Demand side response events in Spain

See analysis plan.

Between January and May 2026, Octopus Energy (OE) will run five Spain “Saving Sessions” events, when the price of the electricity in the day ahead market is deemed extremely high. Events will last 1 hour.

For each event, participating customers will be randomly assigned to one of six groups:
1. Control group: no notification or payment
2. Low payment: customer’s unit rate (approximately €0.12/kWh)
3. High payment: double the customer’s unit rate (approximately €0.24/kWh)
4. Green message: Ask customers to reduce use for environmental reasons, no payment
5. Prepaid reward: Ask customers to reduce as much as possible; notify them that, as a thank you in advance, they will receive a free hour of electricity on Sunday at, e.g., 14:00.
6. Generous prepaid reward: As above, but two free hours of electricity on Sunday, e.g., 14:00-16:00.

Customers will be randomized into one of the above groups for four events. All customers will experience the same treatment for a fifth and final event.

In addition, we will include a separate group of customers who are never treated in any event; this pure control group will serve as the control group for the fifth event.

Those selected are automatically notified via email prior to each session, with instructions on the time window and the reward offered. They are encouraged to shift their electricity usage, such as running the washing machine, charging their electric vehicle, or heating water, away from the designated session period. After each session, they receive a summary of their usage and rewards.

Customers in the financial incentives group will be remunerated for consumption decreases below their baseline consumption. We define baseline consumption as the mean of the 10 (4) most recent working (non-working) days, at the same hour of the day as the event.

Sample: We will access a pool of 108,163 customers who are not:
- Self-employed
- ATR code (in Spanish, Acceso de Terceros a la Red) different to 2.0TD
- Clients who opted-out from marketing communications
- Vulnerable clients
- Clients with open complaints
- Clients who have 80% of their smart meter data in the last 2 months
- Clients enrolled in the Solar Family or Sun Club campaigns (for more information, see: https://octopusenergy.es/blog/sun-club)
- Clients who are enrolled in Intelligent Octopus Tariffs
- Members of the “Comunidad Octoamigos” (for more information, see: https://octopusenergy.es/blog/programa-referidos)
- Clients whose agreement with OE is to be renewed with 60 days after the first session starts
- Clients whose main address is on the Canary Islands
- Clients in distribution zones that are not able to send telemetry in less than 5 days

Note that If a customer switched to another retailer before an event, their consumption for that event will be excluded. If the switch occurred after an event, their consumption for that specific event (and only that event) will be included. Additionally, customers will be entirely excluded if we lack the relevant meter consumption data when conducting our analysis. We predict that these types of attrition are unlikely to be related to the treatment.

2026-01-30

2026-05-30

Associated CUPS (in Spanish,Código Universal del Punto de Suministro or universal supply point number) electricity consumption (import kWh only) at the hourly level during “Saving Sessions” events, in kWh

Consumption during Saving Sessions event hour, measured at the customer level

See analysis plan

See attached analysis plan for details.

Matched pairs randomization; in office by a computer; see analysis plan for details.

Customer email address

Yes

108,163 clusters, where regressions are clustered at the customer email address level

For 108,163, if each event is one hour, with a total of 5 events, this results in approximately 432,795 observations in the main analysis, clustered at the customer level (108,163 clusters).

Pure Control: 9,999
Control: 16,360
Green: 16,361
High: 16,361
Low: 16,361
Prepay: 16,361
Prepay2: 16,360

We conducted simple power calculations of our trial design using electricity consumption data from Spanish Octopus Energy customers. We randomly selected five “event dates” between November 1 and December 15, 2025, and used customers’ actual observed consumption during these events. The sample consists of customers who will participate in our Saving Sessions, with approximately 16,000 customers assigned to each treatment arm. We exclude the pure control group from the analysis, which we believe makes our power calculation conservative. With this consumption dataset, we estimated the same regression model that will be used in the final analysis, regressing electricity consumption on treatment indicators and control variables, as specified in Equation (1). We applied the rule of 2.8, under which the minimum detectable effect size (MDES) at 80% power and a 5% significance level is equal to 2.8 times the standard error of the treatment coefficient. Using this approach, we estimate an MDES of 0.00789 kWh per hour for each treatment arm, corresponding to approximately 1.4% of average hourly consumption during event periods.