Why most difficult conversations go badly
When people improvise difficult conversations, the same three failure modes repeat. They lead with character interpretations instead of observable behavior (“you’re being dismissive” instead of “you cut me off twice”). They bundle three concerns into one conversation, which makes all three feel less serious. And they end without a specific ask, so the other person leaves the room agreeing it was a good talk and changes nothing.
The SBI framework is structurally allergic to all three. By forcing the conversation onto observable specifics, by isolating one situation per conversation, and by ending with a concrete ask, it gives even an inexperienced communicator most of what a skilled one would do naturally.
Situation: anchor to one moment
The first move is to pick one specific situation — a meeting, a conversation, a project moment — and stay with it. Vague framings (“sometimes you…”, “you have a tendency to…”) put the other person on the defensive because they can’t actually verify or refute the claim. Specific moments are checkable. They invite the other person into the conversation as a participant, not a defendant.
Behavior: stay observable
Behavior is the discipline of the SBI framework. The test is simple: could a video camera have captured what you’re describing? “You were rude” fails the test — rudeness is an interpretation. “You interrupted me three times in five minutes and didn’t come back to my original point” passes. The discipline is hard, but the payoff is enormous: behavior described observably is almost impossible for the other person to dismiss.
When you find yourself reaching for a word like “rude”, “dismissive”, or “passive-aggressive”, ask yourself what specifically you saw or heard that made you reach for that word. Use the specific thing instead.
Impact: speak from your experience
Impact has two layers. The first is your felt experience — how the behavior landed for you (confused, embarrassed, unable to do your job, unsupported). The second is the downstream impact on the team, the project, or the relationship. Lead with the felt experience because it isn’t arguable; nobody can tell you you didn’t feel embarrassed. Add the downstream impact only if it’s true and useful.
A common trap: people skip their own felt impact and go straight to the team-level effect because it feels less vulnerable. This is a mistake. Personal impact is more persuasive, more humanising, and more likely to produce a real shift than abstracted “the team is affected” framing.
The ask: end with one specific change
Every difficult conversation should end with one specific, actionable, behavioural ask. Not three, not a rant about how they should be, just one thing the other person could plausibly do differently next week. The structure of a good ask is usually: “Going forward, I’d like to [the change]. Would that work for you?”
The closing question matters as much as the ask. It signals that you’re looking for an agreement, not delivering a verdict, and it gives the other person a place to land. They might agree, propose a modification, or push back; all three are productive next moves. What you don’t want is a one-way speech that they nod through and forget.
Tailoring SBI by audience
The structure is universal. The language varies. A piece of feedback to your manager about not following up on a commitment uses different word choices than the same structural feedback to a direct report; the conversation with a peer about workload imbalance is framed differently than the same conversation with a partner about household load.
The planner ships with curated example phrases for every combination of audience (manager, direct report, peer, partner, parent) and topic (performance, behavior, workload, relationship, decision). Each example is a starting point, not a finished line — the goal is to get you out of the blank-page paralysis and into the work of making the script your own.