Of everything that predicts whether a cash-on-delivery order converts, the strongest single variable is not the product, the price, the city or the agent. It is how long the order sat before anybody rang.
The mechanism is not mysterious. Somebody who ordered eleven minutes ago is still in the frame of mind that made them order. Somebody who ordered yesterday evening has slept, checked their balance, seen it cheaper somewhere else, or simply forgotten. The call is the same call. The person is not the same person.
What that means for a queue
Most order lists sort by order number, which is to say by arrival, which is to say they look right and behave wrong: the oldest order is at the top whether it has been called four times or never. An agent working top to bottom spends their morning on orders that have already been chased and their afternoon on fresh ones that have gone cold while they waited.
So the queue here sorts by time waiting within the stage of the chase. Never-called orders are their own list, ordered by age. Tried-once is another. A callback promised for six o'clock surfaces at six o'clock and not before.
And for the age column
The waiting time is shown in minutes and hours, not as a date. “2 Sep 14:02” requires arithmetic. “7 min” and “2 days” do not, and the colour behind them does the rest of the work before anybody has read a word.
And for notifications
It is also why a new order makes a sound. An order sitting unseen because nobody happened to refresh is the same loss as an order nobody called, and the fifteen seconds between polls is cheap compared with the hour it saves.