Crowdflex: 2025 Summer Utilisation and Availability Trials

CrowdFlex is an innovation project aimed at exploring how domestic flexibility can be harnessed to support the management of the electricity grid. It involves social science trials to better understand consumer behaviour and how households respond to incentives to adjust their energy use. While previous services have explored domestic flexibility, CrowdFlex rigorously tests domestic flexibility through randomised controlled trials (RCTs), examining variables like incentive structures, notice periods, times of day, and event durations. This rigorous causal identification will help policymakers and other stakeholders to assess the role domestic flexibility can play in supporting the UK’s goal of achieving a net-zero energy system by 2030.

Building on the insights from CrowdFlex’s 2024 Summer and 2024-2025 Winter trials, CrowdFlex will conduct Summer trials from July 2025 through September 2025. The two main trials are the Summer Utilisation Trial and Summer Availability Trials. The utilisation trial will use a randomised controlled trial (RCT) methodology will continue to be employed to ensure rigorous evaluation of the outcomes, allowing for robust comparisons between different consumer groups and behaviours. The availability trials will involve a matched control group and randomised treatment groups.

The rewards for participants over the trials will align with payments that could be obtained for the volume of response from stacking of demand turn-up and turn-down markets, services and related capacity payments. The exact reward structure will differ between trial groups to generate evidence on customers’ sensitivity to differing incentive levels and structures. The trial is designed to investigate domestic demand response performance and classify response by key flexibility characteristics. This includes both turn-up and turn-down events.

These trials are designed to provide behavioural insights and learnings for demand side response service providers, particularly in understanding how customers react to price signals and domestic flexibility. The data collected will also inform decision-making in the Control Room regarding grid management at various times of the day and across different seasons. This information will offer a deeper understanding of system challenges, such as peak demand periods, network constraints, and the potential of distributed domestic assets to provide effective balancing solutions.

The Summer Trials are designed to assess domestic demand response performance by testing both availability and utilisation payments. The study will categorise responses according to key flexibility characteristics, including both turn-up and turn-down events.

Different trial groups will be created to test:
* The impact of availability payments on availability and response performance
* The impact of direct utilisation payment vs other forms of rewards on response performance
* The impact of behavioural prompts (availability) and notice periods (utilisation) on response performance

OVO Energy will conduct a trial of each type, while Ohme will focus on an availability payments trial only.

Randomisation in the trial was conducted with stratification by Grid Supply Point (GSP) applied throughout. Stratification ensured balanced representation of geographic areas across trial arms, accounting for potential regional variation in grid conditions and customer characteristics. It was done using a python script with a seed.

OVO utilisation payments trial: 76k customers - about 4,200 customers in each of the 15 treatment arms and notice period and 12,600 in a control group, all assigned using stratified randomisation at the GSP group level.

OVO availability trial: 9,069 participants who were randomised into one of the three treatment groups. 3,173 customers in the matched (not randomised) control group.

Ohme availability payments trial: 18,971 customers randomised into 4 treatment groups using computer randomisation. 5000 customers in a in a matched (not randomised) control group.

External Link(s)

Registration Citation

Citation

O'Donnell, Jim et al. 2025. "Crowdflex: 2025 Summer Utilisation and Availability Trials ." AEA RCT Registry. September 19. https://doi.org/10.1257/rct.16560-1.0

Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Experimental Details

Interventions

Intervention(s)

Customers will receive different incentive levels / structures.

Intervention (Hidden)

See analysis plan.

Intervention Start Date

2025-07-07

Intervention End Date

2025-10-05

Primary Outcomes

Primary Outcomes (end points)

Availability Payments Trial: Weekly plug-ins
Utilisation Payments Trial: kWh consumption

Primary Outcomes (explanation)

Availability Payments Trial: Primary outcome: weekly plug-ins by trial group.
Utilisation Payments Trial: Primary outcome: half-hourly electricity consumption (kWh) during event windows.

Secondary Outcomes

Secondary Outcomes (end points)

Availability Payments Trial: kWh consumption during event windows
Utilisation Payments Trial: kWh consumption outside event windows

Secondary Outcomes (explanation)

Availability Payments Trial: kWh – used in analyses to investigate the extent to which utilisation-like performance (kWh turn-down and turn-up) also varies by trial group – i.e., the extent to which higher plug-in time does, in fact, deliver more genuinely useful availability.
Utilisation Payments Trial: Assess spillover effects of treatments on electricity consumption outside event windows, measuring half-hourly consumption in the 24 hours before and after the events. Spillovers may be positive (e.g., habit formation) or negative (e.g., shifting consumption).

Experimental Design

Parallel-group RCT

Experimental Design Details

See analysis plan. We have three separate RCTs, two administered by OVO and one administered by Ohme.

Randomization Method

Randomization done in office by a computer

Randomization Unit

Customer

Was the treatment clustered?

Yes

Experiment Characteristics

Sample size: planned number of clusters

OVO availability payments trial: 9k customers in the randomised treatment arms, 3k in a matched control group.

Ohme availability payments trial: 19k customers in the randomised treatment arms, 5k in a matched control group.

OVO utilisation payments trial: 76k customers in the randomised treatment arms and a control group.

Sample size: planned number of observations

Analysis is clustered at the customer level. We have hundreds of repeated observations for each customer: * Observations are at the half-hourly level * number of half hour in events: * Utilisation Payments Trial: approximately 12 turn-down (but only one arm has turn down, so 2*13.5k customers) and 48 turn-up events * Availability Payments Trial: over 100 turn-down and over 100 turn-up events * Events are 1-4 hours

Sample size (or number of clusters) by treatment arms

OVO availability payments trial: 9k customers in 3 treatment arms 1. Standard, no prompt, 2. Standard with prompt amd 3. Standard with incentivised prompt. Each treatment arm is further divided between short, medium and long notice periods. 3k in a matched control group.

Ohme availability payments trial: 19k customers in in 4 treatment arms: 1. Standard, no prompt, 2. Standard, with prompt, 3. Enhanced, no prompt and 4. Enhanced, with prompt. We are comparing the treatment arms between them. Additional 5k in a matched control group.

OVO utilisation payments trial: 76k customers in the randomised treatment arms of 4,200 and a randomised control group of 12,600 customers. The groups are: 1. Direct £/kWh turn up and turn down (only group to have turn down events), 2. Direct £/kWh (turn up only), 3. £/kWh + community pot, 4. Free elec (volumetric), 5. Free elec (consistency) and 6. Control.

Minimum detectable effect size for main outcomes (accounting for sample design and clustering)

See analysis plan. Utilisation Payments Trial: 0.003 kWh, or approximately 3% of typical control group consumption per half-hour. Availability Payments Trial: A change in plug-ins per week of 5-7% of the control group average.

Supporting Documents and Materials

IRB