Correlation is co-movement, causation needs a counterfactual. Without a randomized test you approximate one with a natural experiment, difference in differences, regression discontinuity, matching or instrumental variables, each with assumptions you must state.
How to answer it
Correlation says two things move together. Causation says changing one would change the other. The gap between them is the confounder: a third thing that drives both, like summer driving both ice cream sales and drownings.
A randomized experiment closes the gap by construction, because randomization balances every confounder, known and unknown. When you cannot randomize, you borrow structure from the world:
Natural experiment: something outside your control assigned the treatment roughly at random, a policy rollout by region, an outage that hit half the users.
Difference in differences: compare the change in a treated group to the change in a similar untreated group over the same period. Assumes parallel trends before the treatment.
Regression discontinuity: users just above a threshold (a credit score cutoff, a free-shipping minimum) are compared to users just below it.
Matching or propensity scores: build a control group that looks like the treated group on observed features. Only handles confounders you can measure.
Instrumental variables: something that moves the treatment but affects the outcome only through it. Hard to find, and the assumption is untestable.
The interview move is to name the assumption each method rests on and what would break it, not to list the methods.
What they are checking: whether you can say why randomization works and what each observational method assumes.
Common mistake: presenting a regression with control variables as if it settled causation.