How do you audit a story-by-theme matrix for coverage gaps before an interview loop?

## Rows are for building; columns are for auditing The story-by-theme matrix earns its keep in the second reading. Building it, you work along rows — for this experience, which themes does it touch? Auditing it, you work down columns — for this theme, what have I actually got? Almost every hole a candidate discovers mid-interview is visible in the column view an afternoon earlier. For each column, count the cells marked *strong*. Three outcomes matter: - **Zero strong cells.** No rehearsed material for that theme at all. If the theme is one the role plainly cares about, this is the highest-value thing to fix. - **Exactly one strong cell.** Single point of failure. It answers the first ask and nothing after it. - **Two or more.** Covered, including under a follow-up. ## Why one is not enough on the themes that matter The most ordinary follow-up in a behavioral conversation is *'Do you have another example?'* — the second ask that empties a one-story bank. Interviewers use it deliberately, because one good anecdote can be luck or can be the only instance in a career, and a pattern is what they are trying to establish. Answering it with a pause and a visibly improvised half-example undoes the first answer more than it adds. That does not mean two strong cells for all eight themes; that is a bank of sixteen well-rehearsed answers nobody has time to build. Weight it: two on the themes the role emphasises, one elsewhere. ## How to weight the columns The posted role and the level you are interviewing at tell you which columns are load-bearing. A mobile role that stresses release quality and on-call will lean on incident and tradeoff themes; one that stresses growing the team will lean on mentoring and persuading people you do not manage. Reorder the columns so the load-bearing ones are on the left, and audit those first. When rehearsal time runs out, you want it to have run out on the columns that matter least. ## Filling a gap without inventing A blank column is filled by finding real material you had forgotten, and most people have more than they think. Productive places to look: - your own work log, task history and pull request or change history - retrospective and post-incident notes, which record exactly the moments interviews ask about - one-on-one notes, where the awkward interpersonal episodes are written down - release notes and changelogs, which reconstruct timelines you no longer remember - side work: an open-source contribution, a volunteer project, a university project if you are early in your career What does not work is stretching. Marking a cell strong because an experience is adjacent to a theme produces an answer that sounds fine for ninety seconds and then falls apart, because the follow-up questions probe the beat that is not there — the actual disagreement, the actual ambiguity — and there is nothing to say. A *usable* mark exists precisely so you can be honest with yourself about the difference. ## When a theme is genuinely empty Some themes have no material behind them yet, and that is a fact about your career rather than a failure of preparation. A first-job candidate may never have led an incident; a specialist may never have managed a cross-team disagreement. Prepare for that case rather than fabricating one: name the closest real thing you have, be explicit that it is smaller in scope, and say what you would do. That answer is respected. An invented one is not, and it typically breaks under the second or third question because invented material has no detail underneath it. ## Keep the audit cheap and repeat it The audit is a ten-minute exercise, not a project: count strong cells per column, mark the two or three weakest load-bearing columns, and spend the remaining preparation time there. Repeat it before each loop, because the columns get reweighted by the role and because the grid drifts — experiences you now remember only vaguely should be downgraded from strong to usable rather than left flattering you.