Discovery

SPIN: the research behind question order

Neil Rackham's SPIN Selling (McGraw-Hill, 1988) reported research by the Huthwaite corporation, described in the book's catalogue record as a twelve-year study developed from 35,000 sales calls. Its central finding was that the techniques developed for small, simple sales, including pushy closing techniques, did not work in large sales, and that what distinguished successful sellers in large sales was the kind of questions they asked.

SPIN names four kinds. Situation questions establish facts (how many people, which systems, what process). Problem questions surface difficulties (where does it break, what takes too long). Implication questions explore consequences (what does that delay cost, who else is affected, what happens if it continues). Need-payoff questions invite the buyer to state the value of a solution in their own words (if that were fixed, what would it mean for the quarter?).

The practical lesson is about proportion and order. Situation questions are necessary but tedious for the buyer, and in 2026 many of them can be answered by research before the call (Lesson 3.1). Implication and need-payoff questions are where a buyer comes to understand, and articulate, why the problem is worth solving, and that articulation is what they will repeat to their buying group.

What recent call data adds

Conversation-analytics vendors have since measured discovery at scale on recorded calls. Their data is correlational and drawn from their own customers, but it is large and specific.

Gong, analyzing 519,000 recorded discovery calls in 2017, reported that asking roughly 11 to 14 questions over the call correlated with the greatest success, that top sellers spread their questions through the conversation while average sellers front-loaded them, and that the most effective discovery calls uncovered three or four business problems. Fewer suggested the buyer lacked enough pain to act; many more suggested too many priorities to focus on any. A later Gong analysis of calls with C-suite buyers found that asking too many questions there hurt the sale.

On talk time, Gong's 2025 update of its talk-to-listen analysis covered 326,000 sales calls of at least ten minutes. The average seller talked about 60% of the time. Won deals averaged 57% seller talk time and lost deals 62%, top performers kept a steady ratio whether they won or lost, and lost deals often featured long seller monologues. Sellers on won deals asked about 15 to 16 questions per call against about 20 on lost deals. The company's often-quoted 2016 'golden ratio' of 43% talking to 57% listening came from an earlier, smaller analysis.

The pattern across both datasets: fewer, better questions; more back-and-forth; a few problems explored properly. An interrogation of twenty questions is as bad a sign as a monologue.

Running the conversation

Prepare by doing the situation work in advance. Read the account, the person, and the deal history; know what they have already told you or your website. Open by confirming rather than asking: 'From what you wrote, it sounds like month-end reconciliation is the bottleneck. Is that right, or is there more to it?'

Follow the buyer's energy toward problems, then stay there. When a problem appears, ask what it costs, who else feels it, what has been tried, and what happens if nothing changes. Resist the urge to pitch the moment you hear a problem you solve; the implication questions are where the buyer builds their own case.

Ask about the decision as well as the problem. Who else cares about this? How have you bought something like this before? What would need to be true for this to be a priority this quarter? These are the qualification questions from Lesson 1.4, asked naturally.

Close with an agreed next step that involves the buyer doing something: introducing a colleague, sharing data, attending a session with their team. Lesson 4.2 explains why that matters more than a friendly 'let's stay in touch'.

Capturing discovery in Kavanah

Discovery is wasted if it lives only in the seller's memory, and in a buying group of a dozen it has to accumulate across several conversations. Record it on the deal.

If you use Kavanah's call features, meetings recorded inside Kavanah can be transcribed and turned into notes and tasks, and an opt-in notetaker can join your outside Zoom, Google Meet, Teams or Webex meetings, announcing that it is recording, then producing notes, action items and tasks. It is off by default and spends credits and meeting-bot time, so an admin enables it deliberately in Settings. For calls held elsewhere, record_deal_activity appends a note or a manual call log to the deal's timeline; ask the agent to log 'Discovery call with Jane Doe: three problems, ...' and it lands where the whole deal team can read it.

Update the qualification fields from Lesson 1.4 with what you learned, set each contact's next step and its due date with update_deal_contact, and if a colleague owes the buyer something, track_waiting_on lets the agent chase it later. Read back the summary before the next meeting so you open with what they told you, not with what you have forgotten.