Consistency at volume
Updated
On this page7
Roast to a measurement, not to an appearance
Time and temperature recorded, not judged by color through an oven door. Color varies with the lot even at an identical roast, so using it as the control means chasing a moving target and producing genuinely different roasts in the attempt.
Lock the lot into the record
Nuts vary by origin, season and lot, and the same roast gives a different result on a different lot. Recording which lot went into which batch is what lets you tell a roast drift from a raw material change, and those have completely different fixes.
Measure the batch at the end of the grind
Temperature and, if you can, weight in against weight out. A batch that finished much warmer than usual will behave differently in the jar, and that single column explains a lot of texture complaints that otherwise look random.
Fewer products, made the same way
The consistency gains from narrowing a range are larger here than in most crafts, because every additional product adds a changeover, a cleaning record and another roast profile to hold steady. A tight range made well is a better wholesale proposition than a wide one made variably.
What a buyer will ask
Whether the next delivery matches this one, and whether you can trace a jar back if something goes wrong. Both are answered from records, and in this category the second question gets asked more often than in any other property here.
Taste against a retained jar, not against memory
Keep one jar from each batch and taste the new batch alongside the last good one. Drift is easy to miss when each batch is compared only with the one before it, and a reference from three months ago catches the slow movement that nobody notices week to week.