furthermore, she thinks that they are already quite conscious of their energy usage, so they would not be able to change much to adjust their consumption (see Fig. 28 in Appendix C to see her appliances’ overall monthly consumption). Thus, similar to H7, H4 shows a high awareness about the timing of their consumption, but while H7 adjusts their use of appliances to the PV generation during daylight hours, H4 adjusts their appliance usage to the night hours due to the night saver.

Anne used to check the webapp most often than her husband, but she stated that sometimes they have talked about it and checked it together on her phone. Even though Anne found the app full of interesting information, well-designed and easy to understand, she believes that in their case it did not have a huge impact:

“I think if I was starting from kind of the basic level, then yes it would, but I think, I had already been trying to reduce, [...] because we have the night saver…”.

In a way, the webapp made her more aware of using the night saver for electricity. Figures 19 and 20 show how the night use of washing machine and dishwasher prevails in H4. See Fig. 29 in Appendix C to look at the dryer’s usage.

Day/night washing machine usage for the quarter March–May 2023 (H4)

Full size image

Day/night dishwasher usage for the quarter March–May 2023 (H4)

Full size image

When she was provided with the feedback app from REACT, she found it difficult to read, also because her battery was not working yet and therefore she could not see proper information about it. This surely had an impact on her ability to time-shift her practices during the day.

In general, though, she thinks that the provision of information is fundamental: “People aren’t going to change unless they’re made more aware of what’s actually happening.” This is a further confirmation of the importance of providing people with information as an initial step to actually applying (conscious) changes to their daily lives.

H9: Information as a necessary tool to increase engagement with the PV system

In the second interview, Máirtín admitted that he had not checked the webapp with appliance-level feedback, because he was not sure how to use it. Instead, he got very much involved with the REACT app, which provided feedback on the battery status and PV production. His knowledge appeared to have increased considerably in the period between the first and the second interview: “I find that wonderful that even on a rainy day it’s producing more than a light bulb would be using…”, demonstrating that he had gained the knowledge that other households (e.g. H7 and H8) already had, having had PVs for longer. He checks the REACT app few times a day, but especially in connection to the usage of the dryer, as he feels it is the appliance that he has more freedom to decide when to turn it on. Despite not using the appliance-level webapp, he showed great interest in appliance-specific consumption and used the second interview to ask questions to the researcher and see real-time data on appliance usage (e.g. by turning on the kettle and see its real-time consumption).

Interviewer: “What appliances do you think you will be checking more often?”

Máirtín: “The kettle and the dryer. Yeah. Can't do anything about the freezer.”

Máirtín also showed interest in combining the knowledge gained from the two apps, for instance by looking at how much is stored in the battery and how much the dryer consumes. When asked why he is mostly interested in checking the dryer, he said:

“Well, I feel the dryer is a heavy… I don't know that, I'll be asking you to look at, at the thing [the webapp] to tell me what's using the most here. But it's one thing that's, I can decide when to turn it on, you know? […] A lot of the other activities are going to be on and I don't have choice in when they’ll be on, the television… the fridge, the freezer, the, well, the kettle, I should be able to check [both laughing], but, I don't feel I have control over that really, if somebody is tea time, somebody wants tea, whatever…”

Figures 21 and 22 show that H9 mostly used the washing machine and dryer during the day, with few exceptions. Interestingly, they hardly used the dryer in May, specifically from 14 to 28 May, though they still used the washing machine on a regular basis. This may be due to more sunlight to hang their clothes out in that specific time frame, to other ways of drying clothes, or to other occurrences that might have happened.

Day/night washing machine usage for the quarter March–May 2023 (H9)

Full size image

Day/night dryer usage for the quarter March–May 2023 (H9)

Full size image

For more information on appliances’ usage in H9, see Figs. 30 and 31 in Appendix C.

Few authors (Paetz et al., 2011; Baborska-Narozny et al., 2016; Berry et al., 2017) have studied the impacts of combining micro-generation with appliance-level feedback to promote changes in energy usage and everyday practices. This has been the focus of our study. Overall, the previous experience with the PV system may affect the ways in which the households respond to appliance-level feedback. This experience includes e.g. the length of time in which people have been living with PVs; the location of the PV inverter and subsequent accessibility of information; the support households received in understanding the functioning of the system.

Proximity to micro-generation in homes does not lead, by default, to practice changes (Dobbyn & Thomas, 2005). Generally, the provision of clear and accessible information appears to be an important, initial step to promote changes in people’s practices. The same goes in the case of appliance-level feedback and related devices: having them does not lead automatically to practice changes. In our case study, several factors impacted people’s responses to feedback. Firstly, the webapp was designed to let users freely interact with their data (Castelli et al., 2017), and this led to different ways in which households, according to their previous knowledge, understanding, interest and experience, interacted with the app and the use they made of the feedback received. For example, H8 adjusted appliance use based on the information on power consumption, in order to relate this to the power generation of the PVs, which led them to adjust their dishwasher use and to avoid turning more than one machine on at once. H7, on the other hand, based their decisions mostly on cost: this led them to stick to their habits of turning the kettle on frequently and using the dishwasher in the nighttime, seeing that they are not as expensive as they thought. H3, on the other hand, did not respond to the information received and the graphs presented, finding them quite difficult to understand, similarly to few other households that found e.g. the colours used in the webapp misleading and confusing. If, on one side, tailored feedback and giving users the freedom to visualise what they are most interested in can increase the feedback effectiveness, there is, on the other, a need of educating people in such systems and in how to read the information that they receive. Furthermore, the design of the webapp itself, and whether the information is provided in an accessible and user-friendly way, surely impacts whether and how people use it.

Secondly, the duration of feedback provision did not seem to affect people’s ability to change practices. Rather, notifications were indicated as something desirable by all households. The first round of interviews showed how households with a good engagement with their PV system chose to time-shift daily activities in a way that became natural and normalised to them. Similarly, regarding the appliance-level feedback, once the users had learnt about their appliances’ consumption, checking the webapp regularly was considered unnecessary (e.g. H8 checking the webapp only to make sure that nothing is broken). These results partly contradict recommendations that effective feedback should be given continuously, over long periods of time (Fischer, 2008). This connects to the interest of many of the households in shifting the smart plugs to different appliances after having learnt the consumption of the measured ones – in line with Jakobi & Schwartz (2012), who highlight the importance of having flexible switches where different appliances can be plugged in. For instance, some households (like H8) would have liked to shift the plugs to other appliances, even only for few days, just to get an idea of their overall consumption, or to check whether older appliances need to be replaced (H7). Other households would have liked to use the plugs on newly bought appliances to see their consumption, which aligns with the results of Castelli and colleagues (2017) showing that some users would only check the web interface when getting a new device. Furthermore, households agreed that life is too busy to go and check the app regularly, while they all would have appreciated notifications, popping up in cases of high energy usage (H7) or notifying when it is a good time to be using something (H4).

Thirdly, people’s embeddedness in their everyday practices is central in affecting feedback effectiveness, beyond people’s technical literacy and engagement with the information provided (Ellsworth-Krebs & Reid, 2016). The results show the situatedness of everyday practices, deeply embedded in the context where they are performed. This is evident especially when seeing how households with very similar systems and with the same kind of energy feedback, responded to the information provided in different ways, when performing their daily practices. Materials (e.g. the inverter and its location and accessibility in the home) are part of this situated context where daily life occurs, and contribute to form practices. The plugs and related feedback became part of such materiality of the home. Competences also contributed to the shaping of practices, e.g. the ability of people to read and relate to the information provided. For instance, in H8, Jack's technical knowledge was reflected in the selection of the information relevant for them, i.e. kW use, which impacted their dishwashing practice. Finally, the meanings also played a role, in terms of what people consider relevant in their daily life, e.g. H3 finding it impossible to negotiate their daily practices, or doing laundry when the weather is sunny due to their preference of drying clothes out, and H4 considering the kettle as a “luxury item” and finding alternative ways to use it. In general, existing competences and meanings may lead to the formation of new practices, and so of new competences and meanings. Furthermore, basing the analysis on daily practices served also to consider the contingency of households’ social relations, proving how everyday practices are part of more extended bundles^{Footnote 7} interconnected with each other.

In the first interview, people talked about the practices that they considered negotiable and that they time-shifted to match PV production. In general, washing practices are those most commonly time-shifted – especially given the possibility of using timers to programme the starting time of washing machines, dryers and dishwashers. The appliance-level feedback helped illuminate which practices people considered negotiable or non-negotiable in their everyday lives. The interviewees tended to check the appliances related to practices that they thought were most flexible to change. For instance, a few households (H8) were interested in their fridge and freezer, which most other households were not checking as they felt they had no control over them. H7 and H4 were interested in the kettle due to its frequent use, while other households (like H8 and H9) were more involved in checking larger appliances such as washing machine and dryer, as they believe they have more flexibility about when to turn them on. However, most households considered the kettle something that they had no real control over, because it is closely linked to the social life of the home (H3 and H7). The feeling of being able to control the usage of certain appliances is one of the factors that affected what people kept an eye on. In general, it is interesting to see how some households found alternative ways to reflect on their consumption and bring changes to it: for example, by reducing the amount of water put in the kettle, so that it stays on for shorter periods (e.g. H9), or by putting the kettle on the stove in order to heat the water (H4).

Lastly, four main responses to the combination of appliance-level feedback and micro-generation were observed. For households already familiar with their PV system, two scenarios emerged: one is that they found the information interesting, but made few changes, believing to be already aware of their energy usage, and could not change more (e.g. H7). This connects also to the negotiability of practices, as these households, since having PVs, already time-shifted those considered negotiable. For this scenario, providing appliance-level feedback did not lead to major practice changes, and could even lead to unintended consequences (Strengers, 2011; Buchanan et al., 2015), e.g. by normalising the (unsustainable) usage of some appliances (e.g. H7’s kettle use). The second scenario regards those households that used the information from the appliance-level feedback to further maximise PV usage. For instance, H8 stopped turning on appliances simultaneously after getting the information about their actual power consumption, and time-shifting the usage of the dishwasher during the day, after realising its real usage.

For households with limited knowledge and understanding of their PV system, two other scenarios emerged. First, for some households this lack of knowledge is reflected in a general lack of engagement with the appliance-level feedback and resulted in no changes in everyday life (e.g. H3). To some extent, this might be related to the deep embeddedness of energy consumption in everyday practices (e.g. H3 saying “you have to put on your washing machine” or “the kettle has to be boiled”), rendering change difficult. For these households, feedback systems may not have an impact, but there may be other ways to make them involved, e.g. through better in-person support, sensorial feedback and increased social engagement (Martin & Strengers, 2024). Finally, the last scenario is represented by those households who lacked information on their PVs, and who made use of the appliance-level feedback in different ways. For instance, H4 found the webapp useful for increasing awareness and making them stick to their habit of using their night saver. In this case, the webapp also resulted in innovative changes in practices – e.g. heating an old kettle on the stove instead of using the electric kettle to heat the water. H9 instead relied on the feedback on their PVs and battery storage, which, in their opinion, was the main information they needed to start using more consistently certain appliances when it is daylight.

This paper aimed to explore the combined effects of micro-generation technologies and appliance-level feedback on practice changes. This was studied on the basis of two interviews per household: one performed prior to providing them with plugs and a webapp (focusing primarily on existing household practices in connection with the PVs) and the other performed after the households had access to plugs and the webapp for few months (focusing on their experiences with appliance feedback and how this affected their practices). In the analysis, qualitative interviews were combined with quantitative analysis of the plugs’ data.

An important observation is that providing households with information and educating them on how energy is consumed is a necessary prerequisite for them to change their daily practices. However, this study also shows that information alone does not, by default, lead to change, as its effectiveness is conditioned by other factors as well – including the (non) negotiability of practices, system (e.g. app) design, household dynamics and personal abilities and interest in relating to this information. In some cases, households showed increased knowledge and awareness about their energy usage and the consumption of their appliances without changing anything.

Our data revealed varied responses to appliance-level feedback among households already equipped with PVs, depending also on households’ existing interest and knowledge of the PV system: while some felt that their involvement in time-shifting and energy saving through PVs was sufficient, others embraced the feedback, leading to further changes in their energy consumption practices. First, it is relevant that people learn about how their PVs work to fully understand when it is the best time to turn appliances on. Without this knowledge, they find it difficult to relate to the appliance-level feedback – or they find it interesting, but stick to their old habits or find alternative ways to reduce consumption. Second, even if there is need of educating and supporting people in using feedback systems to help them understand the technical information provided, it may not be enough to provoke change. For example, there are also several other types of feedback (as outlined by Martin & Strengers, 2024) that affect people’s usage of energy, which were not discussed in this paper, but that are still important to acknowledge – e.g. the importance of social involvement or sensory experience when using micro-generation and feedback technologies. Third, people’s embeddedness in their everyday practices must be considered as deeply affecting their usage of PVs and appliance-level feedback; not taking this into consideration may render the possibility for change even more complicated. On the other hand, change does not necessarily entail decreasing consumption. Appliance-level feedback combined with micro-generation also involves risks, such as the normalisation of certain practices and the perceived irrelevance of appliance-level feedback because the changes already brought by having PVs are considered sufficient. In general, to promote change, feedback provision in understandable ways and suited to people’s practices through more user-friendly systems, technical literacy, information availability and support in learning about these systems are all necessary starting points, the absence of which will make any change towards a more optimal consumption of intermittent micro-generation difficult. Such factors need to be taken into consideration to promote change towards a better match between local energy consumption and the timing of local (micro-)generation – while keeping in mind the risks of information provision normalising highly consuming everyday practices.

This study presents some limitations that may serve as inspiration for future research on similar topics. Our research lacks the measurement of the energy consumed by the appliances before the feedback delivery: for this reason, change was analysed mainly through the qualitative data collected through the interviews. Ideally, it would have been valuable to start measuring appliance-level consumption over a period of time without delivering any feedback, then offer the feedback and afterwards assess any changes in consumption. One other aspect that was not possible to research in this project and that could be of inspiration for future studies, is about exploring the long-term effects of appliance-level feedback in households with PVs, e.g. by doing follow-up interviews after one year of the feedback provision. New studies in this field context should also consider the recent smart meters rollout, which happened between 2022 and 2023, when fieldwork took place. Some households then may be under a different tariff now, something worth to be considered.