Preparing for a difficult client call and practising it are not the same thing
Most experts prepare for difficult client conversations by thinking about them. They run through the likely objections, remind themselves of the key points, maybe write a few notes. That’s useful groundwork — but it stays in your head.
When a client is pushing back on your pricing, questioning your timeline, or challenging a commitment you made in a proposal, you don’t just need to know what to say. You need to be able to say it — fluently, calmly, under pressure. That’s a different skill, and most experts have never had a low-stakes way to practise it.
ChatGPT‘s voice mode changes that. You can now run a realistic version of the difficult conversation out loud, before the real one happens.
How the setup works
Before switching to voice mode, you upload your proposal or briefing document to ChatGPT and send a prompt that describes the client: their role, their communication style, and the concerns they’re likely to raise. You also give it one critical instruction — to stay in character and maintain pressure even when you make a good point.
Download our prompt template below.
Join the mailing list
While you're here, why not join the AI for Experts mailing list? Get new workflows, prompts, and practical breakdowns to embed AI into your expertise business. Unsubscribe anytime.
That instruction matters. ChatGPT’s default behaviour is to be accommodating. Without explicitly telling it to push back and hold its position, the role-play won’t feel like a difficult client. It’ll feel like a supportive one, which defeats the purpose.
Once the briefing is in place, switch to voice mode and start the conversation. The AI plays the client, asks the tough questions, and challenges the points where your proposal is weakest. You respond out loud — not by typing, but by speaking — which is exactly what you’ll be doing in the real call.
Why the document makes the difference
The document upload is what separates this from generic role-play. When ChatGPT has read your actual proposal — the pricing, the scope, the specific commitments — its pushback can be grounded in what the client would actually be looking at. The questions become specific to your situation, not a generic set of objections.
A vague briefing prompt produces a vague conversation. The more context you give about who the client is and what they’re likely to push on, the more useful the practice becomes. A prompt template you can use is available with this article.
Running it more than once is where the value is
Run the conversation at least twice. The first time, you’re finding your feet. The second time, you’ll notice where you fumble — where you hedge, trail off, or give a weaker answer than you intended. The third time, you’re genuinely preparing.
After the conversation, switch back to text and ask ChatGPT for feedback. It can tell you which responses seemed unconvincing, what concerns went unaddressed, and where you appeared least confident. This kind of external reflection is harder to get from internal review — you know what you meant to say, which makes it difficult to hear how it actually landed.
The goal is fluency, not a script. You’re training how you say things under pressure, not memorising lines. If you start trying to remember specific phrases, the real conversation will sound rehearsed.
The benefit that’s easy to overlook
Beyond the practical preparation, there’s a settling effect. Once you’ve run the difficult conversation two or three times and found your footing, the anxiety that typically builds before a difficult call is reduced. You’ve already been through it. You know where you stand, where you might get challenged, and that you can hold your position when pushed.
That calmness matters as much as the content of what you say. Clients read confidence. Walking in settled rather than quietly apprehensive changes how the conversation goes, even before you’ve said a word.
If you’ve got a difficult client conversation coming up, fifteen minutes with ChatGPT’s voice mode beforehand is worth trying.
