The Skills Gap Field Guide
Behavioural skills show up in the moments people don't plan for. How someone handles a confidently wrong AI answer. Whether they share a rough prototype before it's polished. What they do when an experiment delivers something they didn't expect.
Reading those moments is a skill you can build as a manager. This guide walks through five of them, with one cue for each that tells you whether the underlying habit is in the room.
Part one: The five moments
Each one is a recurring situation at work. For each, the strong version of the underlying habit looks different from the weak version. The cue is what to watch or listen for.
Someone drops an AI-generated summary into the team chat. It reads cleanly and gets used.
Notices something is off, isolates one specific claim, and tries to disprove it before relying on it.
Forwards it on. It reads like it's right, and checking would slow things down.
When AI output lands in chat, does anyone push back on a specific line before it gets used?
Someone is building something new. There's a working version of it, and it isn't pretty.
Shows the ugly version, asks specific questions, opens a real conversation about direction.
Waits for it to be polished. By the time the stakeholder sees it, the direction is locked in.
How often does work get shown before it's "ready"? If the answer is "almost never," that's the gap.
The team is divided on whether to expand something that worked in one place. The conversation has gone in circles.
Proposes a bounded test with a clear question they're trying to answer.
Argues from conviction on either side, or kicks the decision down the road.
When the team disagrees, does anyone propose an experiment, or does it stay a debate?
A project is underdelivering. Everyone knows. Nobody has said it out loud yet.
Holds an honest retro, names what they got wrong, changes direction.
Defends, points to external factors, lets it quietly wind down without naming the lesson.
When something underdelivers, does the team talk about it openly, or does it just disappear?
A number comes in that nobody predicted, or a customer says something nobody expected.
Investigates the surprise, asks what it might mean, treats it as data.
Dismisses it as noise or an outlier and moves on.
When something surprising happens, what's the first instinct on your team: explain it away, or look closer?
Part two: A worked example
A team of seven analysts at a mid-sized consultancy rolled out an AI research assistant three months ago. It pulls together background reads on prospective clients. The lead is trying to figure out whether it's actually helping.
Three things have happened in the last fortnight. None of them is a crisis on its own.
An associate sent a briefing that mentioned the client's recent acquisition of a competitor. The acquisition hadn't happened. The AI had inferred it from a partnership press release and stated it as fact. The client noticed and emailed back politely. The internal conversation was about whether to add a disclaimer to AI-assisted briefings.
One of the juniors built a small dashboard that surfaces the AI's confidence scores so the team can spot the briefings that need closer reading. She's been running it on her laptop for two weeks. The lead only heard about it when someone mentioned it in passing. Asked why she hadn't shared it, the junior said it wasn't ready: bugs, rough styling.
The team has been arguing for a fortnight about whether to use the same tool for post-engagement reports. Two people are strongly for, two strongly against, the rest stay quiet in meetings.
Read together, these three stories say something about how the team handles three particular moments. The briefing mistake is what happens when AI output gets treated as a deliverable rather than a draft. The dashboard on a laptop is what happens when the team's threshold for "shareable" has crept too high. The deadlock on post-engagement is what happens when the team has no shared way to settle disagreement except by arguing it out.
What the lead does next. She asks the junior to demo the dashboard at the next team meeting, bugs and all, and thanks her for not waiting. She proposes settling the post-engagement question with a four-week pilot on three engagements, with one defined question: does time saved outweigh editing burden. And every time an AI summary lands in the team chat now, she asks the same thing: "What's the one claim in here you'd most want to check?"
She works at the level of the moments themselves.
Part three: Your turn
The method only works on a moment you actually saw. Pick one from the last two weeks.
This text stays in your browser. Nothing is sent or stored on our side. Refreshing the page will keep what you wrote on this device; clearing your browser data removes it.
Part four: What to do with what you saw
When you spot the weak version of a behaviour once, that's one data point. Watch the same kind of moment three or four times before you draw a conclusion. People have bad days, teams have bad weeks. A pattern is what you see when the same signal shows up in different conditions and keeps showing up.
Behaviours change through practice in real situations. The most reliable interventions are small and repeated: coaching that meets people in the moment, structured retros that name what happened honestly, bounded experiments that let the team practise the behaviour with low stakes. One-off training tends to land well in the room and fade within a fortnight.
If you see the same weak version of a behaviour across several of these moments, you're looking at an underlying habit your team hasn't built yet. That's worth taking seriously, because it shapes how every new tool or strategy lands on the team.
This guide teaches you to spot the gaps yourself. If you want to measure them rigorously across a whole team and give each person a personalised development path, that's what the i2 Skills Assessment is built for.
Visit i2skills.com