Amazon swapped Grubhub for Uber Eats and said it "continuously evaluates" its partners. The question that tests for it.
On November 5, 2026, TechCrunch reported that Amazon had made Uber Eats its official restaurant delivery partner for Prime members in the United States, effectively ending its existing deal with Grubhub. Under the partnership announced Thursday, Prime members will be able to order food from Uber Eats directly through Amazon.com and the Amazon Shopping app, won't need to create an Uber account, and won't pay a delivery fee on orders of more than $25, the companies said. The Grubhub agreement ends Thursday, according to an Amazon spokesperson, who said the company "continuously evaluates its third-party offers." The relationship being closed was not a casual one. TechCrunch notes that Amazon first struck a deal with Grubhub in 2022, expanded it two years later, and reached a deal with Grubhub's then-owner to increase its equity stake to as much as 18% under what the outlet describes as a "performance-based agreement." Prime members who had already activated Grubhub+ will still be able to use that membership through December 31, 2027.
Strip away the brand names and what remains is a sequence every manager eventually has to run. A relationship that began with investment and optimism was measured against results, found short of an available alternative, and ended, with a runway for the people caught in the middle. Amazon's stated reason is the tell: it said it "identified an opportunity to deliver an even better restaurant delivery experience." That is the language of a standard applied continuously, to a partner who was once good enough and no longer is. Managers run the same evaluation on people, except the stakes are personal, the data is murkier, and the manager's own conduct is part of what gets judged. That is why a question reported in the L8 Loop corpus is simple to ask and hard to answer well.
Why they ask it
The interviewer isn't curious about your HR process. They are testing three things at once. First, whether you notice and act, because the most common management failure is not harshness but drift, where a problem is tolerated for two quarters because confronting it is uncomfortable. Second, whether you diagnose before you judge, since "low performer" is a label and not a cause, and the causes range from a bad brief to a bad fit to a bad month at home. Third, whether you can hold a standard and a person at the same time: be direct about the gap, fair about the path back, and honest about the consequence if the path isn't taken. A senior manager is also being checked for something subtler, which is whether they protect the rest of the team. Every week an unaddressed gap persists, the strong performers are quietly paying for it.
The trap
Weak answers fall into one of two ditches. The soft ditch is the coaching story with no edge: regular one-on-ones, encouragement, a mentor assigned, and a vague ending in which "things improved." There is no observable standard, no date, and no description of what would have happened if things hadn't improved. The interviewer hears a manager who would let the situation run indefinitely. The hard ditch is the opposite: the answer is a termination story told with some relish, where the diagnosis step is missing and the person was gone within a month. The interviewer hears someone who mistakes decisiveness for judgment. There is a third, quieter failure, which is the answer built entirely around process. "I put them on a plan" is not an action, it is a form. The strong answer names the specific gap, the specific expectation, the specific window, and what the manager personally did inside that window.
Applying STAR-T
Situation. Keep it concrete and neutral. A senior engineer on a platform team of seven had, over about a quarter, gone from the most reliable reviewer on the team to someone whose work routinely slipped past its estimates and whose pull requests needed two or three rounds of rework. Two peers had mentioned it indirectly. Say what the signal was and how you noticed it, because noticing is part of the skill.
Task. State the manager's job in one sentence, with the tension in it. The task was to find out whether this was a capability problem, a motivation problem, or a circumstance problem, and to get the engineer back to the team's standard within a defined period without letting the rest of the team carry the cost in the meantime.
Action. This is where the answer is won. Describe the diagnosis first: a direct conversation that opened with specific examples rather than a general impression, and a genuine question about what had changed. In this case, a reorganization had moved the engineer onto work they had never chosen and found uninteresting, and they had not said so. Then describe the agreement: two explicit, measurable expectations, written down and shared, with a six-week window and weekly checkpoints. Then the part most candidates skip, which is what the manager changed. Here it was reshaping the engineer's scope so that half of it was work they had asked for, and rebalancing review load so that peers weren't absorbing the rework. Finally, say plainly that the consequence of not meeting the standard was named in the first conversation, not the last.
Result. Give the outcome in the same measurable terms you set. Estimates were being met by week four, rework dropped to a single round, and the peer complaints stopped. If the honest outcome was that the person left, say that too, and say how the exit was handled. An interviewer trusts a candidate who has told the uncomfortable version at least once.
Trade-off. Name the cost. Six weeks of close attention to one person is six weeks of less attention to six others, and a shorter window would have been fairer to the team but less fair to the individual. A strong answer says where you set that line and why.
The follow-up that breaks weak answers
"How long did you give it, and how did you decide that was enough?" Candidates who told a soft story have no answer, because there was never a window. Candidates who told a hard story have a number but no reasoning. The good answer has both: a period long enough for a real change in behaviour to show, short enough that the team isn't subsidizing it, and a stated criterion, agreed up front, for what "enough" looked like. Prepare that sentence before the interview. It is the one the interviewer is actually waiting for.
Score your answer against the director’s bar
Q: How do you manage a low performer in the team?
Bank your low-performer story with the window and the criterion written down, then rehearse the follow-up until the number and the reason come out together. Try it free →
