What 519,000 sales calls teach us about discovery questions

For six years, Gong Labs has collected and analyzed B2B sales calls from its customers. The 2025 analysis covered more than 519,000 calls from US and European SaaS teams. The conclusions no longer surprise anyone who follows data-driven sales research — but they still surprise most reps. Here are three patterns worth memorizing.

Pattern 1: There is an optimal question volume, and it's narrow

Top-performing reps asked between 11 and 14 questions on a discovery call. Fewer than 11 led to the rep collecting only surface-level information. More than 14 made the buyer feel interrogated. The difference between a rep asking nine and thirteen questions was meaningful in conversion; the difference between thirteen and seventeen was meaningful too, but negatively. It is an extremely narrow window.

That means more questions is not always better. It is a truth up to a point. Beyond that point the call becomes an interrogation and the buyer pulls away. Knowing where the line sits is one of the few concrete benchmarks sales actually has from large-sample data.

Pattern 2: Top performers spread questions evenly

The second observation was even more interesting. Good reps didn't bunch their questions at the beginning of the call — they spread them out. Average reps asked six questions in the first ten minutes and two questions across the last thirty. Top performers asked three questions in the first ten minutes and nine across the rest. That means a good discovery call doesn't have a discovery phase followed by a pitch phase. It is a continuous dialogue.

Pattern 3: The best questions follow an answer

Gong also found that the most successful reps often used follow-up questions — that is, a question that directly built on what the buyer just said. You just said the team is struggling with onboarding. What's the specific part that takes the most time? That signals listening and deepens understanding. Average reps instead jumped back to the next question on their checklist.

What Neil Rackham already knew in 1988

The SPIN model, presented by Neil Rackham in his book SPIN Selling after analyzing 35,000 sales calls, identified the same thing in smaller data: implication and need-payoff questions convert better than situation questions. Gong's analysis 30 years later confirms this in a much larger sample. The difference is that Gong's data could also measure where in the call the questions were asked and how they built on each other.

What it means in practice

Three concrete things to do on your next discovery call. One: prepare 14 questions but plan to ask 11. Two: spread them evenly — if it's a 30-minute call, you shouldn't have asked four more of them after the first ten minutes. Three: after every answer, ask a follow-up question that begins with you just said or you just mentioned. It is the single most underused sentence starter in sales.

The exception: C-suite buyers

Gong also noted that the 11–14 question rule doesn't hold against C-suite buyers. Senior executives have less patience for longer question sequences. With them you need to get past the three standard situation questions within the first two minutes and into implication questions about strategy, risk and time frames. For them, it isn't how many questions but how deep.

How to train this

This isn't a skill you read about. It's a skill you train. Salesprep, which has analyzed more than ten thousand AI roleplays, sees that reps who train one discovery call per week move their average question volume from seven to twelve within eight weeks. The easiest path is to roleplay against an AI prospect, get automatic counting of questions and an evaluation of how evenly they were distributed.

The data from Gong is one of the rare cases where sales research is large enough to be comparable with research in other industries. Use it. And if you want to test your own numbers, Salesprep gives you ten free discovery roleplays where each call gets automatic feedback on question volume, distribution and follow-up quality.