I am going to admit something that should probably disqualify me from giving sales advice: for the first two years of my career, my 'discovery calls' were presentations with a polite Q&A bolted onto the end. I had a 20-slide deck. I walked through every slide. I answered feature questions. The prospect would say 'this is really helpful, let me discuss internally,' and then vanish into the witness protection program of unreturned emails. I thought I was being thorough. I was being a brochure with a webcam.
The wake-up call was literal. A sales ops friend pulled my call recordings and ran the talk-time analysis. I was talking 78 percent of the time on calls I had labeled 'discovery.' Discovery. I was discovering the sound of my own voice. The prospect, the person whose problem I was supposedly there to understand, got 22 percent of the airtime, most of it spent saying 'mm-hm' while I clicked to the next slide.
What Discovery Actually Is
A discovery call is not a presentation. It is a structured conversation engineered to understand the prospect's situation, their pain, and their readiness to act, in that order. If you are talking more than 40 percent of the time, you are pitching, not discovering, and the math does not care how good your slides are. The hard part is that pitching feels productive. Explaining your features lights up the same reward circuit as being useful. It is a trap, and I lived in it for two years.
Building the Guide Before Every Call
Now I run the Discovery Call Question Set That Keeps the Prospect Talking prompt on Gemini 3 Pro Preview before every new prospect. I feed it the prospect's industry, what I'm selling, the call length, and the three pains I most need to uncover. Gemini's strength with structured tabular output is the whole reason it's the right model here. It returns a guide where each theme is a clean table: primary question, follow-up probe, what a strong-fit answer sounds like, and what a disqualifying answer sounds like. I can scan it during the call without losing eye contact, because it's a table, not a wall of prose.
- My prospect talk-time went from roughly 22 percent to over 60 percent
- Qualification got faster, because the disqualifying-signal column told me by minute 20 when a deal wasn't a fit
- Average deal size went up, because I understood the real problem before I proposed anything
- The 'questions to never ask' section killed two habits I had that were quietly stalling my own momentum
The Disqualifying-Signal Column Is the Hidden Gem
Everyone focuses on the good questions, but the column I rely on most is the disqualifying signal. For each question, the guide tells me what answer means 'this is not a real opportunity, stop investing.' Before this, I would routinely get 40 minutes into a call before realizing the prospect had no budget, no authority, and no urgency, the unholy trinity of wasted afternoons. Now the structure catches it around minute 20, and I can graciously redirect to a smaller next step or politely close the loop. The time I save on bad-fit calls is time I spend on real ones.
The 'Questions to Never Ask' List Saved Me From Myself
Every guide the Discovery Call Question Set That Keeps the Prospect Talking prompt produces ends with a short list of questions that sound like discovery but actually kill it. I was personally guilty of two of them. The first was 'What's your budget for this?' asked too early, which makes a prospect clam up and anchor low before they've even felt the pain. The second was the leading question dressed as curiosity: 'Wouldn't it be helpful if you could automate that?' Of course they say yes, and you've learned nothing except that you can lead a witness. Seeing those two on the 'never ask' list, with softer replacements, was like watching a recording of my own bad habits with subtitles.
The softer versions matter. Instead of 'what's your budget,' the guide gives me 'when you've solved problems like this before, how did you typically size the investment?' That gets a real answer because it's a story, not an interrogation. The whole prompt is built around the idea that every question should open a door, never close one, and the never-ask list is the negative space that makes the rest of it work.
I pair this with my Follow-Up Email Built From Your Messy Call Notes prompt, because a good discovery call produces notes worth following up on, and a good follow-up email is just discovery, written down and sent back. Grab the Discovery Call Question Set That Keeps the Prospect Talking prompt on Prompt Dock, build a guide for your next call, and then go measure your talk time afterward. Brace yourself. Mine was 78 percent and I have never recovered from learning that.