The standard advice for principle-driven loops (Amazon's fourteen leadership principles, or whatever values framework the company publishes) is to prepare one or two stories per principle. Take that literally and you end up with twenty-plus rehearsed anecdotes. Which produces exactly the outcome bar raisers are trained to catch: shallow, over-fitted stories that collapse the second time someone asks why.
Candidates who do well run the mapping in the other direction. A small set of deep stories, each mapped to several principles. A story matrix.
Why fewer stories win
A bar raiser probing three layers deep doesn't care that you have coverage. They care whether the story in front of them has depth. Five or six genuinely substantial stories, known all the way down (context, alternatives, numbers, aftermath, lessons), will each demonstrate four or five principles on their own, because real work is like that. A good migration story contains ownership, dive-deep, disagreement, and delivery all at once. Coverage falls out of depth. It never works the other way around.
And repetition across the loop is fine. Interviewers compare notes on consistency, not variety, so telling your best story in two rounds with different emphasis is normal and safe. Contradicting yourself between rounds is what kills loops.
Building the matrix
Pick your five or six: the project you're proudest of, a failure with real consequences, a conflict about substance, a time you were wrong, a time you delivered under a constraint that hurt, and one story where you went far outside your lane. For each, write a fact sheet. Baseline, your specific actions, quantified result, what you'd do differently.
Then draw the actual grid. Stories as rows, the company's principles as columns, a mark wherever a story genuinely demonstrates a principle. Two things show up almost immediately. First, which story serves which question, so when "tell me about a time you took a risk" arrives you're picking from a known menu instead of improvising. Second, and more useful: the empty columns. If nothing you've prepared shows frugality, or hiring and developing others, you have a targeted gap to fill from your history before the loop rather than discovering it live.
Using it in the room
The matrix is prep infrastructure, not a script. In the room, the work is selection and emphasis. The same migration story answers an ownership question if you stress the unrequested cleanup you drove, and a customer-obsession question if you stress why the latency mattered to users. Same facts, different lens. This is exactly what over-fitted per-principle stories can't do. They were built for one column.
One warning: don't announce the mapping. "That's a great Dive Deep example, and speaking of Dive Deep..." sounds coached. Tell the story and let the interviewer classify it. They're better at it than you, and it's their job.
The test of a good matrix
Have someone fire principle-style questions at you in random order before the loop. If every question maps to a row within a couple of seconds, and you never catch yourself reaching for a story you half remember, the matrix works. If two questions in a row land on the same thin anecdote, you need a deeper row, not a wider grid.