Domesticating energy flexibility. Learning from great britain’s 2022–2023 demand flexibility service

Electricity systems have traditionally been demand-led, flexing supply to meet people’s fluctuating needs for energy services. Most of the world’s electricity came from large fossil-fuel or hydro plants and these could be switched on and off with relative ease, especially if gas-fired or water-powered.

Starting in the 1950s, increasingly high and short-lived peak demand, especially in regions with growing air-conditioning and electric heating use, challenged the demand-led model (Bellarmine, 2000). To avoid investing in underutilized generating capacity, utilities began incentivizing customers to shift demand away from peak times, marking the start of a shift towards more flexible demand. With the growth of wind and solar generation, transmission and distribution networks must now adapt to two-way electricity flows, with both supply and demand fluctuating based on weather and time of day, making DR increasingly urgent and valuable.

DR is broadly defined as ‘the change in electricity consumption patterns in response to a signal’ (Element Energy, 2012, p.9). The signal is usually a change in price, however, non-financial factors like trust, perceived risk, and effort also influence the motivation and effectiveness of DR (Parrish et al., 2020).

In the long term and from a system standpoint, DR can reduce the need to invest in generation and infrastructure (Pourramezan & Samadi, 2023). For example, a 2015 European analysis identified peak demand as the main driver of network costs and estimated that a 10% load shift away from the peak could reduce costs for distribution system operators by 5%. There can also be substantial advantages for transmission grid operators from delaying infrastructure investments (Koliou et al., 2015). A study by the Carbon Trust and Imperial College London (2016) estimated cumulative savings of £17-40bn by 2050 from flexibility in GB, depending on levels of demand and interconnection, and on the costs of storage and DR programmes.

In the short term, DR enables system operators to maintain reliable energy services at reduced economic and environmental cost (F. Wang et al., 2017). Peak load shaving – the DR most relevant to this paper – can offer direct benefits to those who own and operate an electricity system, as indicated above. It can also offer direct benefits to customers who participate in DR: reduced bills from cheap off-peak usage, and perhaps the chance to adopt useful new technologies; and indirect benefits to all users, including better system reliability and reductions in overall system costs (Uddin et al., 2018).

While residential loads are relatively small, there is a strong argument in many regions for DR in the residential sector, given the scale and diversity of demand from this sector, especially at peak times (Cruz et al., 2021). The earliest residential DR programmes took place in North America and northern Europe, and they now operate in many countries including China (Z. Wang et al., 2020), India (Ravindra & Iyer, 2014), Iran (Saebi et al., 2022) and Brazil (Oliveira et al., 2023). The type of DR in each programme will vary, depending on characteristics such as climate, supply mix and patterns of demand. It can take many forms, from load reduction or load-shifting a few times a year, when demand reaches a’critical peak,’ to switching smart-enabled devices such as freezers off and on for very short periods, to maintain system frequency. Each form of DR will be associated with tariffs, communications and enabling technologies (e.g. smart meter, in-home display, smart plugs and thermostats) to achieve its purpose (Darby & McKenna, 2012; Parrish et al., 2020).

In summary, there is an established rationale for residential DR as a way of operating electricity grids more efficiently during an energy transition that aims to provide reliable energy services from renewables-based supply.

Analysts have long recognised that DR programmes must attend carefully to the needs of participants (Garcia, 1987), given the ‘multidimensional role that attitudes, habits, and experience have in shaping energy consumption’ (Sovacool, 2014, p.11). Recent studies focus on relational and socio-technical interpretations of flexibility, emphasising interactions between electricity users and artefacts (Blue et al., 2020; Lo Piano & Smith, 2022; Powells et al., 2014). We continue in this direction, using a domestication theory framework (see Sect. "Analytical framework") to map out attitudes, routines, and interactions with artefacts that shape our participants’ experience with DR. Factors that influence residential DR uptake vary with context. For example, a choice experiment study on US and EU participants found that environmental and system benefits can have an impact on willingness to participate (Buryk et al., 2015) and a study in Israel found that users value environmental and national security objectives when deciding to enrol to a time-of-use tariff (Parag, 2021). A Finnish study however found that households enrol their loads in Direct Load Control programmes for economic reasons over environmental benefits (Sridhar et al., 2023). The potential for energy savings was also found to be the strongest motivator in a US study (White & Sintov, 2018), but the authors found that users may have unrealistic expectations of these. A UK study found that incentives-based approaches can overcome some barriers to DR varying by the height of the barriers, electricity-using practice, and incentive approach (Bradley et al., 2016). A study in Australia concluded that time-of-use tariffs are unlikely to reduce peak demand effectively in households with children, potentially exposing them to inequitable impacts (Nicholls & Strengers, 2015).

The concept of flexibility capital, introduced in Sect. "Introduction", has gained widespread acceptance in recent years. It can be applied to continuing participation in DR as well as to uptake. For example, a wealthy, high-consuming retired homeowner with solar PV and an electric vehicle or batteries will be more likely to profit from a peak-reduction tariff than a single-parent shift-working tenant on a low income who must fit activities such as cooking, showering and laundry into limited periods of time.

More recently, the concept of flexibility justice has been developed to show how ‘the costs of doing flexibility work are not evenly and fairly distributed between and within different households’ and how ‘framings of end-user flexibility … would offer an important step to distribute the burden of flexibility work more evenly’ (Fjellså et al., 2021, p.107). Insight into customer household circumstances matters at the level of individual customers; at the macro-scale, so do policies beyond the electricity sector such as those for taxation, housing, and transport (Winther & Sundet, 2023).

The concept of capability has been used in relation to DR and energy justice, highlighting what people can do – their roles and abilities as social beings in societies that rely increasingly on smart technologies (Banks, 2022; Roberts et al., 2020). We will draw on some of these concepts and references in our discussion section.

The rest of this paper unfolds as follows: In the following sections, we provide the context and relevant details of the DFS (Sect. "Overview of the 2022/2023 Demand Flexibility Service"); domestication theory, the framework through which we analysed the DFS, and its application to DR (Sect. "Analytical framework"); collection and analysis of the twenty-five diaries (Sect. "Methodology"); and the mapping of five different pathways for domesticating the DFS (Sect. "Findings"). We then discuss our findings and draw lessons for the design and implementation of future DR response policies and programmes (Sect. "Discussion and implications") and conclude with a research and policy agenda for maximising the contribution of DR to decarbonising the electricity system (Sect. "Conclusions").

We recruited a total of forty-two participants in September 2022, as soon as the DFS was announced, through posts on specialised social media platforms (e.g., Facebook and Reddit groups focused on DR and micro-generation) and through the authors’ professional and personal networks.

Participants received a diary template structured around the three dimensions of domestication reflecting Sørensen’s (1996) framework, as shown in Table 1. Annex 1 provides an example diary from one of our participants who gave permission to have it published alongside this article. Conversations about the DFS could have fit under any of the three dimensions. We chose to position them within the symbolic dimension because we wanted to test how participants convey their attitudes towards the DFS.

Participants agreed to fill in the diary electronically after each DFS session and received reminder emails after each session prompting them to reflect their experiences in the diary even in cases where they did not sign up to the session (in which case the diary prompted them to explain why that was).

Twenty-five participants returned the completed diaries in March 2023 alongside reflections on their overall DFS experience e.g., what they found most difficult and frustrating, what went well, how to improve the programme, and any other information they believed would be relevant for the study. We also held informal follow-up conversations with some of the participants in winter 2023 to understand if they would participate in the 2023/24 DFS. The diaries were immediately pseudonymised, so all the names used in the sections below are pseudonyms. We had a mix of depth and detail provided in the diaries and the example in Annex 1 reflects in the round the extent to which participants engaged with this exercise.

As discussed by Parag et al. (2023), recruiting a diverse sample is challenging when conducting energy research in a hurry. The urgency with which the DFS launched made it difficult to reconcile sample diversity with the depth we were seeking in our data to align with the domestication theory framework.

Our choice of diaries as a method allowed us to collect a longitudinal dataset covering participants’ experience over the entire duration of the DFS in alignment with the analytical framework which views domestication as a process which takes place over time. We prioritised obtaining data from each DFS session (starting from the first one) to be able to track routine formation, learning processes, changes in attitudes etc. over as long a period of time as possible. Due to time constraints, we were also unable to recruit beyond specialised interest groups. Despite not being generalisable to the wider population, however, we believe our approach is valuable in that it gives insight into the raw and detailed reflections of the DR experiences over time. This complements wider studies (e.g., Centre for Sustainable Energy, 2023) more representative of the general population because it provides an in-depth longitudinal view of the DFS experience.

Although not representative of the GB population or of DFS participants overall, our sample does cover a mix of socio-demographic, technological and geographic characteristics. Participants were aged between 25 and 70, including five students, two retired couples, and ten professionals. In terms of geographical location, twenty participants were based in south-east England, two in Wales, and three elsewhere in England. Ten participants had rooftop solar, and five also had home-batteries. Seventeen participants used gas for heating and cooking, and two had heat pumps. Eight participants used technologies which allowed them to automate or remotely control appliances. We have provided detailed participant characteristics in Annex 2.

Given we recruited most participants from specialised interest groups, we believe they had above-average interest in and understanding of the energy sector and energy-related technologies. Reflective of wider DFS participation, most of our participants signed up to the DFS through Octopus Energy, one of the first energy suppliers to join the DFS. We therefore have little insight into the impact of the DFS beyond Octopus Energy’s customer base. Moreover, the sample does not cover participants with pre-payment smart meters (sometimes used as a proxy for energy vulnerability), and no questions were asked about participants’ income or vulnerability to avoid losing participants in the recruitment process. The diaries also reflect the perspective of the main participant rather than that of their entire household, albeit participants did reflect on the impact of the DFS on others in the household. We reflected in the discussion section the main impacts of these limitations.

Step 1: We started our analysis from Hargreaves and Wilson’s (2017) characterisation of the three domestication pathways they had identified (successful, precarious and rejection). Wilson and Hargreaves built their pathways largely on frequency and perseverance of engagement with smart home technologies and perception of and interest in using these technologies. In a similar vein, we looked at continuity and consistency of engagement with the DFS as well as perceptions of the experience, so we defined success in relation to the number of sessions opted in to and looked at whether DFS sessions developed into a fixture of the home e.g., how/if participants planned their actions to accommodate the sessions, how participants and other household members perceived the DFS. This is reflected in Table 2 below.

We also considered whether the kWh reduction in demand by each participant should weigh into whether they successfully domesticated the DFS. The DFS’ objective revolved around reducing peak demand in winter 2022–23, without prescribing what kWh shift in peak is preferred. Since DFS was open to residential users, we inferred that any volume reduction is relevant – otherwise the DFS would have applied only to the industrial and commercial sector which can traditionally achieve higher volumes of demand reduction. The ESO rewarded, thus valued any kWh reductions, no matter how small. DFS providers and the ESO also favour certainty, so a user who is reliably shifting demand away from peak, offers value via that certainty, even if maybe not via volume. Moreover, the DFS may not have provided these participants with significant financial savings, but most will have had an overall positive experience with the DFS e.g., by feeling like they are contributing to a societal goal, or using the experience for fun or family time as will be shown in Sect. "Findings". For these reasons, users who continuously shifted away from peak, even though the volumes they shifted were small, were qualified as having successfully domesticated the DFS.

Step 4: We considered outliers. Three participants successfully domesticated the DFS, but with outcomes that came into conflict with the spirit of DR, as we discuss in Sect. "Gaming Pathway". We created a separate pathway to reflect their experience. We found it difficult to fit this pathway in either successful or precarious domestication. To accommodate this ambiguity, we abandoned the idea of presenting the pathways as discrete and instead set them out on a spectrum from successful domestication to rejection. Annex 3 provides a visual representation and summary of how we defined our five pathways on this spectrum. We thus ended up with a total of five pathways which we detail below.

Our eight participants in this pathway had in common a desire to reduce CO2 emissions through the DFS. They were also aware of, and knowledgeable about, the need for flexibility in a low-carbon electricity system, although they did not all have specialised expertise in this field. For a few participants, environmental motivations went hand in hand with concerns for energy security and a sense of contributing to resolving a national problem.

All participants regarded financial gains as secondary benefits, and no one was deterred by small reductions in peak demand or low financial rewards.

Roberta, for instance, reported that ‘I’m not expecting to save much, but I’m keen to keep participating in the name of contributing to wider system benefits and seeing how it all works.’ Patrick noted he is motivated ‘mainly by contributing to a more efficient and decarbonised grid than [by making] financial gains;’ and Thomas reported a ‘desire to contribute to reduction in peak demand (and associated use of less clean energy sources).’

The strength of motivation of these participants made an imprint on all dimensions of domestication.

On the symbolic dimension, participants perceived reducing electricity demand overall, and shifting demand away from peak times as normal rather than exceptional behaviour. Jennifer, for instance, noted that:

We are used to adjusting our usage. I am part of a local energy scheme […] which charges a premium between 4pm and 8pm, and the cheapest tariff is 8pm to 7am. […] I have been committed to lowering our energy use for many years. I think having a variable tariff for all would make people think about when they use energy […].

Participants tended to give the experience a positive connotation. For instance, they found their participation in the DFS served as an opportunity to reconnect, framing the experience as ‘family time’ or ‘dinner by candlelight.’

Jessica, who has small children and a hectic schedule, used the sessions to bring the household together, reflecting the affordances of the DFS. She created a game in which the children used a phone light to find their toys, noting that 'the boys loved it and were sad when it finished.' Jessica also capitalised on the positive media commentary around the sessions for an educational opportunity: explaining to her children that their participation was helping to save the planet from ‘dirty coal.’

Their attitudes and resolve warmed other members of the household to the DFS even when this did not seem likely to happen. Roberta initially found it difficult to encourage her partner to switch off 'without nagging' and noted that ‘there is an interesting relationship dynamic to these trials’ when she discovered he had not turned off the items that he claimed he had. Thomas too initially found that when his wife oversaw making dinner it was ‘harder to get her to do the smaller things as a) she is less committed to the initiative b) she doesn't have the details.’ Jessica’s husband initially described the DFS as a ‘pain.’ All three partners however joined into DFS efforts as it progressed. Roberta’s partner because of her explanations, and Thomas’ and Jessica’s spouses in conjunction with the positive media messaging around the two consecutive January 2023 sessions publicised as essential for replacing coal generation on those two particularly cold winter days.

From a practical perspective, many participants already knew from experience what actions they needed to take during the sessions. Jennifer already knew which appliances to turn off during the events because they were the same appliances she would regularly avoid using at peak times. Sandra had already been using a smart plug on her refrigerator on a regular basis pointing out that ‘smart plugs will be of most use on appliances that are always on or very frequently used.’

These participants also found it easier to make small adjustments to their routines when needed e.g., working in a sun-lit room to avoid heating the house during DFS events, watching TV on their phone, or boiling water on the stove instead of the kettle.

It was the practical dimension – particularly the things outside of their control – that most tested their resilience. Jessica’s smart meter had an intermittent connection which caused her worry that her efforts are not accurately reflected in the savings: ‘I'm a bit concerned our saving hasn't been captured properly as the Octopus app is saying it can't connect to our smart meter this morning:(’ A few sessions later she wrote: I am annoyed that my smart meter has stopped working again and so I was given an average number of Octopoints.²’

Thomas found it difficult to accommodate longer sessions, sessions on consecutive days, and short notices whilst balancing work and the schedules of three small children. For most of the DFS he found creative solutions to accommodate these constraints e.g., he installed smart plugs, or texted his wife clear instructions and reminders of what to power off and when.

In November 2022, he wrote:

Having not had email from Octopus I signed up via the Hugo app, so I was able to take steps to prepare. Had I not seen via Hugo the Octopus email would have been too late to allow prep. […] I get all the kids at 5pm and then have to go straight home to cook tea (right in the window of demand reduction). I will try to start dinner at 6pm but that makes it hard to get them to bed on time. So, my prep has been to get a pre-made meal (homemade!) out of the freezer that can be microwaved + microwaved packet rice. I assume this saves at least some energy vs heating in oven / induction hob.

Sandra noted after one of the January events:

This was the first 'for real' (non-trial) event and it was well-publicised in several media, including online reporting during the event by the BBC. So I did pay a bit of extra attention, e.g. going to https://grid.iamkate.com/ to see if there was a visible change in demand […] and whether the ESO had had to bring on either of the two coal-fired plants that had been warmed up in readiness for the evening peak. No extra coal-fired power was needed […] which I felt good about. […]

Most participants in this group already had established ways of monitoring their usage (e.g., through mobile apps) and had above average understanding of how much electricity their appliances use. However, due to the calculation of the baseline which we discussed in Sect. "Overview of the 2022/2023 Demand Flexibility Service", they did not fare any better than all other participants at predicting their energy savings or rewards – which caused them disappointment. They also felt frustrated by the three- or four- days it typically took for their DFS provider to calculate the results of their saving session which made it more difficult to compare results between sessions and thus learn how to further reduce demand. Our only participant who switched DFS providers after a few sessions also spotted inconsistencies between the information received from the two DFS providers which caused him to question its accuracy,

The five participants in this pathway had in common the fact that they re-domesticated existing technology to provide DR via the DFS – either large shiftable loads (e.g., electric heaters), or a combination of generation and storage technologies (solar panels and batteries). They relied on automation more than other participants, but still planned their use of technology around the DFS sessions and created new routines to provide DR.

On the symbolic dimension of domestication, participants formed similar positive framings for the DFS like the previous group, with added enthusiasm around using existing technology in new ways. Mason developed what he framed as a ‘battery strategy,’ which involved checking the weather ahead of each session, charging the battery to an adequate level, and keeping an eye on consumption during the session. Phillip was able to enact a similar routine with apparent ease, framing it as ‘one minute’s work.’

Most of the time participants successfully provided DR without other household members even noticing. They enlisted the support of household members only when needed and mostly at the beginning of winter until they figured out how to minimise disruption for the rest of the household.

Philip, for instance, only asked for cooperation from the household once in November 2022, and took a measured approach on where to focus their efforts:

I was working in the office on the day. There was some sunshine on the day, meaning 1.8kWh of solar PV had been used to charge the battery by mid-afternoon. It was touch and go as to whether the battery would still have enough power by the time of the session, so I contacted my wife at lunchtime, and we agreed that she would cook dinner using the gas hob, avoiding our electric oven, electric pressure cooker etc. I also asked her to make sure that we weren't using the washing machine and dishwasher at the time. We didn't take any action with lighting and other electricity uses, as they are minimal, even though we have a large number of Phillips Hue lamps that can be controlled remotely.

The participants in this group had mixed motivations which carried little weight in predicting successful domestication. Henry was primarily motivated by financial gains: when asked after each session if he would opt in again, he noted ‘Yes, especially knowing financial incentives.’ Alisson too had some financial motivations since she attempted to estimate her savings based on what she thought her baseline was for each session and was motivated to participate again when she received significant rewards. However, she gave her rewards to charity via an Octopus Energy-led scheme and noted other motivations like maintaining her streak and because she ‘committed to the cause.’ Like most participants in the previous pathway, Phillip, Matthew, and Mason were already using little grid electricity at peak times, so gained almost negligible rewards, but this did not deter them from participating.

These participants too became warm experts in their personal or social media networks, but their framings were not always as positive as the group above. Mason, for instance, ‘shared saving with [a] friend on WhatsApp (also has solar panels so interested)’ and signed up to a new social media group to discuss his results. However, when Alisson discussed the DFS at work she ‘complained to a few colleagues about the baseline calculation algorithm’ after having gone online to understand why the outcomes of the first two sessions differed so much from her expectations.

Alisson tried various strategies to automate her smart heaters – which she had previously used without automation – tinkering with the GPS and the automated schedules functions on the app. She also tinkered with the smart light bulbs automations and re-defined their interaction with the Google Assistant to re-gain control over some functions that she had effectively delegated to the automations e.g., she de-activated the lights turning on automatically at dusk as this may have coincided with some of the DFS sessions.

For most sessions, Henry noted that he ‘switched off the ground-source heat pump for the hour, simple to do but needs to be done in the house and can't do remotely.’ On several occasions he came home from work early to turn it off. He also ‘bought some smart plugs as well on things which can easily be switched off.’

Similarly, Mason adjusted previous routines and tinkered with automations to accommodate DFS sessions:

Battery will have plenty of charge. We won't put on tumble dryer / dishwasher / washing machine until after 10am today. Probably won't use kettle / coffee maker either. Might turn off heater in conservatory (it's triggered by temperature). Set timer to remind me to turn heater back on.

Most participants planned their use of technology around the DFS sessions in quite a bit of detail. Phillip for instance noted:

The day before, I looked at the status of my battery and the weather - it was clear that there would be no sun and therefore no chance of powering my home from the battery, rather than the grid, during the session. So, I changed the settings on the Tesla app to allow grid charging and temporarily changed the backup reserve on my battery to 50% capacity. The battery therefore charged overnight to 50%. A few hours before the session, I changed the backup reserve level on the battery back to its normal level of 16%. This meant the battery would discharge from 50% to 16% and I was confident this would have more than enough power to cover the 5-7pm period.

For all the planning and automation, this group too found it harder to participate in short notice sessions so either sought to pre-empt the sessions with information from elsewhere (Alisson) or did not bother to accommodate the session at all (Henry and Mason).

Even with above-average understanding of and interest in technology and DR, these participants too found it difficult to predict their rewards. This is despite using technologies such as apps to review their consumption in near real-time, just like participants in the previous pathway. After the first two sessions, Phillip for instance found it:

Very difficult to predict, because I am not clear how Octopus calculate the reference consumption to compare to. This is complicated by my battery - if Octopus use times when we were using battery power or solar PV, rather than grid consumption, as the reference, then it will be really hard to demonstrate savings above this.

With little information available at the start of the DFS on how the baseline is calculated, Alisson noted she feels ‘scammed’ for having received fewer Octopoints in the second session than in the first one despite greater effort. After understanding that the baseline also accounts for in-day consumption, as discussed in Sect. "Overview of the 2022/2023 Demand Flexibility Service", Alisson noted how difficult she found it to translate this into practice and reconcile it with the messaging around the DFS.

We created a separate pathway for participants who, at least on a few occasions, increased their electricity demand in the hours running up to a DFS session (e.g., by charging a home-battery) to reap higher rewards. Borrowing the term from the media, we considered it gaming of the DFS when the increase was both artificial (i.e., participants otherwise would not have used electricity for that purpose at that particular time) and deliberate (i.e., they increased their usage before the session to increase their financial gains during the session). As discussed in Sect. "Overview of the 2022/2023 Demand Flexibility Service", the methodology used to calculate the baseline made gaming possible.

Our study included only three such participants, although the phenomenon generated discussion and debate on special-interest groups on social media, suggesting that it may have been common amongst smart-energy enthusiasts. Gaming, however, appears to require significant effort on behalf of individuals to understand and manipulate the DFS methodology, and to coordinate smart devices so we do not expect it to have been widely employed. A Centre for Sustainable Energy (2023) survey found 16% of respondents attempted to shift usage in the t-4 interval, though it is not clear how many of these did it with the intention to increase their DFS rewards.

William’s family branded the sessions ‘power hour’ and he even noted his ‘wife was a bit stressed about forgetting about 'power hour' on one occasion. Pierce ‘spent some time talking to the children about what we were doing today and the reason—reducing the need for three coal plants. Talked about pollution aspects of coal given high carbon to low hydrogen content.’

He completely changed tack for the next session, a January morning one when he noted:

Given my recent experience of finding out about the in-day adjustment, I was determined to maximise my rewards on this session despite not doing anything different during the actual saving session 9-10. At 6:30 am I put on the washing machine and made my breakfast. At 7am I made my lunch for the week. At 7:30 I hoovered up and put on the electric fire for 30 mins. I did all this to see if my saving session rewards would increase just for doing tasks between 5-8 am […].

He estimated this would be his highest saving yet – which it was – and posted the results on Facebook.

Driven by environmental motivations, Pierce had discussed at length with his friends and family the objectives of the DFS and even exported electricity from his battery into the grid during the sessions to provide additional support. In February 2023, however, he noted:

Reviewed DFS documentation and realised that additional usage between t-4 and t-1 is counted as a saving. Consequently, fully charged battery, heated home to higher temperature etc in this time.

He also noted he felt frustrated that only net imports were measured during this process which doesn't incentivise people to support with extra capacity,' and was ‘surprised by how poorly structured the formula was for calculating savings. It turns out this was very easy to game, which you used to compensate for exporting during the peak period.’

William’s notes were less explicit, but we could see his results spiking up in the second half of the winter. His rewards went from an average of c. £8 per session in the first part of the DFS to an average of c. £60 per session in the second half and earned a total of c. £260 for the winter, by far the highest total reward we have seen in our sample.

As in the previous pathways, he used smart controls to enable the routine: ‘I switched my air-sourced heat pump off via my smart thermostat app prior to the session from my desk in London (I set a reminder on my phone).’ Pierce too consistently noted how he ‘set battery to remain at full capacity prior to the session. Set battery to ensure net export to grid of 100w. Cooked meal before session started. Preheated rooms before the session.’

Gilbert noted how he ‘changed the heating schedule so that it started after the session. Put both sky boxes into deep standby mode. Ensured almost everything switched off and baseload = 22w. Charged my phone.’

Our findings thus show that participants in the gaming pathway domesticated the DFS successfully in the sense of forming routines, collaborating with household members, acting as warm experts, and learning from one session to the next how to better accommodate the DFS into their everyday lives. However, when looking at affordances, or the possibilities for action that they could choose from, we find that the actions they preferred may have run counter to wider energy security, affordability, and decarbonisation objectives.

For instance, in charging his battery from the grid instead of from his solar PV, Pierce artificially increased overall system demand and, to some extent, his own bill – although the latter was compensated for from the additional reward he made during the session. If enough users behaved in this way, the cost of the system overall would increase because more generators would have to come online to feed the additional demand. Emissions too may increase if the generators called upon are gas plants instead of renewables, whose ability to feed the additional demand will depend on the weather. Finally, if this creates a new peak in the interval preceding a DR session, it will have security of supply implications whilst the grid adjusts to the new peaks.

That said, participants who gamed the DFS did not necessarily act in a way counter to the DFS script which rewarded and thus prescribed demand reduction in a specified interval only (and not overall demand reduction). And, in any case, domestication theory does not view successful domestication as contingent upon acting in accordance with the script. Based on this reasoning alone, we have considered the gaming pathway a subset of successful domestication, like the two pathways above.

We noticed however that other participants in our study, when pondering on their affordances, made different choices which accounted for wider values than just the DFS script or their personal gain.

For instance, Roberta in our first pathway, noted:

I noticed that when I’d been using more energy in the previous days, I could save a bit more when it came to participating in the DFS. […] This created a small incentive to use more energy, overall – but I resisted because that goes against my values, and (honestly) the small £ amount I seemed to save after each event was not enough to assuage any guilt for having deliberately used more energy…

We thus believe that the way in which Gaming participants domesticated the DFS, whilst successful from a pure domestication perspective, is not quite in alignment with the spirit of DR. We discuss the implications in Sect. "Flexibility justice".

Most of the seven participants in this pathway were not energy experts and were less likely to express environmental motivations behind their initial decision to take part, and less likely to have domesticated the technologies which made participation easier for those in the second and third pathways. These participants were more likely to express financial motivations, and the pathway is characterised by declining levels of commitment, especially when they perceived the financial reward to be out of balance with their effort. We associated this with precarious domestication of the DFS.

On the symbolic dimension, Nora’s comment succinctly summarised a typical attitude, which is that ‘other things got in the way.' For instance, participants chose to opt-out if guests came over, to avoid social awkwardness – as opposed, for instance, to Sandra in the first pathway who boiled water for her guest’s tea on a gas stove instead of a kettle.

Nolan noted that he ‘unexpected[ly] had guests over during the savings period, so had on more lights, used the heating, and appliances when I should have been saving.’ Scott had family staying over but didn’t tell anyone about because ‘the in-laws were being very helpful doing our cooking and laundry, and I didn’t want to appear ungrateful by asking them to NOT do those things for an hour!’.

Their contributions to peak demand reduction were in fact, not negligible because, as discussed in Sect. "Data Analysis", the ESO derives value from any reduction, regardless how small. Their results from the individual sessions in which they tried to reduce demand were comparable with those obtained by participants in the first three themes. However, even when they participated in DFS to reduce carbon emissions or contribute to energy security, their exclusive focus on financial reward as a proxy for success, resulted in feeling demotivated.

As such, most participants concluded their lifestyles are not conducive to effective participation in DR. Reasons included a tendency to be away from home during peak times, or because they rarely conducted energy-intensive practices (e.g., laundry) during peak hours. This is again, a result of how the baseline was calculated to account for demand during the previous ten days at the same time as the DFS session.

A few things however did keep these participants engaged. Most prominently, the two January 2023 sessions that made headlines for being key in keeping coal off the grid motivated participants to try anew.

Maxwell wrote:

This time […] communication from Octopus Energy clearly encouraged people to participate in the exercise as coal-powered energy plants were needed to rebalance the grid if demand was not reduced. This aspect of communication certainly motivated us to make an extra effort.

Angela too ‘got the message about this session being really key and the reward being higher so just before the start I went round and turned off all the lights and all appliances except the fridge.’ Norah was additionally motivated by ‘maintaining a streak’ and Angela found that delegating responsibility for the sessions to her partner is another way to maintain participation in the DFS. Neither of these however were sufficient to sustain commitment throughout the winter, though interestingly, follow up conversations with these participants showed that all but Maxwell signed up to the DFS in the second winter.

They made little or no effort to convince other household members to partake in the DFS and where they did talk about the scheme, they were unlikely to try to persuade others or advocate for it.

When they ran into an obstacle, they made little or no effort to overcome it. Nolan, for instance, a young tech-savvy participant, said he missed most sessions because he was not receiving push notifications on his smartphone, but had made no deliberate effort to address this, even once he had realised it.

Despite a strong start in Maxwell’s household where all four flatmates.

Switched off as many electric devices as possible (no lights, heating in all rooms). Ensured no device was on stand-by. Cooked tea shortly before. Charged battery of laptop to a sufficient level beforehand. Only devices which stayed connected [were]: fridge/ freezer, Wifi, and TV on standby (plug was unreachable sitting behind a bookshelf).

By the second session Maxwell already noted:

Personally, I was not home. Housemates decided not to participate again (two days in a row seems to be too much disruption/ interference with daily routines, especially during that time of day). One housemate has to attend a virtual Zoom meeting requiring lights, monitor and laptop charging >> important limitations!

Scott too started out enthusiastically and for the first session he noted:

Turned off all but a few lights during session. TV and computer were the only things on. My daughter was at home so needed the TV. Avoided making tea during session. Unplugged laptop charger (although I knew this would be miniscule). Checked IHD and felt frustrated not to be able to get usage below 430w. Everything I was aware of was turned off.

He did not delegate responsibility for the sessions he was not at home for and having eventually noticed no rewards for any of the sessions, he gradually disengaged without having built any routines.

For the first session, Angela noted:

I felt quite positive about the prospect of getting a discount on my energy bill by using less energy, as well as helping the grid. Just before the session started, I went round and turned all electrical items, switched off sockets, except the fridge (which I switched to Economy mode). I turned off all the lights except one.

Only Angela ‘dug into her smart meter data’ and concluded that she was not making any savings ‘because these times of the day aren’t usually when I use a lot of electricity.’ Most participants came to the same conclusion intuitively but did not look to validate it or improve the outcomes. Unlike previous participants they did not seek additional sources of information, and they did not compare outcomes of each session against the actions they had taken.

This pathway only had two participants, which was expected given the sample bias and the fact that the analysis only included those who returned their diaries at the end of the DFS. Like participants in the ‘Not Worth the Effort’ pathway, the two were not energy experts and were mostly motivated by financial gains, albeit one did also have an interest in reducing carbon emissions. They made little or no effort to domesticate the DFS and expressed mostly negative feelings about the experience, so we associate this group with the rejection of the DFS.

Reflecting on her overall experience, Grace noted:

During the energy saving sessions this winter I was going through an unnecessarily complicated gas supplier switch which impacted my desire to join in with these sessions/made it much harder!

Sebastian pondered on his struggles to receive email notifications of the sessions without being the account holder:

It’s designed for individuals – the account holder only gets notified but the whole family need to be involved for it to work. Getting notifications was way more of a hassle than I had predicted.

This extended to musings on how DFS might impact family life:

I’m not always home at the times…so it’s kind of patriarchal… I imagine more men are account holders (would be good to know stats) and more women are homemakers, so the man decides to save money at work but it’s the poor person at home who actually has to make the sacrifice.’

He went on to reflect his resentment towards Octopus Energy ‘I feel like I just want to boil a kettle unnecessarily to make the little Octopus cry,’ and government:

I dislike the personal responsibility [in that] it’s up to me to use less power and ignore the fact the UK gov has a shit energy security plan and hasn’t invested in renewables.’ How much did Greg Hands [i.e., a former energy minister] save on those nights…eh?

Like most other participants, Grace and Sebastian also reflected on the imbalance between financial reward and effort: ‘I like the theory behind it but saving £4 wasn’t worth the hassle’ (Sebastian); ‘I didn’t save a huge amount of energy and so the points didn’t create a huge amount of money off future bills and so the incentive for me wasn’t that high’ (Grace).

We did not get a sense from their diaries that they had discussed the DFS beyond the household.

Grace and her partner missed the first DFS session because she ‘thought it would be hard to complete as we would both be at home.’ She then joined the second session and wrote:

Neither of us was home so it made sense to join this session. Did not use oven/lights/any appliances during this time. I charged my work and personal phones at work. I kept the heating as it is gas powered and as Octopus aren't currently supplying the gas, this didn't affect the session. I didn't think to turn off sockets.

For all other sessions, she wrote that she either forgot or missed that they were happening.

We found that even within our relatively small sample of households, engagement with the DFS was surprisingly varied, driven or enabled by factors such as motivations, access to specific technologies, prior exposure to DR, or family dynamics. This allowed us to identify five distinct pathways for domesticating the DFS which we believe could apply to the wider population, although we do not expect a similar distribution across the pathways to the one in our sample (where most participants were successful in domesticating the DFS), since our participants probably had above average DR capabilities (Roberts et al., 2020) or flexibility capital (Powells & Fell, 2019). In a sample more representative of the wider population, we would expect to see much lower occurrence of successful pathways.

In terms of what enables successful domestication of DR, we noticed that our participants motivated primarily by financial gains were successful only if they believed those gains were worth the effort (e.g., if they had large enough loads) and they tended to be easily disheartened otherwise. This aligns with findings from Bradley et al. (2016) who conclude that perceived barriers to providing DR depend on the financial incentive received. Because domestication theory enabled us to track continuity of participation in DR, we confirm findings in existing research that expectations of financial savings are significant motivators for participation in DR initiatives (White & Sintov, 2018), but also find that they are insufficient for sustained participation. We find that in our sample, the successful domestication of the DFS is largely enabled by non-financial factors, namely motivations (particularly environmental), prior experience of DR, and the availability of DR-enabling technology.

These results corroborate research which indicates that perceptions of social and environmental responsibility outweigh financial motivations as determinants of successful participation in DR (Buryk et al., 2015; Strengers, 2010) We saw that our participants who followed precarious or rejection pathways tended to be less well-informed about the system benefits of DR, while those committed to the DFS were aware of the growing importance of demand-shifting to reduce greenhouse gas emissions and system costs. Additionally, we found that our participants with strong environmental motivations were more likely to continue providing DR even in adverse circumstances e.g., low rewards, frustration with perceptions of injustice. They were also more likely to frame their experience of the DFS in positive terms and act as warm experts and advocates of DR; whilst also finding innovative and ingenious ways to domesticate DR, constituting, as discussed by Nyborg (2015) a source of unrecognised potential for delivering decarbonisation.

In the context of normalising DR, our findings reveal two distinct pathways for successful domestication: manual (first pathway), and automated/technologically-enabled DR (second pathway). This aligns with Darby and McKenna (2012), who theorise the many forms that residential DR can take, from manual to Direct Load Control. There is value in harnessing both types of DR to avoid exclusions (i.e., households which, at the moment, can only provide manual DR) in the nascent rollout of residential DR in GB. Even if the contribution of manual DR to demand reduction is relatively low at present (first pathway), its contribution is likely to become more significant as householders increasingly switch from gas to electric heating and adopt electric vehicles. Rewarding manual DR helps users form routines and learn from experience what strategies best fit their lifestyles. It also prevents exclusions and risk of lock-in to automated-only DR at this early stage in the DR journey. Incentive-only schemes such as the DFS are well suited to normalising DR amongst householders, especially compared to punitive scheme designs which risk generating resistance (Nicolson et al., 2018).

Our results on motivations and prior experience of DR highlights the value of raising awareness and normalising DR amongst households. The DFS received wide news coverage as a novel intervention in the electricity system, but there is a need to raise awareness using broader informational or educational campaigns, and to develop a broader range of initiatives that can enable householders to take part in DR. We believe it worthwhile to raise awareness of the benefits of DR and provide arguments and evidence in favour of demand shifting programmes that warm experts can use in their interactions with friends and family. Our results also show the importance of providing prompts for such discussions to take place – many participants noted that they talked to friends and family about DFS in the context of it being in the news.

Besides general promotional campaigns, there is a need to manage expectations about the potential for energy and financial savings. As discussed by Strengers (2016) energy consumption statistics cannot reflect the complexities of household practices in a satisfactory manner. Even amongst our early-adopter participants, knowledge of the measures that will have most impact on demand reduction was limited, especially in the Not Worth the Effort and Disillusioned Pathways. Providing clear, timely and personalised feedback on DR sessions is essential. Combining energy and financial results with personalised suggestions for impactful actions is likely to reduce the risk of frustration by participants, resulting from the unrewarded effort – as exemplified by Gilbert’s experience.

Additionally, the potential to earn £100 was widely promoted, but few participants came even close to that. Most of the savings we saw in our sample were under £20. Knowing that the potential for savings is a significant motivator for signing up (White & Sintov, 2018), scheme designers must also take care not to generate unrealistic expectations, which may in turn lead to defection as we saw particularly in our fourth pathway.

How best to raise awareness of DR and improve users’ capabilities in this space is, however, an open question. Relying on energy suppliers alone can result in uneven spread and potential confusion especially where a household selects a DFS provider other than their energy supplier. For instance, the one participant in our study who switched DFS providers questioned the accuracy of the information he was receiving because it differed between the two companies. Relying on DFS providers could also be problematic, if they prioritise cheaper communications methods (email, social media) which risk excluding less digitally literate users. We thus note the risk of users disengaging because of confusion and/or excessive information as found by Christensen et al. (2020) and Öhrlund et al. (2019).

Octopus Energy, the largest DFS provider, rewarded participation with Octopoints, which many of our participants struggled to understand despite self-reporting relatively high levels of energy literacy and capabilities (Roberts et al., 2020). Many also felt frustrated with the delay in feedback which prevented them from learning how to improve outcomes between sessions. With 24 DFS providers in the first year, fragmentation in terminology, explanations, and rewards was inevitable. Looking forward, it may be necessary for scheme designers to coordinate or even regulate the provision of feedback from third party providers, building on evidence of effective messaging and engagement processes.

Another enabler of successful domestication in our sample was access to technology, be that large loads or automation. Our findings contribute to the flexibility justice debate. We can infer from the data we have on participants in our Tech-Enabled Success Pathway that they likely sit in the upper right quadrant (i.e., more financial resources, more flexibility capital) on Powells and Fell’s (2019) generalised representation of the interaction between flexibility capital and financial resources. As their research predicted, we noticed that technologically-derived flexibility capital (a subset of flexibility capital) in the form of batteries, solar PV, and smart controls meant these participants mostly did not need to sacrifice convenience to provide DR. We saw, for instance, that some of these participants did not need to involve household members in the sessions, whereas domesticating the DFS for most of our other participants meant negotiating, convincing or collaborating with household members.

Interestingly, access to relatively little flexibility capital in our Motivation-Driven Success Pathway did not hinder domestication, but it did so in our Not Worth the Effort Pathway. This again relates to the discussion above about motivations. We also noticed, however, that many participants in these two pathways had little flexibility capital because they were already using little electricity at peak times. In our sample, this was mostly because they were already on a time-varying tariff, or because their lifestyle meant they were not home at peak times. However, in the wider population we might see this because of the cost-of-living crisis or other socio-economic factors leading users to decrease overall electricity demand. DR, as we discussed in the introduction, is valuable to the system, but many users may have less flexibility capital because they have already reduced as much as they could. This means they get less benefit from participating in programmes like the DFS which reward only per kWh (i.e., they only value the final result) instead of also rewarding participation and consistency (e.g., routine formation) which are important for normalising DR. Rewards do not always need to be financial, and we found for instance that streak-framing worked well in keeping some of our participants motivated.

In the Introduction, we presented the case for DR in the transition to a decarbonised power sector. In our analysis of the Gaming Pathway, we noted that the way in which participants understood and interpreted the role of the DFS in this transition had an impact on how they domesticated it. Our takeaway is that, at least for participants with above-average understanding of, or interest in, energy and/or decarbonisation, the DFS script created confusion by not always aligning with wider societal values and system objectives. In some cases, this confusion correlated with behaviours that led to an increase in overall demand – in particular, where batteries were involved. Moreover, it created frustration for several of our participants and the wider population (as reflected in the media). This may have wider impact on public perception and future participation in DR programmes, as acknowledged by the Centre for Sustainable Energy (2023) study. Perceived injustice risks undermining both users’ appetite for DR and political will to support DR policies (Roberts et al., 2020). Gaming behaviour on a wider scale could also risk diminishing the indirect benefit that all users would be expected to get from peak demand reduction (Uddin et al., 2018). Our evidence suggests that some ‘gamers’ share the concerns over fairness. Targeted messaging around the negative consequences of gaming therefore could be explored to address this issue on the symbolic dimension in case setting less gameable baselines proves difficult.

One of the reasons why the DFS script and the overall implementation of the DFS caused some confusion and frustration is likely due to the multitude and variety of actors influencing the DFS whose agendas may not have entirely aligned e.g., the ESO, government, DFS providers, energy suppliers, user groups etc. The ESO, for instance, tested short notice periods and consecutive sessions which put pressure on and frustrated many of our participants. Like Nicholls and Strengers (2015), we found that the evening peak coincides with a ‘family peak’ of activities. Our participants found it hard to navigate this peak of activities at short notice. It is understandable from the ESO perspective that shorter notice periods are preferable, but we noticed in our sample that this has had a negative impact on our participants and may not be the best way to keep users engaged in the longer term e.g., by enabling routine formation. We thus suggest that paying attention to recognitional justice by ‘asking whose interests are accounted for in the design of the intervention and whose are ignored’ (Winther & Sundet, 2023) will be important for future DR programmes. Like Powells et al (2014) we find that DR changes the relationships between various system actors and users, and we find that without further exploring this there is risk of misalignment between imagined users and actual users’ daily lives and contexts.