
A seasonal energy companion, living inside the Octopus app.
British winters are when energy anxiety peaks. Octopus Energy has the smartest tariffs in the market: agile pricing that rewards households for using energy at the right hours. But that intelligence reached people as tables and graphs. Intelligence without experience is a spreadsheet.
The brief we agreed on was sharper than any feature list: turn winter anxiety into winter control. Without gamifying people's bills. Without punishing anyone. We did not rebuild the app. We built a seasonal product inside it, then operated it through the winter.
Here is how a winter became a product.
We did not set out to make people use less energy. Octopus's tariffs already do that arithmetic. We set out to change how winter feels. That reframe wrote the product's constitution: speak in money and comfort, never in kilowatts.
Two constraints were fixed on day one and never renegotiated. A bill is not a game, so no mechanic may trivialise it. And no household may ever be punished for having a hard week.
Constraints like these only hold if the research backs them. It did.
We read a winter's worth of support transcripts, interviewed households across tariff types, and watched how usage of the live app changed as temperatures dropped. Three patterns shaped everything that follows.
Three findings became three layers.
An assistant fed by spot prices, weather forecasts and household patterns. It speaks in money and comfort, not kilowatts: start heating 30 minutes earlier, ease off at 10pm, same comfort, £4.20 stays in your pocket. Octopus's famously human tone carries into the conversation design. The assistant is not a feature. It is the product's voice. That is what we mean by AX Design.
The signature craft is failure-state design. Forecasts miss. When one does, the copilot does not apologise, it corrects: the miss, the reason, the fix. And every suggestion carries a why. The screen below is live. Open the working yourself.
Every peak hour a household shifts becomes a star. Stars form constellations across the season. An anonymous neighbourhood savings league adds gentle social proof, and in March a shareable recap closes the winter: 214 kWh shifted, £31 saved, your brightest star was in January.
One decision defines the whole layer, and we state it openly: no streaks. Miss a night and nothing breaks, because punishing people who worry about their bills is bad design and worse ethics. Try the sky below.
Wintermode opens in October and closes in March. Deep night blues, warm amber, magenta saved for the moments that matter. Data visualisation warmed up: pounds in pocket, never kWh charts. Cognitive load kept low: one insight per screen, one action per insight.
When spring arrives, the world folds away until next year. Software with seasons feels alive, and a seasonal product is a promise to return.
Shipping in October was the midpoint, not the finish.
The product stayed in the loop through Corexi, our continuous UX monitoring. We do not publish client numbers here. We publish what we watch and why it matters. The share of consumption shifted off-peak, because that is the product doing its job. Weekly actives through the winter months, because a seasonal product must earn every week. Recap share rate, because pride is a metric. And support call volume, because a copilot that works answers questions before they are asked.
In March the recap shipped and the world folded away. By then the data had already written the backlog for winter two: which suggestions people accepted, which corrections rebuilt trust fastest, which stars they cared enough to share. A seasonal product is a promise to return. We intend to keep it.
Wintermode's thesis fits in one line: honesty is a feature. When the model missed, the product said so, corrected itself, and kept the household's trust. The overall feeling we shipped is the one the research asked for: a warm room in a cold winter. Control, not anxiety.

