Description
Stage 2 of the forward-thinking loop: combine tree testing (where), Slip/Lapse/Mistake error taxonomy (why) and signal-to-noise ratio (clarity) into a joint display with decision bars.
Diagnosis locates breakdowns and explains them using three lenses the playbook calls map (tree test), mechanism (taxonomy) and clarity meter (SNR). An unmoderated tree test quantifies findability; moderated think-alouds on the same tasks code each breakdown as Slip, Lapse or Mistake with two coders and Cohen's κ; and an SNR audit measures how much of the screen and of user attention goes to essential elements. Results are synthesised in a joint display linking numbers to quotes and fixes, and judged against diagnosis decision bars: tree direct success ≥ 85%, top misroute ≤ 10%, first-click accuracy ≥ 75%, element SNR ≥ 0.60 and behavioural SNR ≥ 0.70. Qual rigour comes from a participant × episode evidence array sorted by first misroute and error type, and recursive relationship mapping that exposes cascades such as Mistake → extra clicks → frustration → later lapse.
Attributes
- Discipline
- Community-Proposed Frameworks
- Loss type
- Community-proposed
- Tested
- Not yet tested
- Stage
- Evaluation
- Data
- Mixed
- Effort
- Weeks
Steps
- Prepare 6–10 realistic tasks with no label leakage ('Set a price alert for ABC', not 'Go to Price Alerts').
- Build current and proposed text-only trees of depth ≤ 3 and run an unmoderated tree test with N = 80–150, randomising task order.
- Run 10–15 moderated think-aloud sessions on the same tasks and code each breakdown as Slip, Lapse or Mistake; double-code 30–40% and require κ ≥ 0.70.
- Audit SNR on key screens: count essential versus total elements, and compute behavioural SNR from eye, click or cursor dwell if available.
- Build an evidence array (rows participants, columns first click, path nodes, error code, quote, time to first click, SNR snapshot) sorted by first misroute then error type.
- Synthesise a joint display (quant signal → qual mechanism → prescribed fix) and check against the bars: direct ≥ 85%, misroute ≤ 10%, first-click ≥ 75%, element SNR ≥ 0.60, behavioural SNR ≥ 0.70.
- Write a mini joint assertion for Stage 3, e.g. 'Because Mistakes at Notifications co-occur with low SNR and drive extra clicks, moving Price Alerts to top level will reduce abandonment hazard.'
Sources
- Vivek Gopinath (2025), Forward-Thinking UX Research Playbook: A Mixed-Methods, Decision-Ready Approach from Prediction to Validation
- Vivek Gopinath (2025), Tree Testing and Error Taxonomy: A Complete Chapter for UX Researchers
Also known as
- Stage 2: Diagnose with Tree Testing, Error Taxonomy, and SNR
- Map, mechanism and clarity meter
Categories
- Community-Proposed Frameworks