A self-consumed solar kWh is worth the retail tariff it saves you, while an exported kWh is worth only the feed-in credit the municipality applies to it. Those are two different prices for the same physical unit of electricity, and the only honest way to compare them is to strip both back to rand per kWh, at the hour and season in which each one actually happens, under assumptions you write down before you look at the answer. In most South African municipal structures the avoided retail tariff is the larger of the two numbers, which is why a kWh used behind your own meter does more for the system's payback than the identical kWh pushed out through it.
The two rows on the meter
Twenty past six on a July evening. I am at the boundary box with a torch, and the display is scrolling through its registers. The house is doing what every house does at that hour: geyser element on, kettle, the heat pump, every passage light. The import register is clicking over like a taxi meter. I stand there long enough to watch a whole unit go past.
Next morning, half past ten, I am back at the same box with a cup of tea. Nobody is home but me, the washing machine has finished, and the sun is doing that flat white Highveld thing. Now the second register is the one moving. Same array. Same inverter. Same kilowatt-hour. A completely different price.
That is the whole argument in one sentence: at the margin, every exported kWh is a kWh you did not use, so the two are not additive, they are alternatives. The comparison is not academic. It decides where your money went.
What an avoided kWh is actually worth
The self-consumed kWh is priced by the tariff you did not pay. Not the average of your bill divided by your units, which is the number most brochures quietly use, but the specific tariff in force in that specific minute. Three things move it:
- Time-of-use bands. Where the municipality charges peak, standard and off-peak, a kWh avoided at 19:00 is not the same animal as a kWh avoided at 11:00 on a Sunday.
- Season. Winter blocks run heavier than summer blocks on most TOU schedules, so a July self-consumed kWh can be worth more than the identical kWh in January.
- Blocks. Where the residential tariff steps up as consumption rises, an avoided kWh in the top step is worth more than one in the first step. Solar tends to trim the top steps, which is quietly in your favour.
Then the boring question that ruins good spreadsheets: does your number include VAT? Pick one convention and use it on both sides of the comparison, because a VAT-inclusive import tariff measured against a VAT-exclusive credit manufactures a gap that is not really there.
What an exported kWh is actually worth
The feed-in credit is set by the municipality under its embedded generation tariff, and it is not the retail tariff with a minus sign in front of it. Read the schedule, not the flyer. What I look for:
- Whether it is a credit against the bill or an actual payment. Almost always a credit. If your export exceeds your import in a given month, ask what happens to the surplus, because carried forward and paid out are very different animals.
- Whether it is capped, as a percentage of import, a monthly unit cap, or a cap tied to your approved inverter capacity.
- Whether it exists at all before your SSEG application is approved and a bidirectional meter is installed. Until then the number is zero, and any inverter push-back is a compliance problem under NRS 097-2-1, not a revenue stream. SANS 10142-1 sits underneath the installation itself either way.
- Whether prepaid customers are excluded or handled differently. In plenty of municipal schedules they are.
None of that changes the physics. All of it changes the arithmetic.
Put both on one basis
Two columns on one page. Column A is avoided import, in rand per kWh, for the hour and season I care about. Column C is the feed-in credit, in rand per kWh, on the same VAT convention.
The ratio between them tells you how many exported kWh it takes to match one self-consumed one. If A is three times C, exporting is worth a third per unit, and the priority order writes itself: use it, store it, export the remainder.
The numbers I am about to use are invented. They show the shape, not anyone's tariff. Say A is 3 at evening peak and 1.5 mid-morning, and C is a flat 0.8. Call them rand, call them anything. A kWh run straight into the geyser at 18:30 is worth 3. The same kWh exported at 11:00 is worth 0.8. A factor of 3.75.
Then the middle case that generic explanations skip: the stored kWh. If your battery round-trip loses, say, 10%, a stored evening kWh is worth 3 × 0.9, which is 2.7. Still more than three times the export credit. That is the part people miss. The battery is not competing with export. It is competing with the cost of the battery itself, per kWh cycled.
The bill-level trap
Per-kWh reasoning gets the margin right and the bill wrong if you stop there. The fixed charges do not move. Network charge, service charge, meter reading: none of them shrink when your panels work, and none of them grow when you export.
But there is a sharper trap. SSEG registration can put you onto a different tariff category, sometimes a different schedule, sometimes a different fixed charge, occasionally a different connection size. That shift can cost more across a year than the export credit earns, and it never shows up in a per-kWh comparison. Before you weigh A against C, confirm which category you will be billed under after approval. Not the one on the bill in your hand.
Timing is the lever, not array size
If your load and your generation do not overlap, the fix is rarely more panels. It is shifting load. The geyser on a timer at 11:00 instead of 18:00. The pool pump. The dishwasher. The washing machine. Move 2 kWh into the sun and you have converted 2 kWh of 0.8-value export into 2 kWh of off-peak self-consumption at 1.5, and if you can push it into peak band or peak season, better still.
I keep a week of hourly readings whenever I am sizing anything for a household. One week is enough to see the shape: when the kettle goes on, when the geyser recovers, what the base load is at three in the morning. Everything else, the array size, the battery size, whether export is even worth the paperwork, falls out of that one page.
Where export genuinely wins
I will be honest about the cases. If the array is oversized against a small daytime load, if the house stands empty from 07:00 to 17:00, if a battery's levelised cost per stored kWh sits above the spread between A and C, then buying a bigger battery purely to capture that spread is a bad trade. You would be spending capital to turn a 0.8-credit kWh into a 2.7-value kWh, and the battery might cost more per cycle than the gap is worth. Run that sum yourself before anyone sells you a battery as a self-consumption machine.
And if the municipality caps export at zero, or refuses it, C is zero and the answer is simple: size for the load, not for the grid.
What I write down before deciding
A short list, because the decision is only ever as good as the assumptions under it:
- Your municipal tariff category, and the exact schedule you will sit on after SSEG approval.
- Your TOU bands and season definitions, in your own municipality's words.
- Whether the credit is a bill offset or a payout, and what happens to a surplus.
- Whether the credit is capped, and how.
- Your own hourly load profile across a representative week.
- Your inverter's export limit setting, and whether someone has set it to zero.
- Battery round-trip losses, if a battery is in the plan.
Then compute both columns in rand per kWh, write the date at the top, and keep the page. Tariffs get revised every year. The framework does not. When the new schedule lands you re-run two columns and you know inside ten minutes whether the balance has shifted.
I still walk out to the meter box at 18:20 in July with a torch. Not because I doubt the arithmetic. Because the arithmetic is a claim about the world, and the meter is the world.
