The situation
A bakery was receiving complaints about bread that was well inside its printed date. The product was not mouldy and was entirely safe. It was firm, dry in the mouth and crumbling under a knife, and customers were throwing it away on day four of an eight day life.
Why it is hard
Shelf life is two problems and they are routinely managed as one. Microbial shelf life sets the date on the pack. Eating quality shelf life ends the repeat purchase. They run on different clocks and the second is nearly always faster.
Staling is not spoilage. Nothing has grown and nothing has gone off.
The starch that gelatinised in the oven slowly reorders itself on the shelf, firming the crumb and pushing water out as it goes. Which is why a mould inhibitor, however good, does nothing about it: it extends the date that can legally be printed and leaves the crumb to collapse on its original schedule.
What was done
Two targets were set and measured separately: a microbiological limit and a texture limit. Firmness was measured across the whole shelf life rather than at the end, because the customer meets the product in the middle. The anti staling route was chosen against the storage conditions the product actually sees, since retrogradation is fastest just above freezing and a chilled chain ages bread faster than an ambient one.
The two halves were then balanced against each other, because moisture content alone does not establish the water activity or microbiological stability of the crumb.
What it shows
That a specification naming one shelf life has specified half the product, and that the half it leaves out is the half the customer judges.
