This is a resolve-the-paradox question. Two facts are given and must both be accepted as true: fact A, the average time to resolve a problem fell, and fact B, the average satisfaction rating fell. The credited answer must supply a hidden distinction under which both can be true at once, without denying either and without explaining only one side.
The key insight is that the two averages are computed over different mixes of cases once the portal exists. The portal siphons off the easy problems, which customers now solve themselves quickly. Those fast self-service resolutions enter the resolution-time pool and pull the average down, which is fact A. But the cases that still reach a human agent are now disproportionately the hard, frustrating ones, and those are the experiences customers actually rate, so the satisfaction average falls, which is fact B. A shift in the difficulty composition of cases reconciles both facts.
(C) is correct: it names that composition shift. It is consistent with the faster average resolution time, since the easy cases are gone from the agent queue and resolved fast through the portal, and it is consistent with the lower satisfaction, since the remaining rated experiences are the difficult cases. It dissolves the tension by identifying the hidden variable, case difficulty.
(A) is a tempting half-answer. It uses the same diversion idea as (C) but only accounts for the faster resolution time; it says nothing about why satisfaction fell, so the conflict remains. Explaining one fact is not resolving the paradox.
(B) deepens the paradox rather than resolving it. If the portal is well liked and recommended, the decline in satisfaction is more surprising, not less. Positive-sounding feedback feels supportive but cuts the wrong way.
(D) reports a trend for one subgroup, the customers who resolved their problems through the portal alone, and that trend can be true alongside both stem facts, since a rise within one subgroup is a different quantity from the overall average across all rated customers. The problem is that it does no explanatory work: it gives no reason why the overall average fell. If anything, a rise for portal-only customers means the ratings from the remaining customers must have dropped even more steeply for the overall average to decline, and (D) offers nothing to account for that steeper drop. A choice that leaves the aggregate decline unexplained has not resolved the paradox.
(E) is irrelevant. Whether the portal was built in-house or licensed says nothing about why faster resolution coincided with lower satisfaction; it is a topic shift consistent with both facts but doing no reconciling work.