Example: a two-acre market garden going direct to consumers

We grow vegetables for people who live within half an hour of the farm and want to know who grew what they eat. Where we want to be in three years: selling most of what we grow before it is harvested, through subscriptions and standing restaurant orders, rather than carrying it to a market and hoping. That matters for two reasons. Produce sold in advance does not get carried home again at the end of a Saturday, and a subscription paid up front is money in the account in February, when there is nothing to sell and everything to buy. We will know we have got there when more than half our income is committed before the season starts, and when we are no longer deciding what to plant by guessing what will move.

Three groups, and they behave differently. Market customers buy on the day, mostly in small quantities, and choose on how things look. They are how new people find us, and they are the least predictable part of the week. Subscription members pay before the season and collect weekly. They are households who cook most nights and who have decided in advance that they want vegetables to arrive. They tolerate variety far better than market customers do — a member will take kohlrabi and ask what to do with it; a market customer walks past it. Restaurant buyers want the same thing every week, in a known quantity, at a price agreed in advance, delivered on a day they choose. They are the least forgiving about consistency and the easiest to plan around. We are deliberately growing the second and third groups. The first is what pays now; it is not what we want to depend on.

Our competition is not the other growers at the market. It is the supermarket, and we lose to it on price and on convenience every time. What we have that it does not: produce picked within a day of being eaten, varieties chosen for how they taste rather than how they ship, and a person the customer can ask. That is worth a premium to a portion of households and nothing at all to the rest. The plan does not assume we can convert the rest. Among direct growers locally, the ones nearest us grow a similar range. We are not trying to differentiate by crop. We are differentiating by reliability: a member who has had twenty consecutive good boxes does not go looking.

Our prices start from what a unit costs us to grow, which the crop budget works out per pound from our own seed, transplant, labour and packaging costs, and which we recalculate each winter from the previous season's records rather than carrying last year's number forward. On top of that we take the price the local market will actually bear, which we know from what has and has not sold at our own stall over three seasons. Where the two disagree — where a crop costs more to grow than anyone here will pay for it — we stop growing it. Two crops have come off our list that way. Subscription pricing is set so that a member pays slightly less per pound than the same produce would cost at the stall, which is what they are buying with the commitment and the money up front. Restaurant pricing is lower again and fixed for the season, which is what the guaranteed volume is worth to us. We have not assumed a price rise in the projections. If we get one, the plan improves; it does not depend on it.

Weather. We are in a humid subtropical climate with a summer that ends several crops and a hurricane season that can end all of them. Mitigation: two plantings of everything that matters, staggered; the propagation house and the wash-pack shed anchored to rated standards; and a written plan for what gets harvested early when a storm is three days out. There is no mitigation that survives a direct hit, and the plan does not pretend otherwise. Losing a restaurant account. Our largest buyer is a meaningful share of committed income, and restaurants close. Mitigation: no single account is allowed past a share of income we have set in advance, and we hold a waiting list rather than filling every slot. Us. This is a two-person operation and both of us do physical work. An injury in June is the risk nobody writes down. Mitigation: we carry insurance, we have two people locally who have worked the farm and could cover a harvest, and the crop plan for any season either of us is unfit is the reduced one we have already written.

The figures in the financial section rest on four assumptions, which we would rather state than bury. First, that the subscription holds its renewal rate. We have three seasons of our own renewal data and have projected slightly below the worst of them. Second, that the restaurant orders continue at the volume agreed. They are not contracted beyond the season, and we have modelled what the year looks like if the largest one stops in month four. Third, that yields hold near our own three-year average for each crop, from our own records, on our own ground. We have not used any published average, because our soil and our irrigation are not anybody else's. Fourth, that we do not lose a planting to weather. This is the one outside our control, and it is the assumption most likely to break. The cash-flow section shows what a lost summer planting does to the year. If a reader wants to stress this plan, the fourth assumption is where to push.

Money does not arrive evenly and neither do the costs. The tightest stretch is late winter into early spring: seed, potting mix, transplant trays and soil amendments are all bought before anything is sold, and the previous season's sales have stopped. Subscription payments taken in advance are what covers it, which is one of the main reasons we are growing that side of the business. The second tight point is late summer, when heat has ended the spring crops and the autumn ones are not producing yet. We hold back a portion of the spring's income against it rather than treating a good May as money we have. We have not planned around an operating line of credit. If we take one it will be to smooth the first gap, not the second, and the projection shows the position with and without it.