View all features
View all suite

Growing & production

Trade, export & supply

Retail & e-commerce

Beyond horticulture

View all industries
Library Contact sales
EN IT
library · field guide

Last year's plan, again.

Nobody sits down and decides to plan on a hunch. It just happens: the order form comes round in January, last year's number is right there in the column, and there is no time to argue with it.

the lag

You commit long before you find out.

In most trades the gap between deciding and knowing is a week. Grow a 3-litre shrub and it is two years — which makes the production plan the most consequential guess in the business, and usually the least examined.

the quiet default

There is no such thing as not forecasting.

Every nursery forecasts. The only question is whether it does so on purpose.

Sowing the same number as last year is a forecast — it is the prediction that next season will look like the one before it, made silently, by nobody in particular, and never checked afterwards. It has the enormous advantage of taking no time at all, and the enormous disadvantage of being wrong in exactly the ways that hurt: it repeats last year's glut and last year's shortage together, faithfully, year after year.

A real forecast is not a prophecy and nobody expects it to be right. It is simply a number you write down in advance so that you can find out how wrong you were. That is the whole trick: a written guess can be compared to an outcome, and an unwritten one can't.

Growers resist this because the horticultural year feels too uncertain to plan — the weather, the trade, the fashion for a particular genus. But uncertainty is the argument for writing it down. Nobody needs a plan for the parts they already know.

You sold what you had, not what they wanted.

start with what happened

You sold what you had, not what they wanted.

The obvious place to start is sales history, and it is the right place — as long as you read it honestly. Last season's despatch figures record demand capped by supply, and the cap is invisible.

The line that sold out in week three didn't sell 4,000 because 4,000 was wanted. It sold 4,000 because 4,000 was all there was. Meanwhile the line that sat until August looks like weak demand when it may just have been a fortnight late.

A few corrections turn history into something worth planning against:

  • Mark your stock-outs. Any line that ran to zero before the season ended is under-recorded, and by an unknown amount.
  • Track the asks you couldn't fill: the enquiry, the trimmed order line, the substitution. That is the only direct evidence of demand you never met.
  • Look per customer as well as per product. Three garden centres growing their orders and one large account winding down net out to "flat", and flat is a lie that costs you a customer.
Plan backwards from the finished plant.

the arithmetic

Plan backwards from the finished plant.

The commonest planning error in horticulture has nothing to do with optimism about demand. It is forgetting that plants die on the way.

A plan built forwards ("we'll stick in 10,000 and see what we get") quietly hands the shortfall to whoever is selling in two years' time. A plan built backwards starts from the finished, saleable number you intend to have, and works up through every stage of the crop, applying the loss each one actually takes.

Ten percent at sticking, five at potting on, another five overwintering: that is a fifth of the crop gone, and nobody along the way did anything wrong. If you need 8,000 sellable plants at the end, you do not start 8,000, and you do not start 10,000 either — you start what the arithmetic says.

The precondition is knowing your real losses rather than the ones in the textbook. Record what dies, at which stage, every cycle. Nurseries that do this for two seasons usually discover their true yield is several points off the figure they had been quoting themselves for a decade — and once it's honest, the plan stops being short every spring.

the second dimension

A plan without dates is a wish.

How many is only half the question. Every crop occupies ground for a known number of weeks and becomes saleable in a particular one — and a plant that is right in quantity but four weeks late has missed the only fortnight anyone wanted it.

This is where a production plan stops being a spreadsheet of quantities and becomes a calendar. Work back from the week the trade wants the crop, through the growing time the protocol actually takes, to the week you have to start — and then keep going backwards, because the start date has its own dependencies.

Seed, plugs and liners have lead times and finite supply; the good ones are contracted out months before you get round to ordering. Pots, compost, labels and trays have to be in the yard before the crew arrives. Bench and field space has to be free — and it isn't free if last year's overhang is still standing on it. Each of these is capable of turning an achievable plan into a late one, and each is knowable in advance.

Two habits keep the calendar honest. First, plan in ranges rather than points (a good, likely and poor case for each crop) because a single number invites everyone to treat a guess as a commitment. Second, decide in advance which lines are your shock absorbers: the items you would rather buy in finished from a fellow grower than risk growing to a peak. Knowing that before the season starts is planning. Discovering it in May is panic buying.

The plan improves at the post-mortem.

the review nobody does

The plan improves at the post-mortem.

Nurseries that forecast well are rarely better at guessing than the ones that don't. They are better at looking back.

The habit is unglamorous and takes an hour: put the plan and the actual side by side, per crop, once a season. You are looking for the pattern, and there is always a pattern.

  • One propagator's crops consistently yield below plan; the protocol needs changing rather than the person.
  • A whole product group is always over-grown, because it was over-grown in the year everyone remembers.
  • Sales for a given genus land three weeks earlier than they used to, every year, and nobody has moved the sowing date.

None of that is visible in a single season. All of it is obvious across three. A rough forecast that gets reviewed beats a clever one that doesn't.

A forecast sales never saw is just an opinion.

the other half of the building

A forecast sales never saw is just an opinion.

In a great many nurseries the growing team plans production and the sales team sells it, and the two conversations happen months and one wall apart. The growers plan from history because it is the only information they have. The sales team, meanwhile, knows perfectly well that a large customer is doubling its bedding order and another is closing two sites.

Closing that gap is the single cheapest improvement available to most growers, and it needs no software at all to start — just a standing meeting where the plan is read out before it is committed, and the people who talk to buyers get to argue with it.

Beyond that, the trade has a stronger tool it under-uses: selling the crop before it exists. Contract growing, forward orders and pre-season commitments turn a forecast into an order book. They cost some margin and some flexibility, and in exchange the most expensive uncertainty in the business (how much of this will anybody actually take) is someone else's problem before you sow.

atlas core

How Atlas Core supports growing to a plan

Atlas Core carries the production plan from the recipe through to the batch on the bench:

  • Protocols — a reusable recipe per product, with its stages, each stage's duration, the reference batch size, the expected loss and the resulting overall yield percentage.
  • Work backwards from the finished number — the yield chained across the stages gives the expected output of any starting quantity, so a cycle's real finished count is on screen before you commit to it.
  • Production cycles with a start date and an end date computed from the protocol's stage durations, so a plan has weeks on it as well as quantities.
  • Loss recorded as an event against the stage it happened in, feeding a live loss rate per cycle — your actual yield rather than the textbook one.
  • A first-stage material check at creation that compares the protocol's requirements against free stock at the growing location and refuses to start a cycle you cannot resource, then reserves each later stage's materials on entry.
  • Estimated and yield-adjusted output value per cycle, plus overview cards for active cycles, the next crop due and the expected value standing in production.
  • Previously-ordered products per customer, filterable year-to-date, previous year or last three months, and exportable — the corrected demand history a plan is built from.
  • Monthly sales targets by team or salesperson with target-versus-actual attainment, so the commercial forecast and the growing plan are looking at the same year.
  • Purchase requests and purchase orders for the lines you decide to buy in rather than grow, receipted into stock as batches like anything else.

Read further

Sow to a number.

Protocols, yields, production cycles and demand history — see how it fits your nursery.

Talk to us