This is a resolve-the-paradox question. Two facts are given as true and appear to conflict: the average amount spent per recorded transaction rose substantially, yet the chain's total revenue declined over the same year. A correct resolution must be consistent with both facts at once and supply the hidden distinction that lets both be true without contradiction. The key is that the per-transaction average is computed only over the transactions that still happen, while total revenue depends on how many customers there are in the first place.
(D) is correct. If many former customers who made small purchases stopped shopping there because they would not use the app, then two things follow at the same time. First, removing those small-basket shoppers raises the average of the remaining transactions, which matches the rise in spending per transaction. Second, losing a large group of paying customers removes their purchases from the total, which matches the decline in revenue. One hidden change, the loss of low-spending customers, explains both facts without disputing either.
(A) explains only the higher spending per transaction. It leaves the revenue decline unexplained, and if app adopters really do spend more, it actively predicts that revenue would rise, which cuts against the stated fall.
(B) addresses how the per-transaction figures were measured, not why the two trends diverged. A method that consistently overstated the true amounts would inflate the figures in every period alike, so it cannot account for the change: the recorded average still rose, and the overstatement gives no reason why it did. It also says nothing about total revenue, so the decline is left untouched. Taken as true, this choice leaves the rise and the fall exactly as puzzling as before.
(C) points to the higher cost of running the app. That is consistent with both facts but irrelevant to them, since cost affects profit, not the question of why revenue fell while per-transaction spending rose.
(E) would lower the amount spent per transaction on frequently bought staples, which pushes against the stated rise in spending per transaction. It deepens the discrepancy rather than resolving it.
The answer is (D).