## 1. Verification
- 13 records (I1–I4, R1–R6, O1–O3) = 11 unique participants (U01–U11). R2 = I2 (U02) and O2 = I3 (U03) are the same people and events, so they are not new participants (13 − 2 = 11).
- Purposive sample: no rates or population estimates; counts of people only.
- Only U06 (R3) finished independently. U02 (colleague explained), U11 (facilitator explained), U10 (facilitator handed over a mouse), and U04 (someone else clicked) finished with help, so none count as independent completion.

## 2. Evidence by theme (observation = source wording; interpretation = inference)
| Theme | IDs | Direct observation | Interpretation (inference) | Unique | Counter-case | Open question |
|---|---|---|---|---|---|---|
| First-team “Next” | I2, R2, O3 | U02 “누르면 결제가 되는 줄 알았다” (thought it would charge); U11 “다음 버튼 앞에 멈췄다” (stopped before Next) | Plan text may have hidden the step’s purpose | 2 (U02, U11) | U06, R3: “혼자 끝냈다” (finished alone) | U02 cannot reproduce; screen difference vs U06 unchecked; no actual charge reported |
| Keyboard save | I4, O1 | U04 “저장 버튼으로 갈 때 순서가 화면 밖으로 넘어갔다” (order left the screen); U10 “Tab 이동이 저장 버튼을 지나쳤다” (Tab skipped Save) | Possible Tab-order problem | 2 (U04, U10) | None in data (not proof none exist) | Key presses uncounted; shortcut existence unknown |
| Project not visible | I3, O2, R4 | U03 “바뀐 것은 초대 설정이었다” (the invite setting changed); U07 “초대를 다시 받고 보이기 시작했다” | Cause was invitation; no evidence of a search defect | 2 (U03, U07) | I3 “비슷한 이름의 다른 프로젝트는 계속 보였다”; R4 “검색이 느렸는지는 확인하지 않았다” | U03’s pre-change query and recording absent |
| Upload completion | R1, R6 | U05 “화면이 오래 멈춘 듯했다. 시간을 재지는 않았다” (felt frozen; not timed); U09 “완료 안내를 못 찾아 같은 파일을 다시 올렸다” (re-uploaded) | Completion signal may be unclear; speed cannot be judged | 2 (U05, U09) | R1 “작은 파일에서는 못 느꼈다” (small files fine) | Did U09’s first upload fail (“확인하지 못했다”)? Actual delay? |
| Similar document names | I1, R5 | U01 “이름이 비슷할 때는 날짜를 보고 고른다”; U08 “오래된 초안을 열었다”, then re-picked by date | Self-recovered via dates; likely lower risk (inference) | 2 (U01, U08) | I1 finds docs “대체로”; “처음 온 사람도 그렇다고 보지는 않는다” (do not generalize to newcomers) | Sorting? U08 “모르겠다”; U01 never tried very many files |

## 3. Two problems to check first
1. **Keyboard save**: both unique participants could not complete the core action, saving, alone. U04: “혼자 저장까지 할 수 있는지가 내게는 중요하다.” Check: reproduce the Tab order; verify whether a shortcut exists (do not assume one).
2. **First-team “Next”**: two people stalled or quit early (U02 “창을 닫았다”). Since U06 finished alone, find which conditions trigger the confusion.

Priority reflects impact and evidence strength, not frequency. Invisible projects point to invite settings first; upload delay was never timed, so it ranks lower.