Self-consumption is the share of your own solar production that your house actually used, and it comes out of two figures you already have: daily solar yield, and the energy that crossed the property boundary on its way out. Subtract export from yield, divide by yield, and the number is there. The reason it feels hidden is that those two figures live in different places — usually one on a monitoring app, sometimes the other on that same screen, and on plenty of South African systems nowhere at all.
I worked this out on a Wednesday evening at the kitchen table, phone in one hand, municipal account in the other. Six o'clock, kettle on, geyser finished its afternoon cycle about an hour earlier. That last detail is the one I'd underline, and I'll come back to it.
Suppose the app showed 24 kWh of PV yield for the day and 8 kWh of export. Sixteen kilowatt-hours never crossed the boundary. Now check that from the other direction. If household load was 19 kWh and grid import was 3 kWh, then 16 kWh of the load came from something that wasn't the grid. Two routes, same sixteen. When both routes land on the same figure, you can act on it. When they don't, you've just found which number in your app is arithmetic dressed up as a measurement.
The account on its own cannot do this
My meter is prepaid. It tells me units bought, units left, and how fast it's eating them. That's import, and only import. It has never mentioned export, because most prepaid meters aren't built to record direction of flow. So a bill-only approach can show that you're importing less than you used to, which is real information, but it cannot hand you self-consumption. You'd be dividing by a number you don't have.
Registered SSEG systems are the exception. Once the municipality has fitted a bi-directional meter as part of the registration, export appears somewhere on the account and the bill genuinely becomes useful for this. If yours hasn't been through that, don't force the account to answer a question it was never asked.
Where the app is guessing
PV yield is usually a real measurement. Load often isn't. On many systems the app takes the grid reading, adds or subtracts the solar reading, and presents the result as household consumption. Export can be the same story — on inverters without dedicated export metering, the export figure is inferred from a small set of sensors sampling at different points and slightly different moments. None of that makes the app useless. It makes a second route worth doing.
If your inverter has no export metering at all, you will never cleanly split self-consumed solar from exported solar. What you can still do is estimate the whole system: take household load for the day and subtract grid import. Whatever is left came from the array. One figure instead of two, and honest about what it is.
Two catches worth knowing before you trust that subtraction. First, if the system is set to charge the battery from the grid, or if it tops the battery up after a load-shedding outage, that energy sits inside the import figure — and it isn't household consumption. Self-consumption comes out looking lower than it really is. Second, app load figures often carry inverter standing losses and conversion losses inside them. Small, but it means the two routes will never agree to the decimal.
Seven days, not one perfect Tuesday
One sunny Tuesday with the house empty tells you almost nothing. My worst self-consumption days are Saturdays — everyone home, tumble dryer running, washing machine on, pool pump timer landing in the middle of it, and the battery full by eleven so the midday surplus has nowhere to go but out. Sundays have their own shape. Weekdays have another.
So run the arithmetic on totals, not on days. Seven rows in a notebook: date, yield, export, load, import. Add the columns at the end of the week and calculate once on the weekly figures. That percentage is the one worth keeping, because it already contains the household's actual rhythm instead of its best case. It's also the figure you can hold up against anything a proposal claims.
Self-consumption and self-sufficiency are not the same fraction
This is where quotes get slippery, usually by accident. Self-consumption is the share of what your panels made that stayed on site. Self-sufficiency is the share of what your house used that came from the panels. Same day's data, two different fractions.
And they move in opposite directions as you add capacity. Put on more panels and self-sufficiency climbs while self-consumption falls — more production, same load, more of it leaving the property. A proposal that promises to improve both by adding modules is promising something the arithmetic won't deliver. If you only ever track one of the two, track self-consumption when you're judging how well the system suits the house, and self-sufficiency when you're judging how exposed you still are to the grid.
What the number is actually for
It's most useful as a load-shifting signal rather than a score. A low self-consumption percentage with high export means the array is fine and the timing is wrong. Move the big afternoon-shaped loads into the middle of the day and the same hardware delivers a better number without anyone touching the installation.
- The geyser element, if it's on a timer, is the biggest single lever in most homes.
- The pool pump and the dishwasher are next — both are indifferent to when they run.
- The tumble dryer is the awkward one, because it wants to run when the weather is bad.
A battery lifts self-consumption in a different way, by carrying midday surplus into the evening instead of pushing it out. That's a genuinely different mechanism from load shifting, and it's the one that shows up as a smaller export figure rather than a smaller import figure. If neither the export nor the import figure moved after the battery went in, the battery isn't doing the job you were sold.
So when a proposal puts a self-consumption percentage in front of you — 80%, 90%, whatever the slide says — you now have a way to test it that doesn't involve trusting anyone. Take your own week. Add the columns. Do the division. If the claim sits far above what your own household rhythm produces, the claim was about the panels, not about your house.
The kettle had boiled twice by the time I finished that Wednesday. The number I wrote on the back of the account was 67%, and it was the first figure about my system I actually believed, because I'd built it from two records that had no reason to flatter me.
