Short write-ups on the questions that surface once an organization starts measuring engagement seriously: how response thresholds actually protect the people answering, what notification cadence does to participation, and why recognition compounds across a quarter. Written by the practitioners who designed the instrument.
One measures the conditions people work under, sampled often enough to show drift. The other records a judgment about an individual at a fixed point. What each is competent to decide, and what running them as one instrument costs.
Read the distinction → 🔁 Multi-rater methodNot performance — observed behavior from positions with different vantage points. Rater-source variance, why self-versus-other divergence is the most diagnostic column on the page, and where an average stops being an average.
Read the method → 💬 Qualitative methodTheme frequency over intensity, a single articulate outlier versus a distributed pattern, and why the instinct to identify an author costs more than the comment supplied. Includes a five-step protocol for a first read, and how to respond so the channel keeps working.
Read the protocol → 🛡️ TrustManagers see response and participation rates; they never see attribution. The two-path data model, the five-respondent floor, and why evaluations disclose identity while cadence instruments do not.
Where the line sits → 📬 TimingMost non-response is an intention that quietly decayed, not a refusal — which makes timing the cheapest lever on response rate. What a follow-up communicates beyond its words, and why participation compounds in both directions.
Read the argument → ⭐ RecognitionSuccess is silent: problems escalate on their own, good work generates nothing. Why peers see contribution that reporting lines structurally miss, what the karma system is actually for, and how to keep a public wall from becoming a leaderboard.
Read the case →