This is such a useful way to look at psychological safety. The real test isn’t whether people feel comfortable in general, but how quickly they bring you bad news when something starts going wrong. That gap tells you a lot about the environment you’ve created.
I sell agent infrastructure at Dial, so take the conflict as declared: measurement is what we charge for. Which is why the exercise at the end is the part I keep arguing with.
The two dates only exist for problems that reached you. Every gap you can compute has a second date because somebody eventually spoke. The cases you most want are the ones with a first date and no second date, and those never enter the sample at all.
The second study you cite is interviews with people about what they chose not to raise. Researchers had to go and ask them, because nothing inside the organisation held a record of it. The evidence for the hole came from outside the instrument.
Which makes me think the sign runs the wrong way. If your first response genuinely improves, people start bringing you things they have been sitting on for months, and those arrive with long gaps. Get better and your gaps get longer.
Meanwhile the manager nobody trusts sees short gaps, because the only things that reach them are the safe ones that were always going to be reported quickly.
Against myself: an unmeasurable population is not a reason to measure nothing, and your final instruction escapes all of this. Asking the person who closed the shortest gap what made them tell you when they did goes at the decision rather than the record, and that is the one part nobody could have produced after the fact.
So on that one: when you have asked it, has anyone ever answered with something that had nothing to do with you? A deadline arriving, or finding out somebody else already knew.
Great stuff, Ian! Simple and actionable steps to explore and open things open before jumping in and shutting things down.
This is such a useful way to look at psychological safety. The real test isn’t whether people feel comfortable in general, but how quickly they bring you bad news when something starts going wrong. That gap tells you a lot about the environment you’ve created.
I sell agent infrastructure at Dial, so take the conflict as declared: measurement is what we charge for. Which is why the exercise at the end is the part I keep arguing with.
The two dates only exist for problems that reached you. Every gap you can compute has a second date because somebody eventually spoke. The cases you most want are the ones with a first date and no second date, and those never enter the sample at all.
The second study you cite is interviews with people about what they chose not to raise. Researchers had to go and ask them, because nothing inside the organisation held a record of it. The evidence for the hole came from outside the instrument.
Which makes me think the sign runs the wrong way. If your first response genuinely improves, people start bringing you things they have been sitting on for months, and those arrive with long gaps. Get better and your gaps get longer.
Meanwhile the manager nobody trusts sees short gaps, because the only things that reach them are the safe ones that were always going to be reported quickly.
Against myself: an unmeasurable population is not a reason to measure nothing, and your final instruction escapes all of this. Asking the person who closed the shortest gap what made them tell you when they did goes at the decision rather than the record, and that is the one part nobody could have produced after the fact.
So on that one: when you have asked it, has anyone ever answered with something that had nothing to do with you? A deadline arriving, or finding out somebody else already knew.