How to prove a biological input paid, using a field you already have

Most disappointment with biological inputs is really a measurement problem. A grower applies a product across the whole farm, the season is good, the yield is up, and nobody can say why. The following year the season is poor, the yield is down, and the product is blamed. Neither conclusion is supported.

The minimum design that works

A strip trial with at least three replications of treated and untreated, laid out along the direction of planting and harvested with a calibrated yield monitor or a weigh trailer. That is the floor. Below three replications, field variation swamps a 4% effect entirely.

• Run strips the full length of the field. Short strips concentrate soil variation.
• Alternate, do not block. Treated-untreated-treated-untreated captures the gradient; two halves capture the gradient and call it a result.
• Keep everything else identical. Same variety, same planting date, same nitrogen. One variable.
• Record the strip map before planting. Reconstructing it at harvest from memory is how good data becomes unusable data.

What size of effect you can actually detect

With three replications on a uniform field, a difference below roughly 3% is not reliably distinguishable from noise. On a variable field, that threshold rises to 5–6%. This matters because the mean response we publish for cereals is +4.1% — which means a single-season, single-replication comparison will frequently fail to detect a real effect, and will occasionally show a large effect that is not there.

If the trial design cannot detect the effect size you are looking for, running it carefully is worse than not running it, because it produces a confident wrong answer.

Reading the result honestly

Convert the yield difference to margin at your own sale price, subtract what the programme cost you, and write the number down before the next season's ordering conversation. Then repeat the trial. Two seasons of strip data on the same field is worth more than five seasons of whole-field impressions, and it is the only basis on which we are willing to ask a grower to renew.