Choosing What to Run: the Enterprise Matrix

Give "legality" a 15% weight in a scoring matrix and an unlawful enterprise loses fifteen points. It still scores 78% and still comes out ahead of a lawful one on 71%. The matrix has done exactly what you built it to do, and it has told you to break the law.

That is not a hypothetical. It is how South African smallholders end up with enterprises they cannot lawfully sell out of: rabbit meat with no registered abattoir, farm-gate slaughter, a tilapia strain that needs a permit nobody applied for. The arithmetic was fine. The structure was wrong.

#The gates have not changed. The answers have.

You ran the five gates in Module 1 on thin information: market, physical constraint, legality, input cost, capital and cash flow, in that order, stopping at the first honest no. Run them again now. The sequence is identical; what sits behind each gate is not.

Do this on the farm

Gate	What you now have behind it
1. Market	The carcass classification grid, and three named buyers who have written down a volume
2. Physical	Your CARA polygon value, discounted for veld condition and for browse the map excludes, converted into grazing days
3. Legality	The full compliance picture — the abattoir rule, the movement file, the species-specific permits
4. Inputs	Your own fodder-flow shortfall in LSU-days, and the winter lick bill it implies
5. Capital and cash	The transition-year income you will not earn, and the drought you are self-insuring

Score gates 1, 2, 4 and 5. Gate 3 is never scored.

Gate 1 is where the classification grid does its damage, and it is worth putting the numbers up before anyone talks about premiums. In the week ended 17 July 2026, beef A2/3 traded at R69.15/kg, B2/3 at R62.82 and C2/3 at R60.73 — an A-to-C spread of R8.42/kg, roughly R2 000 on a 240 kg carcass. Mutton is harsher still: A2/3 at R111.70/kg against C2/3 at R85.19, a spread of R26.51/kg. Weaner bulls made R47.75/kg that week (RPO).

Read what the grid pays for. It pays for young and appropriately finished. It does not pay for grass, and a slow grass-finished animal that ages into a B or a C class is penalised on that rail rather than rewarded. The research behind this course could not find a single published South African figure for what a customer will actually pay extra for grass-fed beef. If your matrix contains a premium, it needs an invoice behind it, not a belief.

#The column that is not scored

The legal position

Meat. Meat Safety Act 40 of 2000, s7(1)(c): no person may sell or provide meat for human and animal consumption unless it has been slaughtered at an abattoir. s7(2)(a) exempts slaughter for own consumption or for cultural or religious purposes, and s7(2)(b) then bars the sale of any meat from that animal to any person. Registration of an abattoir runs under s8 to the national executive officer, on a certificate lasting no more than five years (Act text). There is no lawful farm-gate slaughter-and-sell route in South Africa, and the Act bites only on the species in Schedule 1 — a list the Minister may amend, so check the current version rather than a course.

Live sale moves the animal, not the meat, and runs on documents: the registered mark under the Animal Identification Act 6 of 2002, a document of identification on disposal, a removal certificate where the transporter is not the owner, and a state veterinary movement permit wherever a controlled-disease condition applies.

Own consumption is lawful, and it is not a market. An enterprise justified on "we'll eat what we can't sell" has failed Gate 1, not passed Gate 3.

Everything that is not meat runs on its own regime. Eggs, wool, mohair, honey and fibre are not caught by s7 at all. Wool and mohair sell into an established auction channel with their own production standards and audits. Honey has a registration duty: beekeepers register annually with the department under a control measure made in terms of the Agricultural Pests Act 36 of 1983, and honeybee movement across the Cape-bee boundary needs a permit — this course did not read the control measure itself, so get the current forms from the department. Fish carry a listing question under the National Environmental Management: Biodiversity Act 10 of 2004 and its Alien and Invasive Species Lists of 2020, plus a provincial conservation permit — and the second applies whether or not the first does.

The discipline is simple and unpopular: a "maybe" in the Gate 3 column is a fail until somebody at the department has put the answer in writing. Draft regulations are not a route. An abattoir that "will probably take small lots" is not a route. A neighbour who has been doing it for years is not a route either — he is an unprosecuted offence.

#Two columns people merge, and must not

Cost per kilogram produced tells you whether the enterprise is efficient. Cash-flow timing tells you whether you will still be solvent when it proves it. They are different questions and they belong in different columns, because an enterprise can win one and bankrupt you on the other.

What it costs, what it earns

Time to the first cheque — the column nobody builds and everybody needs:

Enterprise	First income	Source
Quail eggs	Lay from 6–7 weeks; 250–300 eggs in year one	SA Smallholder — a smallholder publication, so corroborate before budgeting
Outdoor pork	About 12 months to a ±40 kg carcass, 100 kg-plus at 18 months	Wickedfood Earth — one farm's own published figures, not a trial
Cattle, weaner system	At the observed smallholder inter-calving period of 608 days, about 0.6 calves per cow per year	Smallholder benchmark study
Ostrich	"It takes 30 months before money can be obtained from the business"	DALRRD Ostrich Profile 2023

Say the pork figure in matched units or a learner will correctly call it out: ±40 kg is a carcass weight, and the commercial "100 kg in six months" everyone reaches for is a live weight. At Kolbroek-type dressing percentages that carcass is near 50 kg live. So the honest statement is that this outdoor system takes about twelve months to produce a pig at roughly half the live weight a commercial pig reaches in six.

Cost per kilogram, from the same book (Land Bank Livestock Enterprise Budget 2023/24 v2 — 2023/24 prices, and weaners were up 29.8% year on year to July 2026, so these budgets understate current income and must be re-priced): Boer goats on veld plus a modest lick return R1 301.70 per doe and R27.23 per kilogram sold, with labour at 71% of costs. The 500-bird broiler batch returns about R0.05 of gross margin per rand of cost, with feed at roughly 69% of cost — so a 2% feed price move erases about 30% of the margin (Land Bank).

Never set that book's Merino figure of R10 843/ha beside a veld figure. That budget is an irrigated lucerne pasture carrying 13 sheep units per hectare — an intensive forage-crop enterprise with sheep on it. Gross margin per hectare is only comparable between enterprises competing for the same hectares.

Two enterprises can share a cost per kilogram and be entirely different propositions. Quail reach first income in six weeks; the ostrich pen is empty of money for two and a half years, and the winter lick bill still arrives every June. That is not an argument against ostriches. It is an argument for putting both columns up and asking which one your bank account is sensitive to.

#The four this course advises against

Careful — this one bites

Each of these fails on published South African evidence, not on taste.

1. Inland tilapia, on the temperature record. The national water quality guideline gives Mozambique tilapia a growth optimum of 28–30 °C and a lower lethal range of 9–11 °C, and states that "the costs involved in controlling temperature under pond-culture conditions are prohibitive and not feasible" (DWAF Vol 6). A 30-year record for Meyerton computed independently puts the warmest month at about 20.9 °C and June and July at 10.7 °C and 10.4 °C (ERA5 archive). No month reaches the optimum and midwinter sits inside the lethal band. Two honest qualifications: those are air temperatures used as a proxy, since no measured Highveld pond series was available, and a deep dam runs warmer than air on a winter night. Log your own water at 06h00 for one winter before you spend anything. Add to that the permit question — most farmed tilapia other than the two indigenous Oreochromis species, including the red hybrids widely marketed here, are listed and need permits — and the enterprise closes at Gate 2 before it ever reaches Gate 3.

2. Rabbit meat, on the absence of a lawful sale route. Rabbit sits in Schedule 1 of the Meat Safety Act 40 of 2000, so s7 applies in full: eat your own, sell none of it — including as pet food, since s7(1)(c) covers meat for human and animal consumption. Draft rabbit-abattoir regulations were published in 2025, but this course could not establish that they have been finalised or that any rabbit abattoir is registered in South Africa. Until somebody shows you an open route, a rabbit meat business is a plan to break the law.

3. Emu, on breeder-stock speculation. The species is formally recognised in South African standards and the biology works. The market is the problem: no current South African emu population, producer count, oil volume or price could be traced to any authoritative source, and the 1990s boom in the United States and Australia is widely understood to have been a breeder-stock bubble with no end-user demand — a narrative this course could not source primarily, so treat its dates and sizes as folklore. The structural test still applies: any plan whose income comes from selling breeding pairs rather than product is that bubble, repeating.

4. Dairy goats, without a measured pilot. About 1% of South Africa's goats are used for milk — the lowest share in southern Africa — and the reasons given are biological rather than commercial: milk goats are vulnerable to tick-borne disease and to heat stress (African Farming, April 2025). No producer numbers, herd sizes, yields, farm-gate milk price or cheese price were obtainable from any authoritative South African source, which means nobody can budget this enterprise honestly, including this course. It is a value-added cheese business with a goat herd attached. The answer is not "no" — it is a small measured pilot with your own yield, mortality and cheese-yield record for one season, before a rand of processing capital moves.

Contested — here is both sides

Is "no published South African economics" a fair reason to refuse an enterprise?

No, say the challengers. Absence of evidence penalises novelty and locks smallholders into the few commodity enterprises the Land Bank happens to publish budgets for. Somebody had to be the first Boer goat farmer. And the published budgets are not sacred either — the same book carries a Dorper gross margin that does not reconcile with its own income and cost lines, and a broiler page stating two different revenue figures.

Yes, say the sceptics, and their version is narrower than it sounds. The refusal is not "never do this". It is "do not scale this on numbers that do not exist". An enterprise with no published economics has no benchmark, so you cannot tell a bad season from a bad enterprise, and you will find out with your own capital.

Where this leaves you: the pilot is the reconciliation. Run the unproven enterprise at a size whose total failure you can absorb, measure it for a season, and let your own record become the benchmark that was missing.

#A shortlist is still not a plan

Here is the point at which most enterprise matrices quietly fail, and it is arithmetic rather than judgement.

South African specifics

Take 100 ha of Highveld veld at the Gauteng mean of 6.8 ha/LSU. That is 14.7 LSU — about nine breeding cows once you convert at the Land Bank's 1.6 LSU for a cow with a calf at foot and leave room for a bull, or a mixed ruminant load of the same size. Note how much the answer moves with the conversion factor you pick: at the CARA Table 2 figure of 1.21 for a dry medium-frame cow the same veld looks like twelve. Say which factor you used. Now put the shortlist against the year. On sourveld, veld supplies adequate nutrition for roughly five to seven months, and in some conditions only three. Winter lick is 55% of the variable cost of an extensive South African beef enterprise (Land Bank).

So the question the shortlist must answer is not "can I run cattle and goats here?" It is: do I hold enough LSU-grazing-days to carry 14.7 LSU from the last green growth in autumn to the first reliable flush in spring, and if not, what does the shortfall cost at this year's lick price?

If the fodder flow does not balance at the stocking level the shortlist implies, the shortlist is wrong. Not the veld.

The gates tell you what is possible and what is lawful. The fodder flow tells you how much of it you can carry. An enterprise mix that passes all five gates and then runs 20% over the LSU-days you hold is not a plan — it is a lick bill with a business name.

So go back to your shortlist page. Add the Gate 3 column and mark every row pass or fail, with a date and a name against it. Add the timing and cost columns. Then take the surviving rows to your fodder flow and see how many the winter lets you keep.

#Check yourself

3 questions — answers explained as you go

1Your matrix scores five enterprises out of 100 across weighted columns. One scores 81% but you cannot yet confirm any registered abattoir near you that will take small lots of that species. What should the matrix do with it?
Why: Under s7(1)(c) of the Meat Safety Act 40 of 2000 you may not sell or provide meat that was not slaughtered at an abattoir, and the own-consumption exemption in s7(2)(a) is immediately followed by s7(2)(b) barring the sale of that meat to anyone. There is no partial compliance to weight, and a fallback you cannot lawfully sell out of is not a fallback. Confirm the abattoir, in writing, or drop the row.
2An ostrich enterprise and a quail-egg enterprise show a similar cost per kilogram of product. Why is that not enough to compare them?
Why: Efficiency and timing are separate questions and need separate columns. The department's own value-chain profile says it takes 30 months before money can be obtained from an ostrich business, and it adds that ostrich farming "does not lend itself towards subsistence farming". Quail are firmly inside the Meat Safety Act — they are Schedule 1 animals, which is exactly why quail ventures survive on eggs rather than meat.
3A learner's shortlist clears all five gates and implies a stocking level of 18 LSU on 100 ha of Highveld veld with a mapped capacity of 6.8 ha/LSU. What has the matrix not yet tested?
Why: 100 ÷ 6.8 gives 14.7 LSU, so 18 LSU is roughly a 22% overshoot that has to be fed through the very months the veld is worth least — and winter lick is already 55% of variable cost in the Land Bank's beef budget before any overshoot. Species compatibility is a real question and the wrong one here. The map carries one capacity figure for the polygon, not species permissions, and it excludes browse entirely, so a browsing component has to be counted on its own axis rather than argued around.

Choosing What to Run: the Enterprise Matrix

By the end of this lesson you can

#The gates have not changed. The answers have.

#The column that is not scored

#Two columns people merge, and must not

#The four this course advises against

#A shortlist is still not a plan

#Check yourself

Sources for this lesson

#A spreadsheet will eventually recommend a crime

#The gates have not changed. The answers have.

#The column that is not scored

#Two columns people merge, and must not

#The four this course advises against

#A shortlist is still not a plan

Sources for this lesson