Two Seasons, Two Businesses: The Plan I Made in June and What I Did in October

• 92 per centshare of winter import that fell in the free midday window
• 0.014 kWhhighest single-hour meter import in week one, against a 0.03 limit
• 1.65 kWhoven energy counted against export, zero import

In June I planned to change electricity plans twice a year. On 20 October I changed once, to a plan that pays me $1 a day if I buy almost nothing from the grid between 6 and 9 pm, and on 23 October I ran the oven for 61 minutes inside that window to find out whether I’d lose the dollar.

I run a 19.4 kW solar array and a 48 kWh battery on a suburban Melbourne house, and everything here comes from that house and its meter. It is one household’s arithmetic, shared because the method might be useful. Your roof, your habits and your numbers will differ. This is personal experience, not advice.

Two businesses on one meter

The house doesn’t change between January and July. What changes is which way the energy flows. When the panels make more than the house uses, I am a net exporter: over a stretch of time I send more to the grid than I take. When the sun is low and short, I am a net importer. Same wires, same family, two different businesses taking turns.

The billing meter shows the handover. Net flow by month, in kWh: December +613 exported, January +520, February +237, March +31, April +22. Then the flip: May −529 (529 kWh more bought than sold), June −918, July −916, August −628. September came back positive at +163, and October to the 25th is +508.

Two things in that run matter. March and April sit almost at zero, so “exporter from October to April” is too tidy: the export half has a fat end and a thin end. And in June I’d assumed I’d be an importer from May to September, but September was already exporting. The pattern holds; the edges moved.

What I planned in June

The two businesses want different plans. A wholesale pass-through plan, where what I pay and what I’m paid follow the live wholesale price, rewards having something to sell, so it suits the surplus half of the year. A fixed free-window plan, a block of free or cheap electricity at the same hours each day, rewards buying, so it suits the deficit half.

My June model put running each half on the plan that suits it at roughly $140 to $200 a year ahead of the best single plan held all twelve months. That was a modelled figure, specific to this house, and it depended on how much of a wholesale plan’s theoretical value my own automation could bank. The simple plan sat in the half of the year where that skill matters least.

What happened

The winter half went to plan. I accepted OVO’s Basic Free 3 offer on 8 June and was on it two days later. It gave me a free window from 11:00 to 14:00, and I charged the battery from the grid in it. From 10 June to 30 September, 92 per cent of what the meter recorded as import fell inside that window. Most of my winter import was free power bought at midday and stored for the evening.

The summer half did not go to plan. I signed up with GloBird for its ZEROHERO plan on 17 October, and the switch completed on 20 October. I did not go back to Amber or any other wholesale plan.

What I decided on, and what I checked afterwards

What I recorded when I signed was short. I would not join the retailer’s virtual power plant, which means letting the retailer direct my battery for the grid, so the battery stays under my control. And that control stays with Home Assistant, the home-automation software that runs my battery. Behind that sat the work I had finished the day before: the plan I was on had closed to new customers, so leaving it was a one-way move, and on my first objective, buying nothing at a price, a ZEROHERO-based plan was one of only three that came close.

The detailed checks came the day after I signed. They did not hand me a clean answer.

The cooker decided the ranking. The induction cooker sits on the grid side of the gateway, the box that connects the battery and panels to the house, so the battery cannot serve it. On the model as it stood, with the cooker counted against me in the evening test, the cheapest option was still two plans: Flow Power’s 4Free plan for summer, then ZEROHERO, at $629 a year. ZEROHERO all year ranked fifth, at $694. If the meter counted the cooker against my export, ZEROHERO all year came out at roughly $418 and ahead, though that figure was not confirmed by a full re-rank. Rewire the cooker onto the battery side and it drops to $297. The model could not tell me which case applied. That needed real evenings.

Diagram to come

The wholesale summer bought power at a price. Depending on the model, the wholesale summer bought between about 86 and 360 kWh at a price. ZEROHERO’s summer bought about 2.

The dollars were close. When I replayed ten months of Amber’s own forecasts through a controller I built, it banked about 70 per cent of what knowing every price in advance would have earned. At that level the wholesale summer plus a ZEROHERO winter was about $140 a year cheaper in cash than ZEROHERO all year. But it pushes about 2.1 MWh a year more through the battery, and once I priced that wear at 8 c a kWh, the gap flipped to about $90 the other way. Within about $100 either way, then. The year I tested had the calmest evening prices in the record I had, and on earlier years’ prices the wholesale summer came out further ahead.

A fixed schedule is easier to run. ZEROHERO’s terms: free grid electricity from 12:00 to 15:00 (fair-use limit 50 kWh a day, averaged over the billing cycle), $1 a day for importing under 0.03 kWh an hour between 18:00 and 21:00, and 10 c per kWh on the first 15 kWh exported in that evening window (GloBird 2026). Simulated against that fixed timetable, my battery reached 96 to 100 per cent of the perfect-foresight result. On a wholesale plan I measured about 70.

So the model did not pick one plan for me. I chose on control and on buying nothing at a price, and the next day’s checks left the two options close and the cooker as the swing.

Week one on the meter

The new logic went live on 23 October. Three evenings now stand on the billing meter, and all three pass the credit test, the $1 condition above.

The highest single hour was 0.014 kWh, against a limit of 0.03. The 24 and 25 October evenings peaked at 0.001 and 0.000 kWh. On 23 October the oven drew about 1.65 kWh from the grid side over its 61 minutes, and the meter registered zero import in every half-hour it was on. The oven’s energy came off what I was exporting at the same moment: the meter counts the cooker against my export. Cooking again on the 24th, once the evening export had started, also registered zero. On 21 October, before the new logic was live, cooking inside the window had failed the test.

Two things did not go cleanly. On Saturday 24 October I started cooking at about 17:00, with the battery full and not enough sun to cover the cooker, and bought 0.267 kWh at 44 c, about 12 c, in the half-hour before my evening export starts at 17:30. And on 25 October, a cloudy Sunday after a full export evening, the midday free charge was 49.9 kWh at the meter, almost exactly the 50 kWh daily fair-use figure. That limit is averaged across the billing cycle, so one day at 50 is not a breach, but a run of grey days after full export evenings would push the average towards it, and so far I have only three days of history. The first evening’s export also stopped early, at 11.57 of a 13.03 kWh target, because the guard that keeps enough in the battery for the night assumed the house’s overnight use is spread evenly, when most of it comes early in the evening. That is the safe direction to fail in, and it cost about 15 c.

What this leaves me with

The insight from June stands. A solar-and-battery house really is two businesses, and the meter shows the handover in May and September. What did not survive was the conclusion that the response is two plans. The winter plan I’d meant to return to had closed to new customers, the wholesale summer bought power at a price, and a single fixed plan I could simulate, and now measure, came out within about $100 of it and well ahead on my first objective.

The takeaway I’d offer is about method. If your house flips between seller and buyer, rank plans on your own meter, and find the one appliance that breaks a plan’s rule before you trust the ranking. For me it was the cooker, and it moved the answer from two plans to one.

What I haven’t tested is winter. Everything above is spring, with plenty of sun. By next September I’ll know whether a winter of evenings passes the same test, and the cooker rewire that would stop that afternoon purchase hasn’t happened yet.

The spreadsheet doesn’t care what you hoped the answer was. In October it said the gap was small enough to choose on other things, and I chose the plan I could run myself and check on the meter.

References

GloBird (2026) How ZEROHERO works [GloBird app screens, transcribed by the author]. Accessed 19 October 2026.