If your forecast is consistently above or below your actual harvest, one or more inputs do not match what is happening in your greenhouse. Work through these checks in order.

Are your rounds % and color scale set to what you actually harvest, not what you plan to harvest? Source uses these values directly. If rounds % is set higher than your actual rounds, the forecast will overpredict. If color scale is set lower than the ripeness you actually pick at, the timing shifts and the weekly totals will be off.

Update your strategy at the start of each week to reflect actual practice.

Is the registered fruit weight accurate and recent? Predicted kg = predicted harvestable fruits x fruit weight. If the fruit weight in Source is higher than reality, the kg prediction will be too high even if the fruit count is correct.

Register fruit weight as frequently as possible. Check that the value entered is single fruit weight in grams, not truss weight or total sample weight.

Are the measurement plants representative of the full greenhouse? If you measure stems in a section with higher fruit load than average, Source will overestimate harvestable fruits across the entire cultivation. Spread measurement plants across the greenhouse and avoid clustering near edges, heating pipes, or other non-representative areas.

Is waste being registered accurately? If waste is underregistered, Source subtracts less than the actual loss from the gross prediction, resulting in a net prediction that is too high.

Is your crop going through an unusual event, such as virus pressure, heat stress, abnormally low radiation, or a disease outbreak? These conditions cause fruit abortion, stunted development, or reduced set that Source cannot anticipate until the effects show up in several weeks of measurement data. During these periods, the forecast may overpredict until Source catches up.

If you are in the first weeks of harvest, some inaccuracy is expected. Source starts with variety-level assumptions and calibrates to your specific cultivation as actual harvest data comes in. See Why is Harvest Forecast less accurate in the first weeks of harvest?

Use Harvest Forecast Analytics to compare predicted values (fruit weight, harvestable fruits, strategy settings) against actual values week by week. This shows where the gap between prediction and reality originates.

If you cannot identify the cause, start a chat conversation within Workspace and ask for your AISS. Include your cultivation ID.