How to Prepare for a Difficult Conversation: Decide What They Should Believe

To prepare for a difficult conversation, decide what the other person should believe when it ends, write that belief down in one sentence, and plan the route to it before you plan a single talking point. Open on ground you already share, and let the true facts that support the belief come up as context. If the conversation is bad news you owe someone, say it in the first sentence.

Saying and Believing Are Different Jobs

If you prepare for a high-stakes conversation by rehearsing three key points, anticipating the objections and scripting rebuttals for the rebuttals, you’re preparing for a contest won on points. That preparation assumes that if you say the right things in the right order, belief will follow. Belief forms in the other person, from what they see and the order they see it in, so the preparation has to start with where you want their conclusion to land and what they’ll see on the way there.

Order matters, and there’s research behind that. In 1946 the psychologist Solomon Asch gave people the same list of personality traits in two different orders and got noticeably different impressions of the person described. He described a “factor of primacy”: the first traits set a direction that later ones were fitted to. A hard conversation likely works the same way, with the first few minutes setting the frame and later facts read through it. Facts without a frame get interpreted inside whatever frame the other person already holds, and that’s rarely the one you want.

Two Hard Conversations Landed in the Same Week

One of the CEOs in ProvenChaos coaching had two hard conversations land on him in the same week. Call him Marcus; it isn’t his name.

A referral partner pushed back hard on a markup, the kind of pushback where you can feel the relationship starting to crack under the conversation. In the same week, his co-founder had come to believe the business wasn’t profitable, when in fact it was.

The obvious preparation was the spreadsheet: pull the numbers, schedule the confrontation, build the argument. Marcus asked a different question first: what does this person need to walk away believing?

For the referral partner, the answer was that the markup paid for real value to the partner’s clients. For the co-founder, it was that the business was doing well, reached through the co-founder’s own reading of the evidence.

With the partner, Marcus never opened on the markup. He spent most of the conversation on why the overall approach saves the partner’s clients money and how the coordination cuts costly rework. The markup came up late and briefly, inside a context that already explained it. The partner concluded it was fair. Marcus never said it was.

With the co-founder, he didn’t call a meeting to announce that the business was profitable, because that would have opened on the contested ground. He shared the real numbers openly as they came up in ordinary work: a good month, a client win, a cost improvement. Eventually the co-founder was the one crediting the results.

Both conversations landed, and neither felt like a negotiation to the person on the other side. His own team saw both results and put them down to luck.

Pre-Frame Architecture: Three Questions, in Order

Pre-Frame Architecture is the name for what Marcus did: you decide the destination before the conversation departs. It has three components, used in order. The sample lines below are illustrations built on the referral-partner case, not transcripts.

Component	The question	What it sounds like
Target Belief	What should they conclude, in their own words, when this ends?	“The markup pays for coordination that saves my clients money.”
Entry Point	Where do we already agree?	“Let’s start with what your clients need from a project like this.”
Oblique Path	Which true facts can come up as context?	“That last project came in with almost no rework, which is what the coordination is for.”

The Target Belief is what they should conclude, phrased as their own thought. Write it the way they’d say it, for example “The markup pays for something my clients want.”

The Entry Point is somewhere the other person already agrees with you, well away from the disputed number or the profitability question, so the frame starts building before the disagreement is on the table.

The Oblique Path is how the evidence arrives: as context, an observation or a detail in a larger story, which gives the other person room to weigh it before deciding whether to push back.

Two working rules sit inside the method. The first is the 70/30 Split: spend roughly 70 percent of the conversation on the frame and 30 percent on the contested fact. It’s a working rule from ProvenChaos coaching and the percentages are approximate. Putting the frame first and the contested fact late reverses the habit of making the case for most of the meeting and dropping the ask at the end. The second is the corollary: never answer the question that wasn’t asked. Marcus didn’t justify the markup before anyone had asked him to, and by the time the number came up, it needed no defense.

For the shorter story of how this played out, read Pre-Frame Architecture: Why the Invisible Move Wins the Room.

What Should I Do Before a Hard Conversation With My Business Partner?

Write one sentence: what your partner should believe when the conversation ends, phrased the way they would say it. Then pick an opening you both already agree on, and list the true facts that support the belief. Plan to spend most of the time on shared ground and bring the contested point in late, inside that context.

How Do I Stop a Difficult Conversation From Turning Into an Argument?

Don’t open on the point you disagree about. Arguments start when the first thing said is the contested claim, because the other person’s job then becomes defending against it. Start with something they already agree with, let the supporting facts come up as context, and don’t answer objections nobody has raised yet.

How Do You Persuade Someone in a Difficult Conversation Without Manipulating Them?

Use only true evidence, and aim for a conclusion the other person could defend to anyone with all the facts on the table. The test for the Oblique Path: every fact on it is true, and you’d be comfortable if the other person read your written plan. If you wouldn’t be, rewrite the Target Belief before you plan anything else.

The other person also stays free to land somewhere else. If they hear the evidence and still disagree, ask what they’re seeing; it may be something you missed. In Marcus’s case the fix was true numbers, shared openly. Had the numbers been bad, the honest move would have been a direct conversation about them.

Pre-Frame Architecture is for disagreements where the true facts are on your side and the other person hasn’t seen them yet. Owed bad news works differently.

How Do I Prepare to Deliver Bad News to Someone on My Team?

Put the news in your first sentence. Before you walk in, write what the person should believe afterward: that they were told straight and treated fairly. Then plan the reasons you’ll give, in one or two sentences, and the next step you’ll agree before they leave. Don’t build a frame first; with owed news, the delay is the harm.

The plan for this case uses the first three parts of the four-line card below, rearranged:

1. Target Belief: “When this ends, [their name] will believe they were told straight and treated fairly.”
2. Entry Point: the news itself, in the first sentence. “We’re ______, effective ______.”
3. Path: “The reasons, in one or two sentences: ______. The next step we’ll agree on before we leave: ______.”

A termination, a missed commitment, a decision that’s already been made: each of these goes first. Burying owed news under twenty minutes of context is the con version of this method.

If you can’t write the Target Belief because you haven’t decided what you think yet, the conversation is premature. Make the call first, with A Decision-Making Framework for Leaders. If the hard conversation is the one your leadership team keeps having in every meeting without resolving, that’s a decision-rights problem, covered in We Keep Having the Same Conversation and Nothing Gets Decided.

Try This: Four Lines Before Your Next Three Conversations

Before each of your next three high-stakes conversations (a partner, a co-founder, a client, a board member), fill in these four lines on paper before you walk in.

1. Target Belief: “When this conversation ends, [their name] will believe ______.” One sentence, in words they’d use. If you can’t fill it in, hold the conversation until you can.
2. Entry Point: “I’ll open on ______, which they already agree with.”
3. Oblique Path: “The true facts I’ll let come up as context are ______, ______ and ______.”
4. Next step: “The next step I’ll propose before we leave: ______.”

Then one check before you go in: “Would I be comfortable if they read this card?” If the answer is no, fix the card. And if you’re still angry when you write line one, wait a day; a Target Belief written in anger usually turns into a verdict on the other person, and they’ll hear it in your first sentence.

After the conversation, score it on one question: did they say some version of the Target Belief back to you, in their own words? If they didn’t, check the Entry Point first, because that’s usually where the conversation reached contested ground too early.

The Skill Can’t Stay in Your Head

Marcus’s team saw two results and called them luck. When the skill lives only in the CEO, every hard conversation with a partner, a client or an unhappy co-founder gets routed upward because nobody else knows how it’s done, and each time the CEO handles one well, the team learns again that this kind of conversation needs the boss.

That’s one form of The Genius Bottleneck: judgment that lives in one head and can’t be delegated until it’s written down. Writing the card down is what makes the skill teachable, which is why the worksheet above and the team exercise below both put it on paper.

Use This With Your Team: A 20-Minute Pre-Frame

Run this in your next leadership meeting on a real conversation someone has coming up in the next two weeks: a client renewal, a vendor price increase, a partner dispute, a hard message to staff. The person who owns the conversation runs the exercise, and the CEO takes part like anyone else.

1. Minutes 0 to 3: the owner describes the conversation and the person on the other side, sticking to facts.
2. Minutes 3 to 8: everyone writes the Target Belief silently, one sentence each, then reads it aloud. If the sentences differ, that’s the first finding: the team doesn’t agree on what success looks like.
3. Minutes 8 to 12: agree the Entry Point. What does the other person already agree with?
4. Minutes 12 to 17: list the true evidence that supports the belief. Mark which pieces can come up as context. Strike anything you wouldn’t be willing to say directly.
5. Minutes 17 to 20: the owner writes the four lines on one card, including the next step to propose. At the next meeting, they report one thing: did the other person say the Target Belief back in their own words?

Rotate the owner each time, so that over a quarter several people on the team have prepared a difficult conversation this way and the CEO is no longer the only one who can.

If hard conversations with partners, clients or your own co-founder keep landing on your desk because nobody else on the team can hold them, that’s the work of Executive Coaching: weekly one-hour sessions, $3,000 a month, month to month, spent on the real conversations and decisions in front of you and on getting the skill out of your head and into your leadership team. It isn’t for a CEO who wants a script for one conversation tomorrow; the worksheet above does that for free. The first step is a 30-minute conversation.