How big should your home battery be? First the type, then the size

Four hundred and seventy-one euros

Take a pro forma invoice for a fixed home battery. We leave out the supplier. The pro forma carries three numbers: 18 kWh of storage, 10 kW of power, and € 8,470 including VAT, with installation already included.

Divide the total by the storage: € 8,470 divided by 18 kWh is € 471 per kilowatt-hour.

The most expensive plug-in battery in our class comparison costs € 468 per kilowatt-hour. In the video we round that to 470. Between a cabinet an installer fixes to your house and a cabinet you put in your house yourself, the gap per kilowatt-hour is a few euros.

Two viewers asked nearly the same question in the comments. Which home battery fits my house? What type, and how big, with my house and my contract?

Whoever starts to research that gets stuck on everything it depends on: the roof, the contract, the number of phases, the car that may come, and whether the lights must stay on when the grid fails. The thought is then that you need an expert. Until there is one, you take what feels safe. A brand you know, and rather too big than too small.

Doing the maths yourself is no guarantee either. Our own calculator on this site used a curve until 13 September 2026 that sat nine percentage points above the published source at 10 kWh.

The order in this article is the order in which you choose: first the type, then the coupling, then the control, then what premium costs, and last the size.

Which type powers your whole house

The class article lists six categories, from power bank to fixed home battery. For the question here, the first two drop out. A power bank and a power station do not power your house. Two types remain.

The plug-in, categories 3 to 5. It powers your house through the consumer unit. You connect it in three ways:

1. at 800 watts on a circuit you share with the rest of your house;
2. on a dedicated circuit;
3. on a fixed connection up to 4,000 watts, and a certified installer lays that connection.

The line between categories 3 and 4 is the MPPT: with its own solar inputs, your panels go straight into the battery. Category 5 is the Plug-in XL, 7 to 8 kWh and 3.5 to 4 kW behind an ordinary Schuko plug that passes 800 watts out of the box.

Fixed, category 6. The plug comes off, the installer wires it in, and it is grid-forming: if the grid drops, it makes its own grid and your house keeps running.

The two bands for fixed differ. The class comparison holds the price of individual models in August 2026. What the video names is an editorial estimate. How that band was built is not recorded, and it is not a count. The pro forma at € 471 falls below the model band and inside the second band.

What you ask of the battery chooses the type

Which of the two you need follows from what the battery must do for you. Add an electric car, or a heat pump, or a few air conditioners. Then your connection becomes the limit, and fixed wins.

If you also export a lot from your roof, fixed wins again. Three things the plug-in does not have yet in September 2026: a DC charger for the car, the higher power levels, and peak shaving at the level of a fixed system. If your house runs on ordinary appliances, your demand is small and a plug-in on a dedicated circuit is enough.

What I think myself, as an opinion and not as a measurement: an installer would rather sell fixed, because that is where his margin is. And a plug-in that is fully specified and sits on a dedicated circuit now reaches comparable options, cheaper, without an installer and without the overhead that comes with one.

Within the two types, storage does not make the difference. The pro forma showed that. The difference is in three things: how the battery connects to your house, who controls it, and what else it can do.

How it connects to your house: AC, DC or hybrid

How does the power from your roof get into the battery? Along three routes.

AC. The battery hangs on the same grid as your house, through the consumer unit. Nothing changes on your panels or your inverter. You rebuild nothing. Plug-in and fixed can both do this.

DC. Your panels go straight onto the battery, through an MPPT. That means changing your installation. Plug-in and fixed can both do this now as well. With the plug-in, it has mostly been a few panels per MPPT so far, not whole strings. That gap is closing.

Hybrid. One inverter for panels and battery together, with an AC connection to the consumer unit. Next to that, such a system has an MPPT input for panels, or a backup port where you can connect a generator, for example. On this site we use the word "hybrid" only for that DC-coupled inverter, such as a Growatt SPH, a SolarEdge Home Hub or a Sigenergy system. A wired plug-in such as a Zendure SolarFlow Mix or an Anker Solarbank Max AC is AC-coupled, even when the manufacturer writes "hybrid".

One exception to know before you start on DC. On an installation with SolarEdge optimisers, there is no DC retrofit from another brand. EcoFlow lists SolarEdge as a brand on its incompatibility list, and SafeDC holds the string at 1 volt per optimiser as long as the SolarEdge inverter does not take part. DC coupling there means a SolarEdge inverter.

Are there panels already that you do not want to touch? Then you already know your route.

Who controls it

Who decides what the battery does, and when?

That starts with your contract. The contract sets how many jobs a kilowatt-hour from your surplus has.

On a fixed-price contract, energy tax and levies are already in your price per kilowatt-hour. Power you use yourself is then worth what you would otherwise have paid for it, tax included. That kilowatt-hour has one job: to end up in your own house.

On a dynamic contract, it gets a second job. It can be stored, or sold. In 2026 net metering (salderen) still runs. From 1 January 2027 net metering stops. For an exported kilowatt-hour you then get at most the market price, and the law sets a floor: until 2030 at least 50% of the price you pay for electricity without energy tax and VAT. On a dynamic contract, that compensation follows the spot price of the hour or the quarter hour. On a fixed or variable contract that floor is in practice also the ceiling. Of six suppliers with a fixed 2027 rate, four pay around 6.2 to 6.6 cents per kilowatt-hour. The two that offer about double, ENGIE and United Consumers, take almost all of it back through feed-in costs: net, half a cent remains. On balance none of the six offers more than half. In the video we say "50% of the bare EPEX price". That is wrong: the floor is 50% of the bare supply rate on a fixed or variable contract, and on a dynamic contract you get the market price of the hour or quarter-hour.

The difference between the two jobs is in the tax. Whoever stores a kilowatt-hour and uses it later pays the energy tax in one hour instead of another, and the tax cancels out. Whoever stores a kilowatt-hour to sell it gets no tax back. In 2027 the energy tax in the first bracket is € 0.08579 per kWh, excluding VAT. That amount stands fully against you when you sell, and not when you use the power yourself. What that does to trading is in the article on trading with a plug-in battery and the article on feed-in compensation in 2027.

As soon as a kilowatt-hour can do two things, who chooses matters.

In the video, Victron is the example of DIY, for the property and not as a recommendation: the control is open, and you do nearly everything yourself. Which brands are locally open and through which port is in the article on control per brand.

An external party that lets your battery join a virtual power plant brings more financial benefit, in my view. What you give up for it is control, and part of the capacity. How much more it brings is not calculated here. So we give no amount.

On the product page, you check three things: is it closed, can you get at it locally, and may someone from outside steer it.

What makes an expensive battery expensive

If it is not the storage, what makes a battery expensive? Four things come on top:

• a model that controls the battery for you;
• backup when the grid drops;
• modules in the same stack, such as a DC charger;
• how easily you expand later.

One of those four chooses your type for you, whatever size the battery becomes: backup.

Backup on three phases points to fixed

If you want your whole house to keep running on three phases during a power cut, it becomes fixed. Some plug-ins can do backup too. That goes through a separate backup hub, or through a socket on the unit, and mostly on one phase. With three phases you would need three hubs. Then a plug-in overshoots its purpose.

Watch what the backup number means on a plug-in. "3,680 W off-grid" on a plug-in battery is the specification of a socket on the unit, not of your house.

If you buy fixed for the backup, check whether the transfer switch (the ATS) is built in, or whether it is an extra option. One example of that difference: the Tesla Powerwall 3P cost € 7,804.50 including VAT in August 2026, and € 8,833 with the Backup Gateway. Backup there was an option of € 1,028.50.

The other three

The model that controls it. With an AI mode, a model programs the automation, on the weather forecast and on what it has learned from your household's consumption. Whether that works better than what you set yourself, we have not measured.

Modules in the stack. Some fixed systems have a DC charger for your car in the same stack as the battery. The regular SigenStor has one, as an example of the property. Mind the name: the SigenStor Neo cannot take a direct DC charger. Other systems include control of your air conditioning.

Expanding. Here you ask three questions. How much can be added later? What does one step cost? And do the modules go in series or in parallel? With expansion, always name two numbers: the price is for the base unit, and the highest kWh number is the ceiling with separate modules added.

Peak shaving: ask the right question

Peak shaving is on the checklist for fixed. The word means two things, and manufacturers use one word. Shaving a price peak is trading. Shaving a power peak limits the current you draw from the grid.

With a fixed three-phase system, do not ask "can it peak shave", because that gets a yes everywhere. Ask for the unbalance per phase on the datasheet. That percentage decides whether the system can straighten out a lopsided load. Controlling per phase costs efficiency: Victron warns that the system then discharges on one phase while charging on another through the DC bus.

With a plug-in, the question is different: does it regulate on its own phase, or on the netted total of the P1 meter? The kWh meter nets across the three phases, the main fuse does not. And "zero on the meter" is not peak shaving. That mode discharges against every watt, and the battery is empty when the peak comes.

How big: the curve of one house

Then the second half of the question from the comments. How big?

How much of your own solar power do you keep, at which size? That question has been measured, on one house. De Datadame published the measurement in «Een thuisbatterij voor zonnestroom: hoe groot moet die zijn?» (A home battery for solar power: how big should it be?).

The set-up: a year of quarter-hour readings from the P1 meter of her own house, a terraced house on a fixed-price contract. On those readings she simulated batteries. If power went to the grid in a quarter hour, the battery charges, as long as it is not full. If power was drawn, it discharges, as long as it is not empty. Efficiency 90%. She sets the power equal to the capacity: 1 kW at 1 kWh, 3 kW at 3 kWh. The panels generated 3,540 kWh that year. Total consumption was about 2,800 kWh.

The percentages come from De Datadame. Our own conversion is in the kilowatt-hour columns: the percentage times the generation of 3,540 kWh, rounded.

The first 3 kWh shift about 850 kWh per year. Adding 7 kWh brings about 283 more. Another 10 kWh yields about 35. Above 3 to 4 kWh, it runs dead.

Why it runs dead: the calendar

The reason is the calendar, not the technology. De Datadame splits it out per month for a 4 kWh battery. In December and January all the surplus fits in those 4 kWh, and in November 6 kWh is left over in the whole month. A bigger battery therefore adds a few kilowatt-hours at most in those three months. In summer, consumption is the limit: during the day much of the consumption already runs on solar power, and the battery only has to cover the evening and the night. On most days, 4 kWh is enough for that. In her measurement, the step from 4 to 5 kWh is still worth 65 kWh per year.

Only at 1,000 kWh does self-consumption reach 85%. Then you move solar power from summer to winter, and that is no longer about a home battery.

De Datadame adds a warning herself. With quarter-hour data, self-consumption probably comes out somewhat too high. Short peaks above the battery's power disappear when you average over a whole quarter hour, and a battery responds with a small delay. The real curve is more likely a little lower than higher.

Where that house sits against the average

Is that house an exception? The CBS (Statistics Netherlands) makes it possible to place it, with a caveat.

According to CBS table 85005NED, a solar installation on a home averages 3.97 kWp (2023, final). The yield of the whole fleet was between 796 and 843 kWh per kWp in the final years 2019 to 2023. Divide De Datadame's 3,540 kWh by those two numbers, and the roof comes out at about 4.2 to 4.5 kWp. That result is an assumption, and it probably overestimates. The CBS yield is a fleet ratio: annual production divided by the capacity of a fleet that grew by hundreds of thousands of installations that year. New installations that ran for only part of the year push that number down. Applied to one installation that ran all year, it gives too large a roof. Her roof is therefore probably closer to the 3.97 kWp average than above it.

Consumption is harder to compare. De Datadame names her total consumption. The CBS publishes delivery through the grid. Those two numbers do not count the same thing.

What CBS table 81528NED does show is how far other homes sit from this house:

Export is spread over all homes of that type, including homes without panels. From apartment to detached, consumption more than doubles, and export becomes more than twenty times as large. All numbers from 2025 are provisional.

What the curve does not say

One house was measured: a terraced house on a fixed-price contract.

• A terraced house on a dynamic contract: not measured. The size does not change by itself there, but the kilowatt-hour gets a second job.
• A detached house, on any contract: not measured. For a detached house on a fixed-price contract, we expect the curve to break. More surplus means the curve flattens later. How much later, we do not know.
• A heat pump or an electric car: not in this curve. A car that charges from solar during the day takes away part of the surplus that would otherwise go to the battery. A heat pump shifts consumption to winter, when there is the least sun.

So the curve is a shape, not a norm. It shows how the yield per extra kilowatt-hour falls. It does not say where your knee is.

What the next kilowatt-hour costs

Put the curve next to a price. On the product page of a plug-in battery, these promotional prices, including VAT, were listed on 23 September 2026. We do not name the brand: this is about the shape of the price, not the product.

Per kilowatt-hour this promotion even falls below the plug-in band of € 245. Each next step of 4 kWh costs € 799. From 8 to 16 kWh you pay € 1,598 extra.

On the curve, that whole stretch lies in the flat part. At 10 kWh the measured house sits at 59%, at 20 kWh at 60%. The step from 8 to 16 kWh falls mostly in between. The curve has no measured point at 8 or 16 kWh, so we give no exact number of kilowatt-hours for that step. In this house it is in the order of a few tens of kilowatt-hours per year, for € 1,598.

The error in our own calculator

Until 13 September 2026, the calculator on this site carried a reference curve with four points: 30% without a battery, 65% at 5 kWh, 68% at 10 kWh and 70% at 15 kWh. Those numbers were not in the source the tool referred to. At 10 kWh the source measures 59%. The tool was nine percentage points too high there, and the points at 5 and 15 kWh do not exist in the measurement. The reference house's consumption was also set at 3,500 kWh. That was the generation of that house. Consumption was about 2,800 kWh.

We corrected the curve on 13 September. Only the measured points are in it now. Values in between come from interpolation, and not from a number we make up.

Put those nine percentage points next to the curve. From 3 to 20 kWh you go from 51 to 60%. Nine as well.

Our calculation error was exactly as big as everything you gain between a small battery and one of 20 kWh.

Where one of the two has no point, there is no difference. Putting a point of one curve next to a different point of the other would compare two different sizes.

Why the buying size sits above the knee

The curve bends at 3 to 4 kWh. My buying size is higher: around 7 kWh is usually enough, if you have no other requirements and no large consumers. A rule of thumb, not a measurement.

The reason: an installation of 4 kWp fills a 4 kWh battery in one hour of sun at full power, and then it is full. In the video I add a condition. The inverter and the charging side of the battery must be able to take that power too. A battery that charges at 800 watts is far from full after an hour of sun.

A second reason the curve itself points to: De Datadame calculated with power equal to capacity. In her model, a 3 kWh battery discharges at 3 kW, while a plug-in on a shared circuit delivers 800 watts. Whoever buys a small battery with low power does not reach the measured curve. That last sentence is inferred, not measured.

And the 18 kWh of the pro forma?

Is it too big? Not for that house. The house has about twelve panels, a pottery kiln and a lot of electric load. The 10 kW of the inverter and the 18 kWh of storage are both carried there by what hangs on the house. It was not measured the way the curve house was.

Bigger is justified when the load carries it. If the load does not carry it, the money is gone.

One viewer builds 16 kWh in the shed. Another viewer keeps it at two plug-in batteries. Both can be defended. I check the question, not the viewer.

The order, in five steps

The video shows the two lists on screen. Here they are with what you check per item:

Compare brands only after that. If you doubt the seller and not the brand, read on in the article on who you are buying from (in Dutch).

A kilowatt-hour is the same size in every cabinet. You buy what surrounds it, and no more kilowatt-hours than your house carries.

What has not been measured

These points are not stated as fact in the video, and not here either.

The curve for a detached house and for a dynamic contract. The route is set: first reproduce De Datadame's published curve from her own input. If a simulation lands on it, it may run other profiles, with NEDU consumption profiles and PVGIS for generation. Neither source has been opened yet.

How much a virtual power plant brings in. Not calculated.

Whether an AI mode controls better than your own settings. Not measured.

The split between fixed and dynamic contracts. The CBS has no table for it (searched 13 September 2026). So we name no share.

The price band for fixed. The band of € 350 to € 750 is an editorial estimate, not a count. The models in the class comparison cost € 518 to € 756 per kWh.

Accountability

Transparency. This article is based on a published measurement by De Datadame, two CBS tables, a pro forma and a screenshot of a product page, and our own knowledge base. Nothing was paid for this piece. No brand saw it in advance, and there are no affiliate links in it. We do not name the supplier of the pro forma or the brand of the promotional prices, because the article is about the shape of the price and not about a product.

Further reading

• What types of home battery are there? on the six categories and the price per kilowatt-hour
• Home battery too big? on the daily cycle versus annual consumption
• Controlling a home battery per brand on local access
• Feed-in compensation 2027 on what an exported kilowatt-hour brings after 2026
• Trading with a plug-in battery on the second job of a kilowatt-hour

The video for this article will appear on Jay's Desk soon.