A market garden business plan should include a bed-by-bed crop plan, expected yield and price for each crop, the sales channels and their share, labour hours for growing and selling, fixed costs, and a cash budget. AI helps by building the table and testing scenarios; you supply the real yields, prices and hours.
The figure that matters most on a small plot is not profit per acre but revenue and profit per bed-metre and per hour worked, because space and your own time are the two things you cannot buy more of. In the half-acre example below, salad mix earns $63 a bed-metre against $7.50 for potatoes, and the first full budget pays the grower about $17.50 an hour, which is why the plan ends up changing its crop mix.
What the plan has to answer before it is long
A plan for a small market garden is short when it answers six questions, and padded when it does not. Write one paragraph or table for each:
- Land and tenure. How much cultivable area, on what terms, with what water. A half-acre of land is roughly 2,000 square metres, but beds, paths, headlands and a wash area leave a good deal less in growing beds.
- Crop and bed plan. Which crops, how many bed-metres each, and when they go in and come out.
- Markets. Which channels, at what share, at what price and with what time cost.
- Labour. Hours for growing, harvesting, washing, packing and selling, and who does them.
- Costs. Direct crop costs and fixed costs, with the start-up equipment listed separately.
- Risks and what you will do about them. Weather, a lost channel, a price drop, your own illness.
Rules about selling fresh food, licensing, insurance and organic labelling differ by jurisdiction, so check what applies to you with the relevant authority or a growers' association before you rely on any plan. The plan below is about the economics.
How a half-acre turns into 1,170 bed-metres of plan
Suppose a grower has a half-acre, about 2,000 square metres, and uses beds 0.75 m wide on 30-metre runs with 0.45 m paths. One bed plus its path takes 1.2 m of width, so a 30-metre run takes 36 square metres. Allowing space for a wash and pack area, a tunnel, headlands and a compost bay, roughly half of the land is in beds. That gives about 28 beds of 30 metres, or 840 bed-metres of land. Crops that follow each other in the same bed in one season (a quick crop of salad before tomatoes, garlic followed by a late crop) bring the planted total above the land total. Plan with the planted bed-metres, counting each crop separately, and keep the land figure beside it so you do not plant more than the ground allows.
The illustrative plan below has 1,170 planted bed-metres on that ground. If your own beds, spacing and succession differ, change the numbers; the method stays.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowRevenue per bed-metre and per hour: the crop table
Ask an assistant to build the table, and give it the inputs yourself. Every number is a placeholder for your records. The illustration prices at the farmers' market level before channel effects, uses an all-in labour figure (bed preparation, planting, weeding, irrigation, harvest, washing and packing) and counts direct costs such as seed, trays and packaging.
| Crop | Bed-metres | Yield per bed-metre (kg) | Price per kg | Revenue per bed-metre | Hours per bed-metre | Revenue per hour |
|---|---|---|---|---|---|---|
| Salad mix | 150 | 4.5 | $14.00 | $63.00 | 0.90 | $70 |
| Carrots, bunched | 200 | 4.0 | $4.50 | $18.00 | 0.35 | $51 |
| Beetroot | 150 | 4.5 | $4.00 | $18.00 | 0.30 | $60 |
| Kale | 100 | 3.0 | $6.00 | $18.00 | 0.35 | $51 |
| Courgettes | 150 | 5.0 | $3.50 | $17.50 | 0.40 | $44 |
| Tomatoes, tunnel | 120 | 8.0 | $6.00 | $48.00 | 1.00 | $48 |
| Garlic | 100 | 1.2 | $16.00 | $19.20 | 0.25 | $77 |
| Potatoes | 200 | 3.0 | $2.50 | $7.50 | 0.15 | $50 |
Read it twice, because it gives two different rankings. By revenue per bed-metre, salad mix at $63 and tunnel tomatoes at $48 stand far above everything else, and potatoes at $7.50 are poor use of scarce bed. By revenue per hour, garlic ($77) and salad ($70) lead, and courgettes ($44) trail. A crop can be good for space and ordinary for labour, or the reverse, and the plan should know which limit you are up against. On a half-acre worked by one person, labour usually binds before land does, so revenue per hour deserves the heavier weight.
The total at full market price comes to $29,355, with direct crop costs of $2,725 and 520 hours of growing labour. Those are the starting totals for the budget.
A prompt for building the crop table without letting AI invent yields
I run a half-acre market garden with 30 m beds, 0.75 m wide. Build a
crop plan table from my inputs. Columns: crop, bed-metres, yield per
bed-metre (kg), price per kg, revenue per bed-metre, hours per bed-metre,
revenue per hour, direct cost per bed-metre, total revenue, total hours.
Rules: use only the numbers I give. If I give a range, use the lower
end and show the upper end in brackets. Mark any value I left blank as
MISSING and list what I must find out. Check that total bed-metres do not
exceed 1,170 planted (840 land plus succession crops). Show the sums.
MY INPUTS:
[paste crop, bed-metres, yield, price, hours, cost for each crop]If you do not yet have your own yields, ask the assistant for typical ranges and write them in a separate column labelled "placeholder". Then test the plan on the low end. A grower's first season often falls short of the textbook figure, and a plan that only works at the textbook yield is a hope.
Channel mix: market, box scheme and restaurants
The prices above are market prices. The other channels pay differently, and they cost different amounts of time. The illustrative split: 40 percent of the crop by value goes to a weekly farmers' market at the full price, 35 percent to a box scheme at 90 percent of that price (a small discount for a committed weekly customer), and 25 percent to restaurants at 75 percent (wholesale, but delivered in bulk). That blends to 90.25 percent of the full market value: $29,355 becomes $26,493.
| Channel | Share of crop value | Price against market | Revenue | Selling hours a season | Revenue per selling hour |
|---|---|---|---|---|---|
| Farmers' market (26 weekly stalls) | 40% | 100% | $11,742 | 208 (8 hours a market: load, travel, sell, unload) | $56 |
| Box scheme (about 15 boxes a week, 26 weeks) | 35% | 90% | $9,247 | 156 (6 hours a week: pick-list, pack, deliver) | $59 |
| Restaurants (a handful of weekly accounts, 26 weeks) | 25% | 75% | $5,504 | 78 (3 hours a week) | $71 |
The surprise is the one you would not see without the hours: restaurants pay least per kilo and most per hour of selling, because one drop-off moves a lot of produce. The market is the slowest to earn but builds your name and tests prices. A new grower often over-weights the market because it feels like the "real" job, then finds the stall day cost a full working day.
AI is useful here for a different job: the channel plan text. Ask it to draft the weekly box packing list from the harvest forecast, a one-page offer for restaurants listing what you can supply in each month, and the market stall price list. These are routine documents, and for the box side AI for farm shops: weekly boxes, labels and pre-orders goes deeper.
The season's budget: cash profit and profit after labour
The fixed costs below are illustrative. Add the ones that apply to you and leave out the ones that do not.
| Line | Amount |
|---|---|
| Revenue after channel mix | $26,493 |
| Direct crop costs (seed, trays, inputs, bags) | −$2,725 |
| Market fees, 26 stalls at $35 | −$910 |
| Box packaging, 390 boxes at $1.20 | −$468 |
| Water and irrigation parts | −$900 |
| Land rent | −$1,200 |
| Insurance | −$650 |
| Tools and equipment wear | −$1,500 |
| Vehicle fuel | −$900 |
| Website, signs, marketing | −$400 |
| Cash profit before your own pay | $16,840 |
| Hours worked: 520 growing plus 442 selling | 962 hours |
| Cash profit per hour | $17.50 |
| If every hour were paid at $20 | −$19,240 labour; result −$2,400 |
This is the sobering number a plan exists to show. The half-acre, as planned, pays the grower about $17.50 an hour before tax, and would lose money if the work were hired in at $20. The plan has not failed; it has shown that the crop mix and sales mix, not the acreage, are the problem. That is cheaper to learn from a spreadsheet than from a first season.
Changing the plan: what drops, what grows, and the demand limit
The crop table pointed at potatoes: 200 bed-metres earning $7.50 each. Drop them, and plant 100 more bed-metres of salad and 50 more of carrots, taking carrots to 250 and salad to 250. (A bed-metre count of 1,120 results, since the 50 extra carrot metres and 100 extra salad metres use only 150 of the 200 freed.) The revised plan, with 40 more selling hours for the extra salad:
- Revenue after channel mix: $31,637, up from $26,493.
- Cash costs: $9,963, up from $9,653 (more salad inputs).
- Cash profit: $21,674, up from $16,840.
- Hours: 1,080, up from 962.
- Cash profit per hour: $20.08, up from $17.50.
Better, but not rich, and it rests on an assumption the plan must challenge hard: that customers will buy 250 bed-metres of salad, about 1,125 kg a season, 450 kg more than before. Salad sells fastest when it is fresh and has a short life. Before you commit, ask the assistant to turn the extra volume into weekly kilograms and compare it with what the restaurants and market customers can actually take. If the demand is not there, the extra salad sells at a discount or rots, and the plan returns to where it was. Use the assistant to test it at a lower price and a lower yield, as in the next section.
Checking the extra salad against weekly demand
The salad plan is the clearest example of why a crop table is not a business plan. Put the 250 bed-metres on a calendar and the arithmetic changes. Suppose salad is sown in four successions from early spring to early autumn, so about 62 bed-metres are cut at a time, and each bed is cut three times over about six weeks. At 4.5 kg per bed-metre per season, a single succession gives roughly 280 kg across its cuts, or about 47 kg a week while it is producing. With two successions overlapping at times, the weekly volume sits between about 47 and 95 kg.
Now compare with demand. If the market takes 30 kg a week, the box scheme 15 kg (about a 1 kg bag in each of 15 boxes) and the restaurants 20 kg, the total is 65 kg a week: more than one succession produces, and about right for two overlapping. So in weeks when only one succession is cutting you are about 18 kg short of what customers would take, and in weeks when two overlap you have about 29 kg more than they take. Two decisions follow: stagger the sowing dates so the supply is level, and agree a standing order with at least one restaurant before sowing. The assistant can lay this calendar out in a minute once you give it the sowing dates and the weekly demand, and it can show the weeks with a surplus. It cannot tell you whether the restaurant will agree.
The same check applies to any crop that sells fresh: tomatoes, courgettes and kale each peak. Storage crops such as garlic and roots are more forgiving, which is part of why a mixed plan holds up better than one built on the highest earner.
A first-season calendar the assistant can draft and you correct
A calendar turns the plan into jobs. Ask for a month-by-month list for the first year: when to order seed, when beds are prepared, when each crop goes in, harvest windows, the start of each sales channel, and the cash needed each month. A sample line of an illustrative output:
| Month | Field jobs | Sales | Cash note |
|---|---|---|---|
| Late winter | Order seed and trays, prepare tunnel beds, start tomato seed | Sign up box members, pitch to restaurants | Outlay for seed and trays, box deposits come in |
| Spring | Sow first salad and carrots, plant garlic follow-on, transplant tomatoes | Market stall starts, box deliveries begin | First income, still below costs |
| Midsummer | Peak harvest, second salad sowing, weeding | All three channels running | Highest income, highest hours |
The assistant will produce a plausible calendar for a generic climate, and every date in it will be wrong for your plot. Correct it with your local frost dates and your own sowing notes; a growers' group or a local extension adviser can give them to you.
Stress tests for a plan you are going to stake money on
Ask the assistant to re-run the budget with each of these changes, one at a time, and report cash profit and cash profit per hour:
- Salad yield 25 percent below plan (a poor spring).
- Market price of salad down 15 percent (a competitor undercuts you).
- Loss of the restaurant channel altogether, with its share of the crop sold at the market at end-of-day discounts, or not sold at all.
- Selling hours 20 percent higher than planned.
- One tunnel crop failing and replaced by a late cheaper crop.
The point of the tests is to learn which assumption the plan cannot survive. An illustrative result: at a 25 percent salad yield shortfall, revenue falls by about $2,100 on the original plan, and cash profit per hour falls from $17.50 to about $15.30. That is survivable. Losing the restaurants hurts less than you might think on revenue but changes the week's rhythm, since the market and box then carry everything. Write down which of these you would see first, and what you would do at that point.
Mistakes that make a market garden plan look better than it is
Counting every bed-metre once. The opposite error to double counting: a succession crop is a second set of bed-metres and a second set of hours. Missing them understates labour.
Pricing at the best market day. A stall price list is not an average of what you take home. Use realised prices, including the unsold boxes and the second-best produce you sell at a discount. Put a waste line in from the start; until you have your own records, something like 8 to 12 percent of harvest is a placeholder to test, not a benchmark.
Forgetting the hours that are not on the bed. Washing, packing, invoicing, market admin, repairs, the hour in the morning loading the van. An assistant will not add them unless you list them, so ask it: "What hours does this plan leave out?"
Ignoring the calendar. Revenue arrives in bunches (spring salad, summer tomatoes, autumn roots), while costs are spread through the year. A monthly cash view shows the months you run short. Ask for a 12-month cash table, and check that your savings or an overdraft covers the lowest month.
Where the plan's numbers live once the season starts
A plan is a forecast; records are the facts. From the first week, log the yield per bed, hours worked, price actually received and waste for each crop. A notes app, a shared sheet or a farm records tool will do, and a simple template beats an ambitious system that is not filled in. After one season the gap between the plan and the records is your most valuable asset: replace the placeholders with your own figures and rerun. For ideas on keeping records and paperwork without drowning in them, see can farmers use AI for farm records and paperwork.
When you reach the point of choosing accounting software for the business, farm accounting software compared covers the options. Tax and legal structure questions belong with an accountant who knows small farms; do not rely on a plan or an assistant for them.
A one-page version for a lender, a landlord or a partner
Once the spreadsheet is settled, ask the assistant to summarise it in a page that someone outside farming can read. Give it the tables and these instructions: state the plot size and the number of bed-metres; list the crops by revenue per bed-metre; show the channel mix and what each contributes; state expected cash profit, hours and cash profit per hour, together with the low-yield case; list the three biggest risks and the response to each; do not add figures that are not in the tables. Read the result against the sheet, line by line. A summary that rounds $17.50 an hour up to "around $20" is the kind of small flattery that damages trust with a lender later.
If you would like help turning a rough crop list into a working budget and a few checks that AI can run on it, that is the kind of job an AI implementation consultation covers; see the contact page.
Questions new market gardeners ask about the plan
What is a bed-metre and why plan with it?
A bed-metre is one metre length of a growing bed of your standard width. It lets you compare crops on the same scale: yield, revenue and labour per bed-metre. Space is your limit on a small plot, so revenue per bed-metre and per labour hour are the figures that decide what to grow.
Can AI tell me the yield I will get from each crop?
It can suggest typical ranges, but yields depend on your soil, climate, variety and skill, so treat any AI figure as a placeholder. Use your own records, a local growers' group or an agricultural extension service, and plan with a lower figure than the best case in the first season.
Should I include my own labour as a cost?
Yes, as a separate line at a wage you would have to pay someone else. Without it, the plan can show a profit that is really an unpaid job. Show both cash profit and profit after labour, and see what hourly rate your own time earns.
How many sales channels should a new market garden start with?
Two or three. A farmers' market, a box scheme and a few restaurant accounts behave differently on price, hours and risk, and mixing them smooths the weeks. More than three at the start splits your time across too many deliveries and relationships.
Do I need a formal business plan to start?
For a small plot you can start with a one-page crop and channel plan and a cash budget. A fuller written plan becomes necessary when you apply for a loan, a grant or a lease, or when a partner wants to see the numbers. Ask the lender or funder what format they require.
Further reads
- AI Chatbot for Agriculture: A Farm Supply Business's Setup and Cost — What an AI chatbot costs a farm-supply business.
- AI Automation for Agriculture: Rotas, Timesheets and Harvest Messages — Rotas, timesheets and harvest messages for seasonal crews.
- Best AI App for Farmers: Herdwatch, AgriWebb and Farmonaut Compared — Farm apps compared, for when records outgrow a sheet.
- Free AI for Agriculture: What a Small Farm Gets Without Paying — What a small farm gets from free AI tools.
- AI Chatbot for Farmers: Answers From Your Own Farm Notes and SOPs — Answer questions from your own growing notes.
- AI Assistant for Farmers: ChatGPT, Claude or a Farm App? — ChatGPT, Claude or a farm app: which fits?
- AI Crop Monitoring on a Small Farm: Does It Work Without Drones? — How satellite-based crop monitoring works without drones, what it costs per hectare, and why fields under about a hectare need extra care.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: all yields, prices, hours and costs are illustrative planning numbers for a worked example, not market data; replace them with your own records and local quotes.