A battery raises self-consumption only when it is doing the shifting job, not the backup job. Those are two different machines that happen to ship in the same case. A backup battery is held near full and waiting for an outage, so it has almost no room to absorb the midday surplus; a shifting battery is allowed to run down overnight on stored solar and refill the next day, and that discharge is what actually lifts the self-consumption figure. Buy capacity for load shedding and expect the self-consumption percentage to climb, and you will usually be disappointed, not because the battery is poor but because you asked it to do the other job.
I worked that out on my own kitchen floor, a bit after nine at night, with the inverter app open on my phone and the prepaid meter's red display glowing at me from the passage. The house was pulling 780 W. Fridge compressor, three lights, the TV, the fibre box and router, a laptop charger. The battery arc on the screen sat at 91%, and it had sat at 91% since about four that afternoon. Outside, the last of the day's surplus had gone to the meter and been valued at almost nothing.
Ninety-one percent. That number is the whole story. The installer had left the backup reserve at 90% because the family wanted to survive a stage 4 slot without noticing, and a reserve only works if it is genuinely held back. So the pack spent the afternoon full and useless, watching free electricity walk past it to the street.
Two jobs, one case, opposite behaviour
A backup battery is an insurance policy. Its value is in being full at the moment the grid drops, which means it must stay full the rest of the time. Every kilowatt-hour it hands you overnight is a kilowatt-hour of cover you no longer have, so a sane backup setup keeps it parked high and cycles it as little as possible.
A shifting battery is a warehouse. Its value is in how much it moves per day, in and out, all year. It wants to be empty-ish at sunrise, absorb the mid-morning and afternoon surplus, and then feed the evening genuinely, not symbolically.
Those two settings cannot both be optimal on the same pack. You can run both jobs on one battery, and most hybrid systems let you, but the reserve percentage you choose quietly decides which job wins. Set a 90% reserve and you have bought insurance with a small shifting hobby attached. Set a 20% reserve with a proper outage floor and you have bought a warehouse that will also carry you through a two-hour cut, if you keep the essentials lean.
The number on the app called self-consumption is simply solar used on site divided by solar generated. There is no magic in it. The only way to push it up is to put stored energy into real loads after the sun has gone, and the only way to do that is to leave room in the pack during the day.
The sizing mismatch nobody quotes for
Think about what each job actually demands in kilowatt-hours, because they are not the same order of magnitude.
A load shedding battery is sized around essentials for the length of the slot. Fridge, lights, wifi, TV, a phone charger. Add it up and you land somewhere in the region of 1.5 to 3 kWh for a four-hour outage, and that is generous. If you add the geyser or the pool pump to that list you will double or triple the pack for a convenience you can usually schedule around.
A shifting battery is sized around the evening, not the outage. Whatever the house normally draws between sunset and bedtime, plus the overnight baseline, minus whatever you are willing to lose. For a family of four cooking electrically, that is easily 6 to 9 kWh of usable energy per night, and on a winter evening it climbs.
The gap between those two numbers is the whole problem. A pack sized honestly for backup is usually too small to swallow a sunny Highveld or Karoo afternoon, so the surplus still flows out to the meter. A pack sized for real shifting is bigger than backup needs, and you are paying for cycle life and capacity you will never use during a two-hour cut. Neither is wrong. They are just answers to different questions, and the quote you sign usually only answers one of them.
What the round trip quietly takes
Even a perfectly sized shifting battery does not hand back everything it takes in. There are conversion losses in the inverter, losses in the cells, and losses again on the way out. In practical terms you put in a chunk of surplus and get back something less, the difference depending on whether the system is DC-coupled or AC-coupled.
That matters more than people expect, because it changes the arithmetic against exporting. If the municipal feed-in credit is a fraction of what you pay per unit, storing the surplus clearly beats sending it out at nearly nothing, even after losses. If the credit were ever close to your retail rate, the battery's round trip and its wear would eat most of the advantage, and you would be cycling an expensive asset to save very little. In South Africa the credits are generally low enough that storing wins on value, but you should check your own tariff schedule rather than assume it.
I also want to be blunt about something the sales conversation skips: a battery cannot make more solar. It changes when the energy is used. If your array is already fully absorbed by daytime loads, a battery adds nothing to self-consumption, because there is no surplus to shift. It is a timing device, and timing devices are only worth money when something is being mistimed.
Depth of discharge and the cycle ceiling
The spec sheet gives a usable capacity at a stated depth of discharge and a cycle count at that depth. Multiply one by the other and you get a lifetime throughput ceiling, roughly the total kilowatt-hours the pack will move before its warranty is exhausted.
That number is the real size of the shifting job. If you are moving 6 kWh a night, every night, you are spending that ceiling at a fixed rate, and you can work out in a minute roughly how many years of evenings it buys. Subtract the days you are away, the summer nights when the house is empty and the winter mornings when the pack never fully refills. What is left is the honest picture, and it is usually shorter than the ten or fifteen years people assume when they sign.
This is why I push back on the phrase "the battery pays for itself". It shifts value from the export column into the avoided-purchase column. Whether that is a good trade depends on the spread between the two, on how many cycles you actually spend, and on what else you could have done with the money. Nobody can promise that outcome from a datasheet.
The part that caught me out: the loads
Here is the thing I had to learn by watching, not by reading. Shifting only works if there is something to shift into.
My evening baseline, once the kids are down, is under 400 W. That means a 5 kWh pack with a sensible reserve can carry the whole night on a fraction of itself and still be sitting at 60% at dawn. I was congratulating myself on the self-consumption figure while the battery was doing almost nothing. The house simply did not have a night load big enough to justify the capacity.
The fix was not a bigger battery. It was moving loads. The geyser element onto a late-afternoon and early-evening schedule. The dishwasher deferred to after sunset instead of run at two in the afternoon on the timer. That is unglamorous and it works better than another 5 kWh of storage.
So before anyone quotes you a battery for self-consumption, do this. Take a weekday, note the time the sun actually leaves your roof, and write down what runs after that. If your big loads, the geyser, the pool pump, the washing machine, the cooking, all happen in daylight, your battery has nowhere to put the surplus and no load to serve at night. You will pay for a warehouse and use it as a shelf.
Reading it off your own meter instead of the app
I trust the prepaid meter more than the app. The app tells you what the CT clamp believes, and a clamp on the wrong side of the inverter, or facing the wrong way, will produce a beautiful self-consumption figure that has nothing to do with reality.
So I read the meter at the same time each evening, same time each morning, for a week before changing anything. Then a week with the reserve dropped and the geyser deferred. Units imported per day, side by side. That comparison is worth more than any dashboard percentage, and it costs nothing but a note on the fridge.
What you will probably find is this. The battery that lifts your self-consumption is the one you were willing to let run down, and the evening loads you moved are doing as much of the work as the cells are. The capacity itself is only half the decision.
