Often we hear exciting news of crop yields from experimental trials, but then the gains don't show up in the real world. Rachid Laajaj tells Tim Phillips how a group of researchers solved this puzzle, and what this means for how we conduct trials in future.