Commit 50 tons. Deliver 35. Leaving money on the table.
This is the moment commercial growers know too well. The operation is running fine, but your forecast missed reality. The true cost isn't just lost volume, it's damaged buyer relationships, wasted Friday labour, and logistics booked against numbers that didn't hold.
Let's fix your forecast togetherhexafarms Data Platform
6-week harvest forecast
Harvest strategy simulator
Cooperative view
Yield Forecasting exists to close that gap. Not with a better number. With a number you can actually commit to.
Available for these crops:
Three ways to predict what your harvest will deliver
The question is not whether to forecast, it is whether your forecast is tight enough to actually commit to a buyer. That depends entirely on how it is built.
Spreadsheets and grower intuition
Weekly estimates from experienced growers combined into a single number for the sales team. Fast to run, with no setup cost, and often accurate when one experienced grower manages the same variety year after year. The challenge comes when that grower leaves, the variety changes, or changes need to be made dynamicaly.
Works well if:
- Volume is small and forecast misses are affordable.
- One senior grower with long variety history is always available.
- You are not supplying supermarkets on weekly volume contracts.
Standalone forecasting platforms
Dedicated forecasting tools connected to your climate computer via API. More structured than spreadsheets, but most still rely on industry-average yield curves. They know the weather, not your plants. If anything changes, you only find out once it is already too late.
Works well if:
- Close enough is good enough for your buyer commitments.
- Your operation is standard enough that industry averages hold.
- You do not need sensing or pest detection integrated.
hexafarms Yield Forecasting
Part of the hexafarms OS, built on the way you operate.
Trained on your plants, variety, and growing history, not industry averages. Using hardware installed in your operation we look at your plants the same way you do. You can see every building block that makes up the forecast and how reliable it is. Up to 30% lower variance than yield-curve methods, giving you forecasts you can confidently commit to.
Works well if:
- You supply supermarkets on weekly volume contracts.
- Labour planning, logistics, and pricing depend on forecast precision.
- You want a forecast that gets more accurate as time goes on.
Accurate early. Keeps improving as your data builds.
Want to see what this looks like for your crop?
Speak to our teamWhy traditional forecasts keep missing
Every operation is different
Yield-curve based tools work well in standard operations with consistent varieties. The pain shows up when circumstances change, new varieties, unusual seasons, a climate that continues changing.
They look at averages, not your operation
Industry yield curves are built on historical averages across many operations. Your operation is not average. Your variety, your substrate, your microclimate, your management decisions, none of these are captured in a generic curve.
They measure climate, not plants
GDD knows what the thermometer said. It doesn't not know how many flowers set this week, whether fruit is colouring on schedule, or whether the vegetative-generative balance shifted after last week's heat wave.
They report a number, not how they got there
A forecast is built on trust, if a grower can't see what factors are making up their forecast it will not be used. It is a guess with a decimal point. If you don't know how much you can trust your forecast you still cannot commit to a buyer.
Why accuracy alone is not enough
Most forecasting vendors report a 90%+ average accuracy. But averages hide a dangerous truth.
A forecast that is 100% accurate one week and 0% the next still averages out to 50%.
You don't need a single blind guess. You need a forecast that tells you how it was built so that you can commit with confidence.
A mean accuracy figure tells you absolutely nothing about weekly performance and real-world unpredictability.
The mean looks fine on paper, but the variance completely destroys your ability to commit to buyers or plan labor.
This is why hexafarms reports every factor that goes into your forecast and what impact it has. This allows you to steer your crop in the way you want and see the impact on your bottom line in real time.
Three features. One complete picture of your harvest.
4-Week Yield Forecast
Harvest Strategy Simulator
Sales View
4-Week Yield Forecast
A rolling prediction of yield for the next four weeks, per crop and per department. Updated every time new sensor data or fruit counts come in. Reported with confidence intervals, never as a single number.
Harvest Strategy Simulator
Three levers you can pull to see how harvest volume and timing shift in real time: Harvest Cadence, Harvest Percentage, and Ripeness Stage. The simulation allows you to see exactly how much you will harvest if you change your harvest strategy.
Sales View
For sales teams and cooperatives: a roll-up view showing aggregated forecast across locations, with breakdowns per grower, per crop, and per week.
World-class systems require the highest data quality. Pure code cannot see your plants.
Show Me The ForecastWithout the Insight Stations and the AgNodes feeding the system, the forecast degrades to exactly what every other vendor offers: a historical average dressed up as precision.
Solar-powered cameras
Solar-powered climate sensors
Agronomical climate & plant analysis
The three principles that make this possible
↳ Probabilistic forecasting
Farm-specific training
Biology-driven models
The three principles that make this possible
Probabilistic, not point-estimate
Reliable ranges, not historical averages.
Traditional models offer a single static number, leaving you with zero safety buffer. We provide a forecast combined with a mathematically valid confidence band.
Smart limits:
Instead of predicting an exact 50 tons, we give you a secure range (e.g., between 48 and 52 tons).
Commercial freedom:
Commit to the lower bound (48 tons) with confidence, and safely negotiate trade on the upper potential.
Continuous tightening:
As your data builds, our forecast tightens by up to 30%.
Trained on your plants, not on averages
Tailored to your operation, not industry generalities.
Industry averages and past season yield-curves are just raw inputs, not final outputs. Our AI models learn directly from your unique operation.
Hyper-local validation:
The AI trains on your historical data and validates it live against your current season setup.
Absolute specificity:
Every calculation adapts dynamically to your operation, your exact crop variety, climate strategy, and harvesting decisions.
Climate adaptability:
Unpredictable anomalies like a heat wave won't break the model, they simply feed it more real-time data to improve the output.
Built on plant physiology, not simple climate data
Deep crop physiology breakdowns, optimising for your crop and your strategy.
Standard tools rely on historical averages, meaning any change in your operation will result in poor performance. hexafarms operates in a completely different scientific category.
80+ Live parameters:
We track actual fruit and flower counts, crop biomarkers, and precise microclimate conditions.
Dynamic growth tracking:
The model continuously factors in the microclimate history and the exact biological growth stage of the plant.
True physiology:
Your forecast is derived directly from real crop behavior, rather than approximated from historical averages.
Case Study
Full Case StudyStrawberry cultivation
Germany
"Where before we were working with uncertain estimates, we now have reliable forecasts that guide every decision."
Dominik Janssen
Managing Director, Edelrot GmbH
The situation:
Edelrot is a leading German strawberry grower, but manual yield estimates were highly reactive, relying mostly on grower intuition and basic weekly checks.
The trigger:
Wild forecasting errors caused costly product waste during surpluses and severe labor shortages during unexpected harvest peaks.
Looking ahead:
Three years into the partnership, Edelrot is expanding the setup to real-time climate sensing and pest monitoring. Yield forecasting was just the entry point, the ultimate goal is transitioning to a full greenhouse operating system.
The choice:
hexafarms Yield Forecasting, integrating fruit and flower counts, historical production data, climate, crop biomarkers, and 80+ parameters into live, per-week predictions.
The outcome:
Forecast accuracy soared to 85-90%. Waste and labor shortages were drastically cut, allowing perfectly aligned staffing, logistics, and consistent supermarket deliveries.
The ROI
4.0×
Case Study
Full Case StudyThe weekly average looked fine. Friday collapsed to −52%.
Strawberry cooperative
Germany
ELO is a cooperative of 26 strawberry growers in Germany with supermarket contracts, including Lidl, with financial penalties for missed delivery commitments.
Existing forecasting relied on manual weekly estimates from each grower, aggregated by the central sales team.
The situation:
Calendar week 27: across the week, deliveries ran 6% above commitment. But Friday collapsed to −52% of what had been promised to the buyer. What the sales team called "Vertriebsseitige Kürzung." The weekly average looked fine. Reality broke a supermarket promise.
The choice:
hexafarms is now in pilot with ELO across 14 strawberry growers, replacing manual aggregation with real-time per-grower forecasts that roll up into one reliable number the cooperative can stand behind.
Preventing one bad forecast already pays for the cost of an entire season
What you get:
- Forecast accuracy: 85–90%
- ROI for growers: 4–10×
- Up to 30% variance reduction versus statistical methods
- No upfront investment required
- Accurate from month one. Continues improving as data builds.
"hexafarms allows us to proactively sell our growers' produce, ensuring a better price over the entire season and more revenue for our growers."
Ricky Gommans
Business Analyst, ZON
Growers who already run hexafarms AgNodes do not pay twice for climate data. The sensor foundation is already in place, forecasting extends it. For growers starting with Forecasting, a sensing setup is included in the price.
The growers with the best forecasts are the ones with the best bottom lines.
Most growers start with Yield Forecasting as their first module, it solves the most visible commercial problem immediately. The data builds. The forecast improves. The other modules come later, on a foundation that is already working. Every module that is added improves the forecasting performance.