Reading time:
Energy retailers must segment to survive
The electrified home is just some segments in a trench coat
Article written by
Julie Radu
Back in the heady days of 2023, rooftop PV customers were treated like the dream segment. Green! Sticky! Premium tariff! Featured prominently on the homepage next to a stock photo of a smiling family! Two facts now complicate the family picture:
Basically a non-trivial share of the kWh the lovely PV households exported last year cost energy retailers money.
You priced this on a flat or net-metering tariff that assumed the opposite. A flat-rate PV household exporting at midday in June, into a negative-price hour with balancing costs stacked on top, is (and I say this with love) a position you would not voluntarily take if you saw it on a trading screen. You'd back away from the screen and ask who was working that desk.
You wouldn't. But your Marketing team did. Last quarter. They called it the 'solar customer acquisition campaign' and they hit their target.
The point isn't that PV customers are bad (they're great and we love them). The point is that customer acquisition decisions are now portfolio decisions, and you cannot make portfolio decisions without the data Trading already has. Which, by the way, is also the data Marketing already has. They just don't talk to each other. They probably don't even Slack.
How we got here
On 1 October 2025, the EU's day-ahead electricity market switched from hourly to 15-minute trading intervals across the Single Day-Ahead Coupling. Add in intraday and balancing markets (which had already gone 15-minute, like the punctual kids in class) and you get one structural fun fact:
The wall between your Retail business and your Trading desk no longer makes economic sense.
For thirty years that wall did a great job: trading bought wholesale and hedges and retail sold flat tariffs to households while an internal transfer price absorbed everything in between. Trading worried about MWs. Retail worried about NPS. They sat on different floors, reported to different leadership members, and had remarkably different opinions about what 'a customer' was.
In 2026, every input that used to live with Trading (weather, market price, settlement granularity, balancing exposure) is now an input to a customer-level decision. And every customer-side asset (the EV, the heat pump, the PV, the home battery, the slightly suspicious crypto-mining rig in the attic) is now a position in a portfolio your Trading desk would otherwise be buying or selling externally.
What the data layer needs to look like now
To make any of this work, you need one view of every household built from inputs both sides need:
Smart meter electricity and gas, at 15-minute or better. Hourly data is now the data equivalent of paying with a cheque. You can still do it but people will stare.
Sub-meter signals where available. A meter that distinguishes the EV charger from the heat pump from the household base load turns asset disaggregation from informed guessing into knowing.
Home profile data. Building age, heating system, occupancy, the boring bits. This is what explains why two identical load curves come from physically different houses and why one of them is one cold snap away from googling 'heat pump grant Sweden'.
Weather and forecast. For separating 'this customer is leaving us' from 'it got cold.' Without weather, every churn model is just a thermostat in disguise.
Tariff and market price, both sides. The whole point of customer intelligence is knowing which customers' tariffs are tragically misaligned with what you're paying for their consumption hour by hour.
Ground truth data. Asset-detection models are only as good as the labels they trained on. Most retailers have terrible ground truth on which of their customers own EVs, heat pumps, batteries, or PV.
The output is a per-household, per-quarter-hour cost-to-serve curve that three teams use three different ways.
How to stop buying hedges from yourself, and other modern energy retailer hobbies
Trading. Forecast the portfolio's aggregate flex by quarter-hour. The MW you can shift in your customer base across the next 96 intervals is a position. Yesterday that position was invisible (buried inside the gross load forecast) so the trading desk over-hedged the peak and someone else monetized the spread. The corporate equivalent of paying for your own drinks at your own bar. Today you can harvest it internally, if Customer Intelligence gives you a number you can take a position on.
Sales. Stop running churn campaigns on contract anniversaries. Run them on asset events. The week a customer first looks like they're charging an EV is the week to pitch a smart-charging tariff. Eleven months before renewal, when no competitor is in the inbox. The asset shows up in the meter data weeks or months before the customer thinks to mention it.
Marketing. High-value segment used to mean high consumption. Now it means high and well-timed consumption. A 12,000 kWh customer whose load piles into peak hours costs you more to serve than an 8,000 kWh customer whose EV charges at 3am. Marketing's job is to acquire the second household at scale. This kills the lookalike audience you're running on Meta. Sorry to the lookalike audience.
Marketing, Sales, and Trading walk into a bar. They each order separately
The 15-minute settlement made the average household economically extinct as a unit of analysis. A retailer running Marketing, Sales and Trading on three different data layers (and viewing their book in hourly averages) is paying for that fragmentation in balancing, churn, cost-to-serve, and missed upsell. All three. Simultaneously. Every quarter-hour. Forever.
You already have the meter data. You almost certainly don't have the layer that turns it into intelligence the three teams can use without a translator.
That's what the Eliq Customer Map is. One view of the household, eight inputs in, three teams out, no translator required.
Julie Radu
Article written by
Julie Radu
