Listening as a technical skill

Sales Academy · from no sales experience to landing an AE job, with a UK life-sciences and health-tech specialism. All companies and people in the examples are fictional unless stated otherwise.

Every other module teaches a technique: how to qualify, run discovery, handle an objection, build a business case. This one teaches the raw skills those techniques are made of — and in practice they come first. A beginner with an excellent process and poor communication still fails: the discovery framework is sound, but they talk over the answer that mattered; the follow-up is well-timed, but nobody reads past line two; the demo is tailored, but the delivery is a monotone gallop and the room disengages at minute four.

Communication craft is also the part of the job you can practise today, alone, with no prospect, product or quota — which makes it the fastest-moving item on your development list, and the most visible in an interview, where the interviewer is not only listening to your answers but watching how you handle a conversation.

What you'll learn

• Listening as a technical skill — levels of listening, what "active listening" means in practice, note-taking that doesn't kill rapport, how long to hold a silence, and how to summarise without sounding like a chatbot.
• Questions that open people up — open and closed, funnelling, the follow-up that gets the real answer, labelling and mirroring, leading questions, and where curiosity becomes interrogation.
• Writing that gets replied to — the clarity rules, the "so what" test, and three worked before/after rewrites.
• Phone and video presence — pace, pitch, volume, filler words, smiling on the phone, camera and lighting, nerves, and recovering from a stumble.
• Structuring what you say — the 30-second version of anything, the pyramid principle, the rule of three, story shape, and how to answer a question you can't answer.
• Reading the room — buying signals versus politeness, disengagement on video, when to stop talking, multi-person dynamics.
• Deliberate practice — how to actually improve, with a concrete four-week plan.

1. Listening as a technical skill

Why this is the load-bearing skill

Listening feels passive, which is why beginners treat it as the thing that happens between their turns. Everything of value in a sales conversation — the pain you'll quantify, the phrase you'll reuse in the proposal, the name of the person who really decides — arrives through listening, and most of it arrives once.

Ralph Nichols, who ran listening studies at the University of Minnesota from the 1940s, put average listening efficiency at roughly a quarter: immediately after hearing someone talk, the average listener remembers about half, falling to 25% two months later.

Voice from the field

"Listening is a skill. It can be improved through training and practice, just as can reading, writing and speaking."
— Ralph G. Nichols & Leonard A. Stevens, Are You Listening? (1957)

The levels of listening

Covey's five levels (The 7 Habits of Highly Effective People, 1989) run ignoring, pretending, selective, attentive, empathic — only the last, listening to understand rather than to reply, does any work. Scharmer's four levels (Theory U, developed at MIT) are more useful on a call: downloading (hearing only what confirms what you already believe), factual (noticing what contradicts you), empathic (seeing it from inside their world), generative (the conversation changes both parties).

Downloading is the rep who has decided this is a "compliance play" and hears every sentence as evidence for it. Empathic is noticing that the lab manager's real problem isn't the software but being left to finish someone else's implementation. Make empathic your working standard.

What "active listening" actually means

Four behaviours, all observable, none of them nodding:

1. Attending — nothing else on screen. Attention is audible; people can hear you typing.
2. Following — short encouragers that cede the floor: "mm", "right", "go on". Fewer than you think; British buyers find constant American-style affirmation grating.
3. Reflecting — giving back what you heard, in their words, to test whether you got it.
4. Not preparing your reply — the hard one. Composing your next question drops you to factual listening at best. Fix it mechanically: write the question down as three words and get back into the conversation.

Field note

The most reliable tell that you weren't really listening is asking a question the buyer already answered. It happens constantly to new sellers, and the buyer's read isn't "inexperienced" — it's "they weren't paying attention to me." Don't cover it: "Sorry, you did say that and I moved past it. Can I go back?" Buyers forgive the slip and remember the recovery.

Note-taking that doesn't kill rapport

You need notes — a call you can't recall is a call you can't follow up. But heavy note-taking destroys what it captures: the buyer can see you've stopped listening and started transcribing.

• Capture their words, not your summary. Write "the year we lost", not "negative prior experience". Verbatim phrases are the raw material of every follow-up you'll write.
• Nouns and numbers, not sentences. You can reconstruct meaning; you can't reconstruct a figure.
• Say what you're doing, once: "I'll take notes while you talk — if I go quiet, that's why."
• Look up on video when something lands, and write the call up within ten minutes, while recall is intact.

Silence, and how long to hold it

Silence is the highest-return skill here, and the hardest, because conversation is tuned against it. A large cross-linguistic study (Stivers and colleagues, PNAS, 2009) found that across ten languages, responses cluster within roughly 200 milliseconds of the end of a question; every community studied avoids long gaps. Holding three seconds therefore feels like a mistake. It isn't: the first answer is the prepared one, and the second, after the pause, is the real one.

• After you ask a question: hold until they speak. Not three seconds — until they speak.
• After they finish an answer: count two. This is where the sentence beginning "and honestly…" arrives, usually the most valuable sentence on the call.
• On the phone, silence feels twice as long — no visual channel confirms anyone is still there.
• Don't hold silence to manipulate. If someone is genuinely stuck, rescue them: "Sorry — big question. Let me narrow it."

The drill: for one week, count two before you speak in every internal meeting. Unbearable, then invisible.

Summarising to confirm — without sounding robotic

Summarising proves you listened, corrects misunderstandings before they get expensive, and makes the buyer confirm or refine their position aloud. "So what I'm hearing is…" is fine once; three times and it sounds like a technique. Four fixes:

1. Vary the opener. "Let me check I've got this." / "Playing that back —" / Or no preamble at all: "The blocker is the migration, not the software."
2. Summarise the meaning, not the transcript. A robotic summary lists what they said in order; a good one compresses it into the shape of the problem.
3. Include something they didn't say explicitly. "…and after last time, you're not taking a vendor's word for it." Wrong, and they correct you, which is also information.
4. End on a genuine check. "What have I got wrong?" beats "Is that fair?", because it invites correction rather than agreement.

Summarise after developing a major pain, before changing topic, and before agreeing next steps.

2. Questions that open people up

Open, closed, and the beginner's error

An open question can't be answered yes/no or with one fact: "How does a new access request move through the lab?" A closed question can: "Do you use a LIMS?"

The beginner's error is asking closed questions when they want information, because closed feels safer: shorter, implying you already know the answer. It also produces one-word answers, so you ask twelve more — which is how a conversation becomes an interrogation. The subtler error runs the other way: "How do you feel about next steps?" is a worse close than "Can we put 45 minutes in for Thursday with your data manager?"

Rule of thumb: open to explore, closed to confirm and commit.

Funnelling

Funnelling means going broad to narrow inside one topic instead of hopping between topics:

1. "Walk me through what happens when a researcher requests samples." (broad, open)
2. "Where does that usually get stuck?" (narrower)
3. "What happens when the consent status isn't in the system?" (narrower still)
4. "Roughly how many samples are in that state?" (closed, quantifying)

Four questions, one thread, ending in a number. One question each on samples, staffing, budget and timelines yields four shallow facts instead. The discipline: don't leave a topic until you've hit a number or a dead end.

The follow-up question that gets the real answer

Most beginners take the surface answer and move on. The value is nearly always in the second question, and nearly always one of five:

• "Tell me more about that." The lowest-effort deep-dive there is; works after any answer.
• "What happens when that goes wrong?" Turns a process description into a consequence.
• "What's driving that?" Softer than "why", which can sound accusatory in English.
• "Can you give me an example?" Abstractions become stories, and stories contain specifics.
• "What have you tried?" Reveals the history and the political landmines.

Labelling and mirroring

Two techniques from Chris Voss, the FBI's former lead international kidnapping negotiator, whose Never Split the Difference (2016) is the standard source. Both are tools of what he calls tactical empathy: understanding the other side's perspective well enough to name it, which is not agreeing with it.

Mirroring is repeating the last one to three words, with an upward inflection.

Buyer: "It didn't go well." You: "It didn't go well?"

It reliably produces elaboration without steering, because you've added no content of your own. Use it when the buyer says something loaded and you don't know which part is loaded.

Labelling is tentatively naming the emotion or dynamic you observe: "It sounds like…", "It seems like there's history here." Labels invite expansion when you're right and get corrected when you're wrong — and correction is more informative than agreement.

Two cautions. Don't stack them, or you sound like you're running a script. And hedge the opener: "You're frustrated" is a diagnosis and invites denial; "It sounds like this has been frustrating" invites detail.

Leading questions, and spotting your own

A leading question contains its own answer: "So faster turnaround would be valuable, wouldn't it?" The buyer says yes, you write down "values turnaround", and you've learned only that people are polite. A yes you engineered is worth nothing in a forecast.

Three tells: it ends in a tag ("…right?", "…wouldn't it?"); it names the benefit your product happens to deliver; you already know what you want them to say. The fix is to strip the answer out — not "would faster turnaround help?" but "what does the current turnaround cost you?"

Curiosity versus interrogation

The line isn't how many questions you ask — it's whether you give anything back. Interrogation is question, answer, question, answer: you extract, they deplete, and they can feel it. Curiosity is question, answer, reaction, deeper question — the reaction being a mirror, a label, a short reflection, or a summary.

Analysis of large call-recording datasets by Gong, a conversation-intelligence vendor, points to 11–14 questions as the productive range on a discovery call — a correlation within their customer base, not a law. The test, as in Module 04: at the end, did the buyer get something?

Annotated example: the same moment, mishandled and handled

Fictional. Alex is a new AE at "Lattice Bio", selling sample-tracking software to research labs; Dr Ruth Adeyemi manages the Marchwood university biobank. Both takes start from the same sentence, ninety seconds into a first call.

Take one — mishandled

Ruth: We looked at something like this about two years ago. It didn't go well.

Alex: Right, yes, and honestly a lot of people say that — the market has moved a long way since then. Two years ago most of these systems were really built for pharma QC labs and just didn't fit academic biobanking, whereas ours was designed from the start around consented human samples, so the chain-of-custody piece is native rather than bolted on afterwards. Was it a chain-of-custody problem you had, or was it more of an integration thing, or was it cost?

→ Alex hears "objection" and defends. Then he stacks three questions, letting Ruth pick the easiest — or none. And every word of product content is a guess.

Ruth: A bit of everything, I suppose.

→ The disengagement answer: vague, agreeable, closes the topic. It means "I'm not unpacking this for you."

Alex: Sure, and that's exactly what we hear. So what I'd say is, where we're genuinely different — I'll show you properly in the demo — is the audit trail. Every movement of every aliquot is logged, and the ethics reporting comes out as a one-click export. How many samples are you holding at the moment?

→ Pitching into a vacuum, then a topic hop because the thread died. A demo built on this call would target a problem Alex invented.

Ruth: Around four hundred thousand.

Alex: Great, that's a really good size for us.

→ "Good size for us" evaluates Ruth out loud, and tells her the conversation is about Alex's pipeline.

Take two — handled

Ruth: We looked at something like this about two years ago. It didn't go well.

Alex: It didn't go well.

→ A four-word mirror, then nothing. Alex doesn't know what's behind that sentence, so he adds nothing of his own.

Ruth: No. We spent about eight months on it. The system went in, the vendor's implementation lead left halfway through, and we ended up running the old spreadsheet alongside it anyway. Eventually we switched the thing off. My team still calls it "the year we lost".

→ The mirror bought a paragraph: a duration, a root cause, a workaround, an outcome, and a phrase in her own words that Alex will reuse for months.

Alex: (pause) It sounds like the software itself wasn't really the problem — the problem was being left to finish somebody else's implementation.

→ A hedged label, going one step beyond what Ruth said. Wrong, and she corrects him, which is also useful. Note he did not counter with "our implementation team is excellent": claims are worthless here.

Ruth: That's exactly it. The software was probably fine. Nobody could tell us how to migrate fifteen years of legacy records, and that was the part I actually needed help with.

Alex: Tell me more about the legacy records.

→ Seven words, and the whole call now turns on the answer.

Ruth: Fifteen years of an Access database, three naming conventions, and about forty thousand samples where consent status only exists on paper in a filing cabinet upstairs. Anything we put in has to reconcile with that, and nobody wants to touch it.

→ Quantified pain, without a single quantifying question.

Alex: So — let me check I've got this. The blocker isn't choosing a system. It's that nobody has offered you a credible plan for migrating fifteen years of mixed-quality records, including forty thousand where the consent only exists on paper. And after last time, you're not taking anyone's word for it.

Ruth: Yes. If you can show me how you'd do that, I'll handle the budget conversation myself.

→ The summary compresses rather than repeats, and adds the unspoken part ("after last time"). Ruth responds by volunteering to run the internal sale.

The comparison

Take two is longer and contains zero product content — both features, not costs. The toolkit: one mirror, one hedged label, one "tell me more", one summary, two silences.