Marketing demand-side and time-of-use programs to reluctant customers

# Marketing demand-side and time-of-use programs to reluctant customers

In November 2022, Octopus Energy began paying households to switch things off at a named hour, on a day's notice. Its Saving Sessions ran inside National Grid ESO's Demand Flexibility Service: customers with a smart meter reporting half-hourly reads opted in, dropped consumption below a forecast baseline for a window of about an hour, and collected OctoPoints worth several times the retail price of the units they did not use. Hundreds of thousands of households took part. (Octopus sells both retail tariffs and the flexibility products discussed here.)

The enrolment figure is the least interesting part. Look at the ask. Nothing about who supplies you, nothing about your contract, only when you run the dishwasher. Choosing a supplier is one decision made once, on the triggers the foundations lesson describes. This is a request to rearrange a household routine and keep it rearranged on a wet Tuesday in February.

Why these programs stall

Utilities push three related products at customers who did not ask for them:

• Time-of-use (TOU) pricing: electricity costs more during high-demand hours (often late afternoon and evening) and less overnight.
• Demand response (DR): customers agree to cut or shift usage during peak events, usually for a bill credit or a smart thermostat.
• Efficiency programs: rebates for insulation, heat pumps, LED lighting and appliances.

All three ask for short-term inconvenience or perceived risk in exchange for a benefit that is delayed, invisible or uncertain.

The reluctance is rational, and the measured effects are modest. Across two decades of dynamic pricingdynamic pricingAutomatically adjusting prices in real time based on demand, competition or user behaviour to optimise revenue, margin or conversion.View full definition → pilots, a plain TOU rate with no enabling technology tends to move peak demand by a low single-digit percentage. Sharper event-based prices do better. Rates paired with a device that acts on the customer's behalf do several times better again. That finding should decide your creative brief: you are selling automation, not willpower.

Diagnose the real objection

Before designing incentives, name the barrier. Resistance falls into five kinds:

1. Loss aversion. People fear a higher bill more than they value an equal saving. A TOU rate that could save money still reads as a threat.

2. Effort and habit. Shifting laundry to 9pm is a daily chore nobody volunteered for.

3. Distrust, the category-level problem the trust lesson anatomises, which here attaches itself to the meter and to the phrase "dynamic pricingdynamic pricingAutomatically adjusting prices in real time based on demand, competition or user behaviour to optimise revenue, margin or conversion.View full definition →".

4. Invisibility. Nobody watches kilowatt-hours the way they watch a fuel gauge.

5. Nothing to shift. A house with an EV, a heat pump or a battery can move several kilowatt-hours a night. A one-bed flat with a fridge, a kettle and a laptop cannot, whatever the messaging says.

The fifth is where campaigns quietly waste money. A transparency problem does not respond to a larger rebate, and a household with no shiftable load does not respond to anything.

Design the incentive around loss aversion

The core finding from Daniel Kahneman and Amos Tversky is that losses hurt roughly twice as much as equivalent gains feel good. Two design moves follow.

Use bill protection on new rates. When customers move to a TOU rate, guarantee that their first year will not cost more than the old flat rate. This removes the downside, so trying the rate stops feeling like a gamble. Several US utilities used this guarantee during default TOU transitions and measurably reduced opt-outs.

Frame the peak event as a reward. A "Peak Time Rebate" that pays you for cutting usage during an event outperforms a "Critical Peak Price" that charges you more, even when the arithmetic is identical. Same money, opposite emotion.

Now the counter-example, because the rule is not absolute. EDF's Tempo tariff in France has been sold for decades and is unapologetically penalty-shaped: 300 blue days, 43 white and 22 red in a year, with the red-day peak rate several times the blue-day off-peak rate. It survives because the sanction is legible (one colour, announced the evening before) and because much of the response is automated by relay-switched water heaters rather than by discipline. Its failure mode is structural: red days cluster in cold snaps, when electric heating is hardest to give up, so the cost of non-compliance lands on households with the least room to move. Any penalty-shaped rate needs a hardship path built before launch, not after the first February complaint wave.

Choose the enrollment model: opt-in vs opt-out

This is the biggest single lever, and it is a marketing decision disguised as a policy one.

• Opt-in: customers must actively sign up. Enrollment is low, often single digits to low teens as a percentage.
• Opt-out: customers are enrolled by default and can leave. Enrollment stays high, frequently above 80 percent, because inertia works in your favour.

Regulators increasingly allow default TOU precisely because opt-in adoption is so weak. The catch is that defaults move the failure downstream: you can hold 85 percent enrolled and still shift almost no load, because most of those people never noticed the rate changed. Enrolled and inert is not a win, and it is expensive to unwind once complaint volumes reachreachThe number of unique people exposed to your message in a given period. Unlike impressions, reach counts each person once, no matter how often they see it.View full definition → the regulator.

The classic behavioural framework is worth internalising. See the UK Behavioural Insights Team's EAST framework: make the behavior Easy, Attractive, Social and Timely.

Make it easy: automate the behavior

Asking people to change habits daily loses. Asking them to enroll once and automating the change wins.

The smart thermostat is the model: the customer enrolls, the utility nudges the setpoint a couple of degrees during an event, the customer can override and mostly does not. A daily decision becomes a single sign-up. OhmConnect built a consumer business on that logic in California, giving away smart plugs and thermostats, paying for reductions against a forecast baseline and wrapping the whole thing in lottery-style rewards while aggregating the homes into a virtual power plant. (OhmConnect sold demand response to households; its business was folded into Renew Home, a venture involving Google's Nest Renew, in 2024.) Two things to take from that arc: gamified rewards do pull people into an event, and the margins in consumer DR are thin enough that acquisition costacquisition costCustomer Acquisition Cost (CAC) is the total sales and marketing spend divided by the number of new customers gained in a period. It measures how efficiently you grow.View full definition → per household has to be judged against the kilowatts you can credibly bid.

One engineering detail with marketing consequences: if every enrolled thermostat releases its setback at the same minute, the compressors all restart together and you have manufactured a second peak. Stagger the release, or your load-shift number evaporates in the reporting.

Make it social and timely

Social proof moves energy behavior. The best-documented case is Opower (now part of Oracle), whose home energy reports compared your usage with similar neighbours. Peer comparison produced average reductions of roughly 2 percent, small but persistent across millions of homes.

Use it carefully. Telling efficient customers they are below average can nudge them back up. Pair every comparison with an approving signal (a smiley, "Great") for the already-frugal so you do not undo your own work.

Timely beats generic. A text alert on the morning of a hot day outperforms a brochure mailed in March. Amber Electric in Australia passes half-hourly wholesale prices straight through for a monthly subscription and pushes a notification when the spot price spikes or turns negative, then sells automation that dispatches a home battery so the customer never has to read the alert at all. (Amber sells wholesale-exposed tariffs.)

Fix distrust with radical transparency

If the barrier is trust, no incentive overcomes it.

Octopus Agile makes transparency the product: next-day half-hourly prices published every afternoon, including occasional negative periods where customers are paid to consume. That legibility is also the risk. Exposure cuts both ways, and through the 2021-22 wholesale spike Agile's per-unit cap was raised repeatedly while cap-protected standard tariffs looked cheap by comparison. Amber's version of the same risk is starker: the administered price cap in Australia's wholesale market sits above A$15,000 per megawatt-hour, hundreds of times a normal retail rate. Sell the downside in the acquisition creative, or you will pay for it in churn and complaints.

Concrete moves:

• Show customers their own dataown dataData collected directly from your own customers and prospects through your own channels: your most reliable and privacy-compliant source.View full definition → in plain language. "You used the most power between 5 and 8pm" beats a rate table.
• Be explicit about privacy: what is collected, who sees it, that it is not sold.
• Show earnings near real time. "You earned £4 in credits during yesterday's session" closes the invisibility gap.

Segment, because one message fails everyone

Segment on shiftable load first, motivation second.

• EV and battery households: the highest-value segment by far, since one overnight charge dwarfs everything else in the house. Lead with automation and scheduling.
• Cost-motivated renters with a dryer and a dishwasher: lead with pounds and the bill guarantee.
• Values-motivated homeowners: lead with grid impact ("shifting your usage helps avoid firing up the dirtiest peaker plant").
• Vulnerable and fixed-income customers: lead with protection, and coordinate with assistance programs so a dynamic rate never becomes a hardship.

Same program, four campaigns.

Sequence the rollout

Failures come from bad sequencing more often than bad offers. A defensible order:

1. Check the plumbing. No half-hourly data, no program. Britain's smart meter rollout is still incomplete a decade in, which means a large share of the base is ineligible before marketing spends a penny.

2. Educate before you price. Give customers months of usage insight first. Let them see their own peak.

3. Run a shadow bill. Show what they would have paid on the new rate versus the old, with no money moving. People self-select, and the offer stops sounding like a trick.

4. Default them in, with protection. Opt-out enrollment plus the first-year guarantee.

5. Automate, remind, then report. Devices, event alerts, visible credits and peer comparisons to hold engagement past the novelty phase.

Skip steps 2 and 3 and you end up defending meter accuracy in public, when the customer's actual grievance is the surprise.

Measure what actually matters

Enrollment is a vanity metric. What counts:

• Load shift net of rebound: kilowatts moved out of the peak, minus the snapback in the hour after.
• Retention: who is still enrolled at 6 and 12 months.
• Complaints and opt-outs: your trust signal, and the one regulators read.
• Cost per avoided kilowatt, against what peaking capacity would have cost.
• Baseline integrity. Paying against a forecast baseline rewards anyone who runs the dryer in the hour before an event. Watch pre-event consumption; if it is drifting up, you are paying for theatre.

Run pilots as controlled experiments with a comparison group, so you can attribute results to the program rather than the weather. The US Department of Energy publishes evaluation guidance through its Smart Grid resources, a solid free starting point.

Key takeaways

• The failure is behavioral, not technical. Meters work. Programs stall because nobody planned the habit change or the disclosure.
• Sell automation, not willpower. Plain TOU moves peak by a few percent; a rate plus a device that acts for the customer does several times better.
• Use the default, then check it did something. Opt-out plus a first-year bill guarantee beats opt-in, but enrolled and inert shifts no load.
• Frame gains, and if you must use a penalty, make it legible and attach a hardship path. EDF's colour-coded Tempo works that way; its red days still land hardest on the least flexible homes.
• Front-load the downside. Wholesale-exposed rates like Agile and Amber's win on transparency and lose badly when the customer meets a spike they were never warned about.