Unsafe acts are the behavior half of the act and condition equation, and they are only correctable once they are seen. The seeing is the job of the supervisor, and it is a skill, not a personality trait. The supervisor who knows what to watch, how to watch without disrupting the work, and how to turn the observation into feedback is the supervisor whose floor’s at-risk behavior falls. This post is the technique guide: the observation method that the behavior-based safety program on this site depends on.
What to watch
The observation checklist writes itself from the injury data and the unsafe act catalog. Operator behaviors: speed at intersections and in pedestrian zones, horn use at blind corners, load handling and stability, seatbelt and PPE compliance, and the inspection actually being done. Pedestrian behaviors: walkway use, intersection checks, and proximity to moving trucks. The list is finite, which is the point: the supervisor who watches ten behaviors, every observation, builds a data set that the weekly review can read. The how to identify unsafe acts question is answered by the checklist, because you see what you look for.
How to watch
The watching technique has three rules. First, watch the work, not the worker: the observation is about the behavior in context, not about judging the person, and the posture shows. Second, watch from a position that does not disrupt: stand at the edge of the aisle, at the intersection, where the observation does not change the behavior, because the operator who knows they are being watched behaves differently and the data lies. Third, watch the whole cycle: the approach, the maneuver, and the exit, because the unsafe act often lives in the transition, the speed carried into the turn, the load raised too early, the phone checked at the stop. The psychology of unsafe acts post on this site explains why the transitions are where the risk hides.
The observation record
The record is a simple tally: the behavior observed, safe or at-risk, the location, and the context. The tally does not need a paragraph; it needs enough to aggregate. The weekly roll-up shows the at-risk behaviors by type and location, which answers the two questions the program runs on: which behaviors are the problem, and where do they cluster? The behavior-based safety software on this site automates the tally with cameras and telematics, but the supervisor’s manual observations are the ground truth that the automation is measured against.
The feedback
The observation without feedback is surveillance. The feedback rules are immediate, private, and behavior-focused: the conversation happens close to the observation, out of earshot, and names the behavior and the safer alternative, not the person. The positive observations get the same treatment, because the reinforcement of safe behavior is half the mechanism. The supervisor who delivers ten feedback conversations a week builds the habit; the one who saves them for the monthly meeting builds a form. The BBS feedback section on this site covers the conversation structure.
The distinction that decides the fix
The observation cadence
The observation program runs on a cadence, and the cadence is the commitment. The supervisor schedules the observation slots, typically two or three per shift, at the high-exposure times, the shift change, the dock rush, and the aisle peaks. The slot is protected, the supervisor is not interrupted, and the observation is recorded in the log before the next slot. The cadence produces the data that the weekly review reads, and the KPI is the observation count per week, with the target set from the baseline. The supervisor who runs the cadence for a month has the pattern that the program needs; the one who observes when reminded has anecdotes. The cadence is the difference between the BBS program that works and the one that is claimed.
The observation data has to answer one more question before the fix: is this a behavior problem or is the environment producing it? The intersection where every operator rolls through without stopping is probably a sightline problem, not a training problem, and the fix is the mirror, camera, or detection system rather than more reminders. The supervisor who can make that distinction, using the observation data and the Risk Scorer on this site, is the one whose feedback changes the floor rather than just documenting it. The observation is the skill; the distinction is the wisdom, and the program needs both.
