As a few people have already mentioned - there are a lot of factors that could account for the 95-105% range. Whether its your bulk systems, or just the accuracy of your traceability system. Some companies also want you to treat this as a more real life scenario in which you would encompass of buffer for your recall, including a couple batches before and after the affected ingredient/ material was used.
One example of the +/- range is due to bulk/ comingling of ingredients. If you are asked to trace 50,000 pound lot of bulk dry ingredient you received in, but it was placed into a silo with 2 other 50,000 pound shipments, you will likely not have an exact 100% recovery rate. Even if you performed the mock trace for the whole silo - the weights will likely not add up to be the exact amount you received in. There could be leaks or scales being slightly off that you would just not match the weight.
Also short shelf life products tend to have some issues. Like in bakery - we used the same color kwik lok (plastic tab) 2 days in a row. If orders changed, we would be running the same product 2 days in a row with the same color tag. The way our trace system was set up, we could not distinguish between which production date was pulled so we would have to account for both lots which increases the amount recalled. It was a CI opportunity, but if there was ever a recall event, everything would be pulled from the shelves so that's when the auditor's exercise would not match what we would do in the event of a recall event. We could still trace everything - but in reality everything from from the depots and their stores would be pulled off the shelves. Just a business decision to protect consumer incase investigations expand the scope.