There's a specific kind of dread that settles in the night before a difficult phone call — the one where you have to push back on a client, negotiate a higher rate, or address something that's been quietly building tension for weeks. I used to handle that dread by rehearsing the conversation in my head at 1 AM, which helped nothing and ruined my sleep. Now I spend 15 minutes with ChatGPT before any high-stakes call, and I show up to those conversations calmer, sharper, and genuinely prepared for whatever comes back at me. This post is exactly how that system works.
TL;DR — Key Takeaways
- Mentally rehearsing difficult conversations in your head is the least effective preparation — it's unstructured, biased, and always ends the way you want it to
- ChatGPT can simulate the other person's likely responses, pressure test your talking points, and help you prepare for objections you haven't thought of
- Free tools: ChatGPT (free tier), Google Gemini (free), Otter.ai (free tier for post-call review)
- Paid tools: ChatGPT Plus (~$20/month), Otter.ai Pro (~$16.99/month), Fireflies.ai (~$10/month)
- This system works for any high-stakes call: rate negotiations, client conflict resolution, difficult feedback conversations, partnership discussions, or late payment follow-ups
- The role-play prompt is the core of the system — everything else is built around it
The Call You Keep Putting Off
You know the one. It's been sitting on your to-do list for three, maybe four days. You think about it, you draft the opening line in your head, and then you find something more urgent to do instead. Not because you don't know what needs to be said — but because you don't know how the other person will respond, and you're not sure you'll handle it well when they push back.
For solopreneurs and freelancers, this kind of avoidance is especially costly. You're not just one person on a team — you are the team. Every difficult conversation you delay is a delayed decision, a delayed resolution, or a delayed rate increase that quietly costs you money every single month you don't have it.
And here's what makes it worse:
The longer you wait, the higher the stakes feel, the more you overthink it, and the more likely you are to either over-prepare with a rigid script or under-prepare by winging it and losing the thread the moment they say something unexpected.
What Actually Happens When You Wing It
I've had three genuinely bad client calls in my solopreneur career. All three had the same pattern: I knew what I wanted to say, I had a rough idea of how I'd open, and I told myself I'd "read the room" for the rest. And in all three cases, the moment the other person said something I hadn't anticipated, I went off track — conceded something I didn't need to concede, or doubled down on a point that wasn't my strongest, or simply got flustered and wrapped up the call without getting what I called for.
The problem with winging a difficult conversation isn't that you don't know enough. It's that your brain, under social pressure, defaults to the path of least friction — which is almost always the path of least outcome for you.
Preparation isn't about memorizing a script:
It's about having already thought through the hard moments before they happen in real time, so your nervous system isn't making decisions for you.
Why Mental Rehearsal Alone Doesn't Work
Most people prepare for difficult conversations the way I used to: by running it through in their head. The problem with this is that your mental simulation is hopelessly optimistic. You imagine yourself saying the right thing, the other person responding reasonably, and the call ending cleanly.
That's not a rehearsal — it's a fantasy.
Real preparation requires the other person to push back, say something unexpected, or respond badly. And the only way to experience that before the actual call — without a human sparring partner available on demand — is to simulate it with AI.
Cognitive science research on mental simulation shows that when people rehearse scenarios where they control both sides of the dialogue, they significantly underestimate resistance and overestimate their own persuasive clarity. ChatGPT fixes this by being a genuinely unpredictable interlocutor — especially when you instruct it to be difficult.
The Role-Play System I Use Before Every High-Stakes Call
This isn't complicated. It's a three-prompt sequence that takes 15–20 minutes and completely changes how I show up to hard conversations.
Prompt 1: Set the Scene
Before any role-play, I give ChatGPT a complete picture of the situation. The more specific I am here, the more realistic the simulation.
Here's the prompt structure:
"I have a difficult phone call coming up and I want to role-play it with you to prepare. Here's the situation: [describe the relationship — client, partner, contractor, etc.], [what the call is about — rate increase, conflict resolution, overdue payment, scope creep, etc.], [what outcome I need from the call], and [what I know about this person's likely reaction — are they defensive, friendly, confrontational, conflict-avoidant?]. You'll play the other person. Stay in character the entire time — don't break to give me advice mid-conversation. Be realistic, not cooperative. If this person would typically push back, push back. If they'd deflect, deflect. I'll start the call. Ready?"
Then I actually start the call — typed out, as if I'm speaking.
What comes back is often uncomfortable, which is exactly the point. ChatGPT plays the client who says "I'm not sure we have the budget for that" or the partner who pivots immediately to "well, I think the real issue is..." — the kinds of responses that throw you off in real conversations.
Prompt 2: The Debrief Prompt
After the role-play ends — either when I've reached a resolution or when I want to stop and analyze — I run this:
"Step out of character. Analyze our role-play conversation and tell me: (1) Where did I lose momentum or credibility? (2) Were there moments where I conceded too quickly or said something that weakened my position? (3) What was the strongest thing I said and why did it land? (4) What's the one thing I should say differently in the real call? Be direct."
This debrief is where the real learning happens.
In one recent rate negotiation prep, ChatGPT identified that I'd apologized twice before stating my rate — a pattern I hadn't noticed but that signals uncertainty to the other person before you've even made your ask. That single note changed how I opened the actual call.
Prompt 3: The Objection Prep Prompt
Once I've done the role-play and debrief, I run one final preparation prompt:
"Based on the scenario I described and our role-play, list the five most likely objections or difficult responses I'll face on this call. For each one, give me: (1) The objection exactly as they might phrase it, (2) The underlying concern driving it (what they're really worried about), and (3) My strongest 1–2 sentence response. Don't make my responses sound scripted — they should sound natural and confident."
Here's what that output looks like for a rate increase call:
| Their Objection | What They're Really Saying | My Prepared Response |
|---|---|---|
| "This feels sudden — we've been working together for a year." | I'm surprised and a little blindsided | "I should've raised this sooner — that's on me. But the rate I quoted you last year no longer reflects the scope we're actually working at." |
| "We don't really have budget flexibility right now." | We might, but I need a reason to justify it internally | "I get that — when would be a realistic time to revisit it? I'd rather plan ahead than have this catch either of us off guard." |
| "Can we do a smaller increase for now?" | I'm open to negotiating | "I'm open to that. What number are you thinking, and let's see if we can find something that works for both of us." |
Walking into the call with this table means nothing they say is truly unexpected.
The Optional Round 2: Make It Harder
If I know the person I'm calling is particularly difficult, I add this instruction before running the role-play a second time:
"This time, play the most challenging version of this person — be more resistant, more defensive, and more likely to try to redirect the conversation. I want to experience the worst-case version of this call before I have the real one."
This is deliberate stress inoculation — a technique used in performance coaching where you practice under conditions harder than the real event, so the real event feels manageable by comparison.
The Tools — Free and Paid
Free options:
- ChatGPT (free tier) — All three prompts work on the free version; the role-play quality is solid for most scenarios
- Google Gemini (free) — Good alternative for a second role-play simulation with a different "personality" interpretation
- Otter.ai (free tier) — Record and auto-transcribe your actual call afterward (300 minutes/month free); review the transcript to compare your real performance to your prepared talking points
Paid options:
- ChatGPT Plus (~$20/month) — GPT-4o is noticeably more nuanced in playing difficult characters; better at maintaining realistic resistance without breaking character
- Otter.ai Pro (~$16.99/month) — Unlimited transcription plus AI-generated call summaries; useful for identifying patterns across multiple difficult calls over time
- Fireflies.ai (~$10/month) — Records, transcribes, and provides AI analysis of your calls including talk-time ratios and topic tracking
Total monthly cost:
$0 on free tools. $26.99–$46.99/month with paid stack.
Before vs. After: The Shift That Actually Matters
Before This System
- Spent the night before hard calls running imaginary best-case scenarios in my head
- Went into calls with an opening line and no plan for what came after it
- Lost negotiation ground in three separate client calls by conceding points unnecessarily under pressure
- Avoided one difficult client conversation for 11 days — which let a fixable problem become a real one
After 4 Months Using This System
- Every high-stakes call gets a 15-minute role-play session the morning before
- Show up knowing the five most likely objections and a prepared response for each
- Rate negotiation success rate: 4 out of 5 calls resulted in an accepted increase or an agreed review timeline
- The 11-day avoidance habit is gone — when I know I've practiced, the call doesn't feel like a threat
The change that surprised me most wasn't my performance on the calls:
It was how quickly the dread disappeared. When you've already had the hard conversation — even with an AI — the real one stops feeling like stepping into the unknown. You've been there. You know how it goes. And that feeling of prepared calm is worth more than any specific script.
The difficult call you're avoiding right now is probably not as hard as you think. But you won't know that until you've practiced it somewhere safe first.
Your Turn
If you try this system before your next difficult call, I want to hear what happened — especially if the role-play surfaced something you hadn't anticipated. Drop a comment below: What kind of call are you preparing for? What did the debrief prompt reveal about your communication patterns? And if a prompt didn't produce a realistic simulation for your situation, tell me exactly what you typed and I'll help you rework it.
Comments
Post a Comment