A discovery call framework you can read off a second screen

I opened Gong’s discovery research and Craqly’s own pricing page on 20 Aug 2026 to re-check the numbers behind this discovery call script, and two of them had drifted from the versions that keep getting quoted around sales blogs. One drifted far enough that the standard advice may now be backwards. What follows is the part you can read off a second screen, with a note under each block saying what that block is doing, so when a call goes sideways and you skip something, you at least know what you gave up.

Read it out loud once before you use it. Scripts that look fine on the page fall apart in the mouth.

What each block of a discovery call script is for

Discovery has four jobs and only one of them is asking questions. You earn the next 25 minutes during the first two. You find a problem the buyer already agrees they have, rather than the one you’d prefer they had. You get a number attached to that problem, phrased their way, so the person who wasn’t on the call can repeat it in a budget meeting without you standing there. Then you work out who signs and by when. A deal with no date attached is a hobby.

The spine here is SPICED, the Winning by Design framework: Situation, Pain, Impact, Critical Event, Decision. It’s ordered roughly the way a real conversation runs, which is why I’d rather script against it than against a qualification checklist. Winning by Design’s own comparison describes MEDDIC as “direction driven; tells you what needs to be done next” and SPICED as “action driven”, and there’s a longer SPICED blueprint PDF if you want the full version. If you sell into enterprise you’ll likely run both, SPICED on the first call and MEDDIC or MEDDPICC later when someone inspects the deal.

The discovery call script, block by block

Block 1. The frame, about 60 seconds

Say: “Thanks for making time. Here’s what I had planned. I’ll ask how [process] runs for you today for about twenty minutes, then if it looks like we’re useful I’ll show you the two parts that would matter, and we’ll sort out next steps. If it doesn’t look like a fit I’ll tell you. Does that work, or is there something you want to make sure we get to?”

The block buys you permission to ask a lot of questions, which is the thing that otherwise makes buyers defensive around minute nine. The second half is doing quieter work: it drags out the agenda they came in with. Sometimes that answer alone reroutes the call.

Block 2. Situation

Ask: “Walk me through how [process] works today, start to finish.” Then: “Who else touches it?” Then: “How long has it run that way?”

You’re building a baseline you can point back at later. Cap it at three questions. Situation questions feel safe for both sides, which is exactly why nervous reps hide in them and burn eleven minutes learning things a public help page would have told them.

Block 3. Pain

Ask: “Where does that break?” Then: “When did it last break?” Then: “What did you do about it that time?”

The second question is a recency test. Pain that last happened fourteen months ago isn’t pain, it’s a memory, and it won’t survive a procurement review. The third question tells you what they’ve already tried, which is also the list of things your pitch shouldn’t sound like.

Block 4. Impact

Ask: “Roughly what does that cost you in a quarter, in hours or in dollars?” Then: “Who feels it?” Then: “What happens if nothing changes before [their fiscal date]?”

This is the block reps skip, and skipping it is why deals stall at the business-case stage with nothing to put in the business case. If they can’t produce a number, build one out loud with them, slowly, then repeat it back and ask whether you got it wrong. The number doesn’t have to be right. It has to be theirs. Write it down verbatim. A buyer who says “call it six hours a week across four people, so maybe thirty thousand a year once you count the contractor we hired to paper over it” has just written the opening line of your own business case, and no estimate you produced for them would carry the same weight in a room you aren’t standing in.

Block 5. Critical event

Ask: “Is anything forcing a decision by a particular date? Renewal, audit, launch, a headcount plan?”

Urgency that belongs to them survives the quarter. Urgency you manufactured with a discount deadline does not. If there’s no critical event, that’s fine, but write it down, because it’s the single best predictor of the deal quietly aging.

Block 6. Decision

Ask: “If you decided this was worth doing, what actually happens next inside [company]?” Then: “Who else needs to see it?” Then: “Have you bought something like this before? How did that go?”

The last one earns its place. People describe their procurement process accurately when they’re describing the past, and vaguely when they’re predicting the future. You’ll hear about the security review nobody mentioned.

Block 7. Closing the call

Say: “Let me play back what I heard. [Pain], costing you roughly [impact], with [critical event] landing in [month], and [name] would need to sign off. What did I get wrong?”

Then book the next thing while you’re both still on the call. Not “I’ll send some times.” Calendars are where discovery calls go to die.

How many questions should a discovery call script have?

Gong’s analysis of over 519,000 B2B sales call recordings found that “you have the greatest shot at nailing your discovery call when you ask between 11-14 (targeted) questions”, with success falling back toward average once you push past that range. The script above holds 17 prompts, 13 of which are questions, so it sits inside that band with two spare. That’s deliberate. You’ll want the room.

The count matters less than the spacing. The same Gong write-up notes that average salespeople “front-load” their questions “as if they are working their way through a checklist”, while stronger reps spread them across the call. So don’t run blocks 2 through 6 as a solid wall of interrogation. Answer things. Tell a short story between block 3 and block 4. Let them ask you something and actually stop talking.

The talk-to-listen number is shakier than people think

Everyone has heard the golden ratio: talk 43% of the time, listen 57%. On 20 Aug 2026 I read Gong’s talk-to-listen page, drawn from 326,000 calls of at least ten minutes, and the figure it now shows for closed-won deals is 57% talk time, which is the old ratio flipped. I don’t know whether that’s a new dataset, a different way of segmenting reps, or something about how those calls were sampled, and one page read on one afternoon isn’t enough to settle it.

My read, and you’re welcome to disagree: the ratio was never the mechanism. Speaker switches per minute is closer to the thing that matters, because two people trading turns is a conversation and one person holding the floor isn’t, whichever of them is holding it. We wrote up the ratio question separately in talk-to-listen ratio on sales calls. Use it as a smell test after the call. Don’t sit there watching a meter mid-sentence.

Where the thirty minutes go

Reading a script live without sounding like you’re reading

The reason reps abandon scripts isn’t that scripts don’t work. It’s that reading a doc, taking notes and listening are three things competing for one attention budget, and listening loses. That’s the gap Craqly sits in: the script and the next question stay on your screen while the call transcribes itself, so you’re not typing during block 4 when the buyer finally says a number out loud. Craqly’s pricing page, checked 20 Aug 2026, lists a free Starter plan at 20 credits a month, with “1 credit = 1 minute of live session · credits reset every month”, and Pro at $38 a month or $19 a month on the yearly plan. Stealth is on the free plan too, not held back for paid. Twenty minutes of credits is enough to run this script once and see whether reading from a screen changes how the call feels.

If you’d rather see the wider case for assisted discovery first, we covered it in AI for discovery calls.

What to write down afterwards

Within ten minutes, while it’s still in your head: the pain in their words, the impact number, the critical event date, the names, and the objection you fumbled. That last one is the one people leave out, and it’s the one that repeats. Our meeting summary templates cover the format, and the objection you fumbled belongs in a running objection list you review on Fridays.

One warning about the script. If you run it word for word twice and it works both times, you’re probably talking to two buyers with the same problem, which means the script isn’t the interesting part of what happened. Change one block next call and see what breaks.