Ever wish you could pressure-test your sales pitch before you put it in front of a prospect?
I’m not talking about recording yourself in front of a mirror or dragging your spouse and/or dog through yet another rehearsal of your talk-track. (They love you. But remember: Love has limits.)
I mean running all your messaging—emails, cold calls, and everything that comes with it—through a simulation that’s good enough to call out your mistakes and point out where your pitch is likely to resonate or fall flat with the people you’re actually trying to sell to.
…Because guess what?
With AI, you can do that now.
I know, because I’m doing that now with a few of our companies: Asking ChatGPT to use mini-simulations to QA our sales playbook as we build it.
Let me show you how it works.
The Key Prompt: Kicking Off the Simulation
Here’s how I started to tune our sales playbook - without wasting my time, without hiring an expensive focus group, and without burning my credibility with the sales team.
Prompt: “I want you to simulate ten cold calls with the messaging I’m thinking about using in our prospecting talk-track (attached below). Act as if you are our key personas at our target companies, and simulate the back and forth between you and our salesperson. Look for dead spots in the introduction, language or jargon that turns you off, discovery questions that engage or land flat, transitions that seem too abrupt or clunky, and objections you’re likely to give (plus include some suggested responses that would work well to field those objections)."
First, if you’re wondering how I write my prospecting talk tracks, see this article.
And second, if you’re wondering, “Did the AI actually do that?” Yeah. As far as I know, it did simulate the calls. And as a result, it was able to pinpoint the exact areas where our messaging and talk-tracks needed some extra TLC. In less than a minute, it gave me a set of well-organized bullets summarizing what I should consider changing and why. I kept most of the changes, rejected a few, and asked ChatGPT to fold in the edits. It did a fine job, and saved me a bunch of retyping. All before we ever called a customer or let an email fly. Nice.
That’s when I decided to go a little deeper.
The Double-Click Prompts: What to Ask
After the initial simulation, I started drilling into both our pitch itself and the plan for deploying it. Below are the double-clicks I focused on, including the exact prompts I used.
1. Simulating Skepticism
Once we had the core messaging dialed in, the next step was to stress-test it against a “tough crowd.” Why? Because it’s easy for reps to get comfortable delivering a pitch when they know their audience is friendly or disengaged. But when the stakes are high—say, in front of a skeptical executive or a combative procurement team—the cracks in your messaging start to show.
I set up the AI to play the role of a particularly tough prospect:
Prompt: “Imagine you’re a 26-year-old salesperson delivering this pitch to a VP at one of our prospects who’s extremely skeptical about adopting new tools. As you read through the talk track, flag any part that feels weak or where a rep might get tripped up in a higher-pressure scenario.”
This prompt wasn’t about editing the content for clarity; it was about pressure-testing our pitch. The AI highlighted several areas that seemed fine on paper, but after closer inspection could easily derail a conversation in real life. For example, it noted that certain jargon-heavy phrases would likely lead to pushback. It also noted where my transitions felt too abrupt - potential sour notes that might annoy a more ornery potential buyer.
Once we had those weak spots mapped out, I added a follow-up prompt to help me generate some coaching points.
Follow-Up Prompt: “Now, based on those flagged areas, I want you to suggest coaching points that would help a rep navigate through those tough moments. Keep the focus on making the messaging more resilient and easier for them to deliver under pressure.”
This worked well. The AI added tweaks like transitional phrases to smooth out the flow and specific responses to use when our pitch hit resistance. With these adjustments, the playbook was no longer just effective—it was pre-emptively battle-tested for tougher audiences.
2. Mapping Insights to Discovery Questions
When you’ve got a treasure trove of unique data at your fingertips, you want to make sure you’re using it to drive real conversations. But here’s the thing: if you lead with a great insight and then follow up with a generic discovery question, you’ve just spoiled the magic trick. A strong data point needs an equally strong question behind it—one that keeps the momentum going and prompts the prospect to reveal more about what’s going on.
This company has access to some powerful customer feedback data. We wanted to use it to help our sales reps start conversations that would resonate, but we knew that simply showing up with a few “what’s happening” data points wasn’t enough. The key was pairing each insight with a discovery question that could unlock more context and lead the conversation naturally.
Here’s how we approached it:
Prompt: “List five to ten potential insights we might see in a prospect’s data. For each one, suggest a specific discovery question that could help us better understand what’s driving that trend.”
This wasn’t about running another simulation or mapping out a script. It was about building what I call a “pitch map”—an if-then decision tree that sales reps can follow in real time. If the insight looks like X, then ask about Y. It’s a way of turning raw data into a more fluid conversation that leads to a natural “I think we can help.”
The AI’s initial list was a solid starting point but needed refinement. Some of the suggested questions were too vague, others too aggressive. So, we added a follow-up prompt:
Follow-Up Prompt: “Refine these discovery questions to be more specific. The goal is to build on the insight and prompt the prospect to consider not just ‘what’s happening,’ but why it’s happening and what it means for their business.”
Here’s a semi-redacted example of what we ended up with:
This approach created more than just a list of questions. It gave our reps a kind of conversational roadmap, training them on not only what to say, but how to recognize the signals that make tailoring their pitch much more natural. Think of it as a flexible structure that reacts to how the prospect responds, letting you discover the story instead of deciding on it upfront.
The biggest pitfall we avoided? Checklist discovery. You know, the type of discovery call where a rep runs through a series of generic questions without considering the answers they’re getting and without getting curious with follow-ups, creating a disjointed and awkward interaction. Instead, our AI-polished pitch map encouraged reps to let the insights guide the flow and gave them the flexibility to go deeper based on what the prospect shared. Nice.
3. Coaching + Making It Stick:
Once we had the core messaging dialed in, the next step was to ensure that our sales team could actually execute it. That meant pinpointing where reps might get tripped up.
I started with a simple analysis:
Prompt: “What is a 26 year-old salesperson most likely to be confused about in this talk track? What will they struggle with when it comes time to use it in front of a prospect or customer?”
To be clear, I wasn’t asking the AI to run another round of simulations. This was my way of getting an outside perspective on the “last mile” of this sales enablement project. More specifically, I was asking the AI to guess where sales reps might struggle to put the talk-track to use - where they might feel stuck, uncertain, or less confident. The AI flagged a few spots where the language was too vague or abstract, suggesting that junior reps might have trouble clearly articulating the value or knowing when to pivot in the conversation.
Next, I needed to address these areas in a way that wouldn’t overwhelm the reps, but would give them some guardrails to follow:
Follow-Up Prompt: “Ok, so based on those areas where people are likely to struggle, I want you to insert some short coaching points in the right sections that correspond to the areas of potential confusion. Focus on making it easy for salespeople to (a) execute the talk-track and (b) make it their own.”
This prompted ChatGPT to insert concise coaching tips directly into the talk track—little reminders like:
“After this question, pause for a few beats. Let the prospect process and respond before jumping in with more details.”
“If the prospect hesitates here, try asking ‘Does that align with your priorities?’”
These small adjustments turned the script into a more usable “choose your own adventure” guide that reps could easily follow and personalize, making the delivery feel smoother, more confident, and more authentic.
The Results
After just a few rounds of this “simulate + tweak” approach, we had refined messaging that:
Ditched the jargon and honed in on pain points faster.
Tightened up our intro so we wouldn’t lose people in the first 20 seconds.
Built us a short + long version of our company intro, giving our sales team easy-to-reference options that help them “run the option” based on the appetite of the prospect
Built out a set of go-to responses for the most common objections.
But here’s the kicker. We did all of this in a matter of hours, not weeks.
No guessing. No wearing out the sales team. And no putting my credibility at risk. (No more than usual, anyway.)
Just rapid iteration, immediate feedback, and most importantly, a polished “what to say” talk track I now feel much better about sharing with our sales team.
Want to read about the other ways I’m using AI to help my portfolio companies? Check out my previous AI how-to guides: Boss Your Robot Around and How To Do Cool Stuff With AI.
Love this - crafting and chipping away at the pitch is hard for most. Nailing it and getting the story arc to land well is one of the biggest ROI skills 💲 Thanks for sharing ⭐
Would love to demo how https://www.syntheticusers.com/ can help with this use case. Let me know if that's something you'd be interested in.