The Feed-in Tariff, Properly

• +21 c/kWhnetwork step in the buy price at 15:00, every day
• 83 %less daily grid import after solar than before it (27 → 4.5 kWh)
• $142net cost over 130 days on the wholesale plan, as modelled

There is a chart on my wall now. It is the half-hourly feed-in price I was paid for exporting solar over 130 days, plotted against time of day, with the sign corrected so that up is money in my pocket and down is money out of it.

It looks nothing like the chart most people imagine. It is not a flat 6 cents. It is not even always positive. From roughly 10:00 to 14:30 it sags toward zero and sometimes plunges negative: the grid paying me to stop. From 15:00 it lifts into a peak window between about 16:30 and 21:00 where the average sits in the high single digits and the spikes go to 30, 50, sometimes 180 cents. After 21:00 it collapses back to nothing.

Everything else follows from that chart.

---

1. What the feed-in tariff actually is

I had this wrong for months. The first draft of this post did too.

The feed-in tariff on a wholesale plan is wholesale-only. When my solar exports a kilowatt-hour, I am paid the wholesale spot price for that half-hour. No network charges are added on the way out. Amber’s invoices show a single “Solar Exports” line and no network adjustment, and the four invoices reconcile to my half-hourly data to the cent. Over that window I exported about 2,070 kWh and earned $157.

The buy price is not symmetric with that. What I pay on import is wholesale plus network charges, environmental certificates, hedging, market charges and retail margin. Network alone adds about 10–12 c/kWh off-peak and roughly 21 c/kWh more in the 15:00–21:00 peak block. That +21 c step at 15:00 is the largest feature in the buy-price data. It snaps back at 21:00, and it is the same every month.

The data has a sign trap. Amber’s API shows a credit to me as a negative number and a charge as a positive one, which is upside-down for a homeowner. Every chart here negates it. My first draft built its central argument on the wrong side of zero.

So forget the “6 cents flat” model. The feed-in price varies by half-hour, has spikes, and goes negative often enough to matter for any control strategy.

---

2. Spike capture is the whole game

Of the $157 I earned over 130 days, $97, sixty-two per cent, came from 205 half-hours in which I exported at a price above 10 c/kWh. That is about 3.3 per cent of the half-hours earning nearly two thirds of the revenue. Seven of them were super-spikes above 30 c/kWh, and those seven alone earned $20.

The long, flat midday belly of the chart pays close to nothing, and sometimes less than nothing. The money is in the evening shoulder between 16:30 and 21:00, when wholesale climbs and a battery that has sat full since lunchtime can dump 20 kWh into the grid at premium rates.

I missed the biggest spike of the dataset. On 27 January 2026 the price briefly hit $1.86 and my battery did nothing, because I had no automation to react. That is a gap, and the target for the next round of automation work.

You are not optimising for export volume. You are optimising for export timing. A 10 kWh battery dispatched into seven minutes of a price spike is worth more than the same battery trickling out across a sunny afternoon at 2 c/kWh.

---

3. The curtailment story (and the 5 April failure)

The flip side of “the price goes negative” is that the system has to know when to stop exporting. When wholesale crashes through zero, usually in the middle of a sunny, mild day, every kilowatt-hour out the meter costs me money. Not earning less. Costing.

The automation watches the price, and when it goes below zero it drops my export limit from 15 kW to a 100 W trickle. When the price recovers, normal export resumes.

I wanted to know whether this is worth doing. I curtailed an estimated 1,000 kWh of solar over the window. The net financial impact was about minus $1. I saved roughly $22 by not exporting during negative events, and lost roughly $21 by curtailing at the edges of those events after the price had turned positive.

That is not the result I expected to write. The automation is doing its job and the result is break-even. It correctly refuses to pay the grid to take my electrons without bleeding revenue from the positive intervals either side. It is duller and more reassuring than a story where it prints money or burns it.

Then there is 5 April 2026.

On that Sunday afternoon Home Assistant restarted in the middle of a deeply negative price window. The startup behaviour cleared the curtailment helper, and the trigger was supposed to re-evaluate straight away and put the export limit back down. It did not. It was a numeric_state trigger, which only fires on a change across a threshold, and the sensor came back from the database already below zero. No crossing, no trigger. For about an hour the inverter exported 8.7 kWh into a grid paying between minus 4 and minus 8 cents.

The incident cost about 50 cents. It mattered far more than that. It was the first time my own automation moved real money the wrong way, and every threshold-triggered automation in the stack had the same gap. The fix is a baseline re-evaluation block plus a five-minute polling fallback, and it is deployed on the curtailment automation. The matching gap on the opportunistic-charge automation is still on the list. I am keeping this incident in the numbers rather than smoothing it out. The point of this series is to show what happened, including where I lost money to my own bug.

---

4. Wholesale versus the alternatives

In April 2026 the offers on my network included several alternatives to a wholesale passthrough plan.

The baseline

For my network (AusNet, postcode 3806) the regulated Victorian Default Offer for 2025–26 is $1,908 a year at the 4,000 kWh reference usage. The cheapest single-rate market offers landed around $1,500 a year, about 21 per cent below it. For a household whose battery already shifts evening load out of the peak window, the cheapest time-of-use plans are cheaper still, around $1,400 a year.

So the fair comparison is not wholesale against the Default Offer. It is wholesale against the best fixed plan a careful shopper can land on.

The comparison

I ran my actual half-hour import and export through each tariff: same kWh in, same kWh out, same timing, over 130 days. The five comparators:

1. 1st Energy Total Plan Saver. The cheapest single-rate plan on the network: 26.50 c/kWh, 5 c flat feed-in.
2. AGL Solar Savers. The mainstream capped plan: 8 c feed-in on the first 10 kWh a day, 0.04 c above the cap.
3. Engie premium. A higher headline rate: 11 c feed-in on the first 8 kWh a day.
4. 1st Energy time-of-use. The cheapest two-rate plan: 35.60 c peak, 18.21 c off-peak.
5. A fixed-rate free-window plan. The only structural sibling to wholesale: 15 c feed-in on the first 10 kWh exported in the evening peak, free import 11:00–14:00, and a $1/day credit if peak imports stay near zero.

The result

Plan	Net cost	Wholesale wins by
Wholesale (actual)	$142	n/a
Fixed-rate free-window plan	$158	$16
1st Energy time-of-use	$184	$42
1st Energy single-rate	$204	$62
Engie premium (11 c capped)	$252	$110
AGL Solar Savers (8 c capped)	$258	$115

The mainstream capped plan costs $115 more over 130 days than running the same household on Amber. The summer is flattering, though, and the seasonal table below shows why I would not annualise it.

The cap on a headline 8 c feed-in is doing nearly all the damage. My system exports an average of 15.9 kWh a day. The first 10 kWh gets 8 c. The remaining 5.9 kWh gets 0.04 c, which is nothing. On the AGL plan that is $58 of feed-in revenue against Amber’s $157. The cap takes $100 out of my pocket. Engie is no better: the cap is tighter, so more leaks out at 0.04 c.

Wholesale also wins on import. Import cost on wholesale was $100. On the cheapest fixed plan it was $150, for identical kWh. That $50 gap is the value of buying at the wholesale-linked rate in the solar middle hours and shoulders, when the fixed plan is still charging 26.50 c. Where the battery cannot cover load, whether early morning, the bottom of an overnight discharge, or the cooker on the grid side, wholesale delivers kWh at 8–12 c.

Why this collapses in winter

The 130-day average hides the seasons, and the seasons are the most important thing here. Amber’s monthly advantage in dollars:

Month	vs 1st Energy single-rate	vs AGL	vs Engie	vs 1st Energy ToU	vs free-window plan
Dec 2025	+$45	+$71	+$69	+$29	+$22
Jan 2026	+$35	+$54	+$52	+$29	+$20
Feb 2026	$0	+$7	+$6	−$2	−$6
Mar 2026	−$14	−$15	−$15	−$12	−$13

December and January carry all of Amber’s advantage and more. In February the time-of-use and free-window plans were already slightly ahead, and in March every comparator beat wholesale by $12 to $15 for the month. The summer peak does all the work.

So whatever wholesale does for me in December, I cannot assume it does in July, when there are no battery-full middays and little surplus for the evening peak. Whether it still wins depends on whether I can catch enough cheap morning-trough intervals to charge the battery for the evening. I do not know yet.

What the free-window plan tells me

The free-window plan is the only alternative that comes close, about $90 behind wholesale. It does the same fundamental thing, paying the household to deliver into the evening peak with a battery covering the import side. The difference is the price model: a fixed 15 c capped at 10 kWh in a defined window, against whatever the spot price is with no caps and no windows.

Its $1/day peak credit paid out on 25 of the 130 days, about one in five, and that is $25 of its $158. Without it, the plan loses to the cheapest time-of-use plan.

For my household the choice is not mainly about money. It comes down to three things:

1. Flexibility. On a strange day, such as a heatwave, a spike outside the plan’s window or a negative midday, wholesale captures it and the fixed plan does not.
2. Forecastability. The fixed plan is easier to budget and easier to explain to a household member who does not care what wholesale is doing this afternoon.
3. Volatility. Wholesale exposes you to both sides of the market. The 5 April incident shows the bad days move real money. The fixed plan caps the upside and the downside.

There is one thing I have not modelled that could change the answer. The free-window plan includes free grid import from 11:00 to 14:00. In summer that is dead weight: the battery is already filling from solar. In winter it inverts. On a Melbourne overcast day the battery may not fill from solar alone, and free grid in the solar shoulder means topping up to 100% at zero cost ahead of the peak, instead of paying about 11 c overnight. That could turn the window into the headline feature of the plan, and it is the biggest reason I want winter data before calling wholesale the right answer year-round. The fact sheet sets eligibility rules I need to read carefully before banking on it.

I am on wholesale because the modelling work I am doing on this house only makes sense against a market-linked feed, and because the summer margin is too large to leave on the table for a household willing to automate. Without the rest of the project, the free-window plan would be the rational pick. I will reassess after June and July.

---

5. What this means for the household

Stripped to one principle: the value is in avoiding the peak buy, not in export revenue.

I can show that with a number, because my network metering data goes back two years and contains a clean experiment I did not know I was running. Three regimes fall out of it.

In the gas era, to July 2025, daily grid import averaged 17.5 kWh. Gas carried heating, hot water and the cooktop.

In the all-electric, pre-solar months, August to November 2025, gas was gone, six HVAC zones and the heat-pump hot water were live, and the solar and battery were not yet commissioned. Daily import jumped to 27 kWh, about 10 kWh a day more than the gas era. That is the cost of full electrification with nothing to offset it.

In the post-solar period, late November 2025 to March 2026, daily import collapsed to 4.5 kWh, an 83 per cent cut. The peak-hour share of import fell from 33 to 25 per cent, which is the battery deliberately shifting load out of the 15:00–21:00 block. Same household, same loads, same retailer, same meter.

The peak block is where the money lives. Before solar I imported 8.8 kWh a day in peak hours. Now it is 1.2. At a peak buy price of about 32 c/kWh that is roughly $2.50 a day, or about $900 a year, saved on the peak block alone. Add roughly $600 a year on the off-peak block and the system removes about $1,500 a year of grid import before counting a cent of feed-in revenue.

Two caveats. The post-solar window is summer and early autumn, so winter import will climb back, and I do not know how far. And the $1,500 is annualised from four sunny months. It is a number to think with, not to bank.

That gives me a hierarchy to optimise against. Peak buy on this network is around 40–55 c/kWh once everything is added up, so a kilowatt-hour the battery covers in that window is worth that much in avoided import, plus whatever I would otherwise have been paid to send it out. The same kilowatt-hour exported at midday earns 1–2 c, or is curtailed at zero. That asymmetry is the entire game.

The first priority is self-sufficiency: cover the household load from on-site generation and storage, so as little as possible crosses the meter at peak prices. The second is covering my fixed costs, about $50 a month, which summer spike capture handles easily and winter will need deliberate arbitrage for. The third is paying for the battery’s eventual replacement: exports beyond fixed costs must clear the cycling cost before they count as profit. Anything left is pure profit. There is not much yet, and there does not need to be.

Spike capture, curtailment, peak-import avoidance and the chart on the wall all sit inside that hierarchy. Curtailment can be break-even and still be right because it serves the first two tiers without costing the third. Spike capture is the headline because it is the only lever that earns revenue at the scale needed to fund tier three. And the 5 April incident matters more than its 50 cents because any failure in the automation layer threatens all four tiers at once.

The next post is about the battery as a peaker plant: what it does in those 16:30 to 21:00 windows, and what the data looks like when it does.

---

Sources

• Plan comparison: public comparison ranking for the AusNet network and the plans’ fact sheets, as at 7 April 2026.
• Victorian Default Offer: Essential Services Commission, 2025–26 final decision.
• Everything else: my smart-meter network data, Amber’s half-hourly price and usage data, four of Amber’s invoices, and my own automation logs.