Batch separation: the unglamorous decision that decides throughput

Ask what a scanner does and you will be told a pages-per-minute figure. It is a real number and it is almost never the constraint. In production capture the constraint is separation: how the system knows that these eleven pages are one agreement and the next four are a different one.

The three workable approaches

• Barcoded separator sheets: reliable, cheap, and the standard choice for mixed backfile
• Patch codes: faster to prepare where you control the source paper, common for daily inbound
• Content-based separation: the system detects a new document from layout and first-page features, best where preparation time is scarce and document types are consistent

Most sites end up with a mix. Backfile boxes get separator sheets during preparation because the paper is unpredictable. Daily post uses content-based separation because volume per batch is lower and document types repeat.

Where the time actually goes

Preparation is the real cost centre: removing staples, flattening folds, pulling out fragile originals for flatbed capture, inserting separators. On a typical legal backfile, preparation takes several times longer than the scanning itself. This is why buying a faster device rarely changes the weekly total, and why preparation deserves the same design attention as the pipeline.

Measure it directly. Time a representative box from opening to verified output and note where the minutes go. Firms that do this almost always discover the same thing: the machine is idle for most of the day, and the answer is not a bigger machine.