How to Separate Observation from Inference
This playbook helps leaders keep evidence clean. The work is simple but demanding: write what happened before deciding what it means, label your interpretations, and check the predictable ways judgment can bend a record.
Developing
Start here. Build the foundation.- 1
Write the action or the actual words before any conclusion. Use entries like 'spoke over the customer twice while they raised concerns' instead of 'was dismissive.' The record is clean when another person can read it and form their own view.
- 2
Create a separate line or field for interpretations. If you think the behavior showed impatience, label that as your judgment rather than blending it into the factual note. This keeps the evidence available for inspection even if the interpretation changes.
Proficient
Build consistency and rhythm.- 3
When a strong first impression forms, list the moments that support it and the moments that challenge it before you score. Do this for favorable impressions too. The signal is that your first read no longer decides what you notice next.
- 4
Before summarizing, scan the whole observation window from start to finish. Ask whether one recent or vivid event is pulling the score harder than the full pattern supports. Adjust only as much as the evidence warrants.
Mastered
Operate at the highest level.- 5
Name the bias most likely to affect the observation and the correction you used. It might be giving extra benefit to someone you like, explaining behavior by personality instead of situation, or drifting toward the middle of the scale. Mastery means you can teach another observer to run the same check.
Common Pitfalls
Avoid the common failure modes.- Recording labels like 'lazy,' 'aggressive,' or 'engaged' without the observable actions behind them. Labels are not evidence.
- Letting a strong opening moment set the tone for the entire rating. First impressions are useful signals, not final verdicts.
- Explaining a one-off slip as a character flaw. Ask what the situation required before deciding what the behavior means.