Demo engineering — what you show, when, and the conversion differential.

What the demo is actually for

The demo is the structured second call. It exists to convert a qualified discovery into a working opportunity — a prospect with named pain, named stakeholders, named timeline, and a champion who can describe the product to a colleague without the operator in the room. Not a product tour. Not a feature parade. The moment at which the buyer transitions from understanding what the product does to understanding what it does for them, specifically.

A demo executed against that frame converts at 50-65%. A one-way walkthrough converts at 15-25%. Same product, same buyer, same price point: a thirty- to forty-point spread, governed by which architecture the operator chose to run.

The demo-during-discovery anti-pattern

The most expensive demo failure happens before the demo exists. On the first call, the buyer asks "can you show me what it does," and the operator — anxious to provide value, differentiate, avoid the appearance of stalling — shares the screen and walks the product.

This is the demo-during-discovery anti-pattern, and it drops advancement to second call by 20-40% on cohorts where it occurs. The mechanism: the discovery questions that would have surfaced the buyer's specific pain go un-asked, because the call is now a product conversation rather than a pain conversation. The buyer leaves with an unstructured impression of the product. The operator leaves with no qualification data. The second call, if booked at all, begins with the operator having no anchor for which features matter to this buyer specifically.

The operational discipline — covered in Chapter 1 — is to refuse the demo on the discovery call. The phrasing is uniform: "I could share my screen now, but it would be the generic walkthrough. The version that's worth your time is the one where I show you exactly how we handle the three things you just told me about. Let me book thirty minutes next week." The advancement rate on calls where the operator holds this line is 1.3-1.6x the rate on calls where the operator capitulates.

The structured second-call demo architecture

A 30-35 minute demo call has a fixed architecture. The per-section time-boxes are not suggestions; they are the discipline that produces the upper band.

Opening — 2 minutes

Agenda, attendees, time-box. The operator names the structure: "We have 30 minutes. I want to spend the first three confirming what I heard last week, then 20-25 walking through how we'd solve it for you specifically, then five for questions and a next step." The buyer hears the shape of the call and tacitly agrees to it. Most common error: skipping this and opening with "let me share my screen."

Context recap — 3 minutes

The operator surfaces what was learned in discovery and confirms it. Not "as we discussed last time" — explicit recall: "Last week you mentioned the three things slowing down your renewal cycle were X, Y, and Z. The lift target was 15% within two quarters. Is that still the frame, or has anything shifted?"

This section does two operationally critical things. It signals that the operator listened, the single highest-impact trust signal in a B2B sales motion. And it surfaces any drift — the buyer who says "the priority shifted to W" has just saved the operator twenty wasted minutes of demonstrating the wrong product.

Demo — 20-25 minutes

Structured around the customer's specific problem, not the product's standard tour. The rule: show only features that directly address what the customer named in discovery. A product with thirty features and a buyer who named three problems gets three feature areas. The remaining twenty-seven exist; they are not shown.

The buyer told "we'll show you how we handle X, Y, and Z" who then sees exactly that — and only that — leaves with a coherent narrative. The buyer who instead sees a forty-five-minute walkthrough of all thirty features leaves with no narrative at all, only a vague sense that the product is large.

Q&A — 5-7 minutes

Real questions, not curated. The operator solicits explicitly: "What didn't I cover that you needed to see?"and pauses long enough for the buyer to actually answer. Most common Q&A failure: the operator answering a question they were asked with the answer to a question they wished they had been asked.

Next step — 3 minutes

A named next step with a named owner and a named date. The phrasing: "Based on what you saw, the next step is X. I'll do A, you'll do B, we'll reconvene on this date." Without this section, the demo converts at the lower band regardless of how well the prior twenty-seven minutes went.

The customer-in-driver-seat pattern

The structural difference between a demo that converts at 30% and one that converts at 55% is, more often than not, who is driving. The pattern is the discipline of asking the customer to make routing decisions throughout.

Before each section, the operator asks: "Where do you want me to start?" or "Which of these matters most?" or "Should we go deeper, or see the next piece?" The buyer who has just told the operator what to show next is, by their own choice, paying attention to what comes next. The buyer being passively walked through a curated sequence is, by minute fifteen, mentally elsewhere.

High-friction for operators with a rehearsed fixed flow. Also the difference between a buyer who later describes the product as "it does the three things I asked about" and one who describes it as "some software the rep showed me."

The demo-as-pitch anti-pattern

The opposing failure mode: a one-way feature parade. The operator shares their screen, narrates for forty minutes, asks "any questions?" at the end, books a follow-up. The buyer was a passenger. No engagement, no routing decisions, no surfacing of which features mattered.

Demo-as-pitch is the most reliable way to produce a 15-25% demo-to-opportunity rate on a product that should convert at 40-60%. It is the default for operators not trained on conversation architecture, because it is the lowest-cognitive-load option — read the script, run the slides, answer questions at the end. The buyer's silence reads as engagement; it is disengagement, and surfaces days later as the unreturned follow-up email.

Multi-stakeholder demo architecture

The customer who brings three or more attendees is most likely to close. The one who brings a single attendee is most likely to ghost. Multi-thread engagement — covered in Chapter 4 — begins on the demo call.

The per-attendee pattern: in the opening, the operator asks each attendee what they want from the call. As routing input, not formality. The economic buyer wants ROI math; the technical buyer wants integration architecture; the end user wants daily workflow. A demo that addresses all three with explicit hand-offs — "this next section is for the technical side of the room" — converts at the upper band. A generic walkthrough regardless of who is in the room converts at the lower band.

The rule: name each attendee's question in the opening, address each in the body, confirm each was answered before closing. A three-attendee demo that fails to engage the third will, in 60-75% of cases, be killed by that third attendee in the review the operator does not attend.

Screen-share discipline

Per-second, where does the customer look. The operator's screen contains, at any moment, the answer to the buyer's current question — and a dozen distractions. The discipline is to pause on important screens longer than is comfortable, move past unimportant ones faster than is rehearsed, and never narrate navigation.

Mechanics: maximize the relevant pane before sharing, hide notifications, close unrelated tabs, never demo with a debug console open. On an important screen, the operator pauses — five seconds of silence while the buyer absorbs is a feature, not a gap to fill. When transitioning, name the destination before clicking, so the buyer's attention arrives where the cursor does.

Demo environment — production vs sandbox

A production environment carries the operator's actual data — recognizable customers, real volume, the texture of live use. It also carries every imperfection of a live system: stale records, a feature mid-deploy, a stakeholder's name in a comment field. A sandbox is clean and controllable and demos the same way every time. It also reads, to an experienced buyer, as theater.

The pattern: production demos for buyers who have asked product-depth questions; sandbox demos for early-stage buyers who need the clean shape of the workflow. The failure mode is the sandbox that does not reflect production — the demo runs perfectly, the buyer signs, the implementation encounters the actual product, and the renewal twelve months later is a churn conversation. A sandbox that diverges from production by more than 10-15% in workflow or UX is a leading indicator of first-year churn at the 20-35% rate.

Recorded-demo handoff

A per-customer recorded version, sent post-call within 24 hours, is the highest-leverage multi-thread enabler available to the operator. The champion cannot, in practice, re-explain the product to four colleagues in the form the operator delivered it. The recording does that work. A five-minute custom intro — "this is the demo we ran for [champion]'s team last Tuesday, addressing X, Y, and Z" — turns a single demo into a re-deliverable artifact.

The lift on multi-thread engagement is 1.5-2x: deals where the recording was sent and forwarded reach three-plus stakeholders 50-70% of the time; deals with no recording reach three-plus stakeholders 25-40% of the time. The cost to the operator is fifteen minutes of editing.

Demo-as-conversation vs demo-as-presentation

The discipline that distinguishes the upper band: 90-second segments with conversation between them. Operator shows a feature for 60-90 seconds, stops, asks a routing question, waits. The buyer responds — a question, a comment, a redirect to a different area entirely.

Five 90-second segments with four 30-second interstitials is structurally different from a 30-minute monologue. Total information transferred is similar. Recall, engagement, and probability of advancing is materially different. Operators trained on demo-as-presentation resist on grounds that "I won't get through everything." The reframe: not getting through everything is the goal. The buyer who has time-blocked an unshown feature for the next call is the buyer who has committed to a next call.

Customer-led demo

The customer who asks to drive themselves is signaling either deep interest or deep skepticism. The handling is the same: hand them the keyboard. A buyer who can articulate on the call what they want to click and what they expect to happen has done more product internalization in five minutes than the operator can produce in twenty.

The risk is the buyer clicking into a broken or unfinished area. The mitigation is the pre-call walk-through — run the demo flow once, an hour before the call, note any in-progress areas to steer around. The conversion lift on customer-led demo segments, when they occur, is 1.4-1.8x on subsequent advancement.

Empirical demo length

20-25 minutes of demo content is the optimum, inside a 30-35 minute call. Beyond 25 minutes of demo content, recall drops, engagement drops, the buyer's mental model gets less coherent rather than more. Above 45 minutes it is observably counterproductive: the operator who runs a 60-minute demo is 30-50% less likely to advance than one who runs 25 minutes against the same buyer.

The reframe for operators who feel the product needs more time: every demo is followed by another call. The follow-up is not a failure mode; it is the structure. A buyer who needs to see more, after the right 25 minutes, is a buyer who is advancing.

The technical-deep-dive separate call

When the technical buyer needs depth — integration architecture, API behavior, security posture, data model — it does not belong in the business-buyer demo. The two audiences are listening for different things; satisfying both in the same call satisfies neither. The pattern: surface the technical depth requirement in the demo call's Q&A, name it, book a dedicated technical call within the week.

A separate technical call signals seriousness, brings in the technical resource (founder, solutions engineer, technical lead), and produces a per-technical-stakeholder relationship independent of the champion. This is the multi-thread architecture Chapter 4 builds on.

Common operator failures observed in production

• Demoing in discovery. Buyer asks, operator shares screen, discovery never happens. Advancement drops 20-40%. Highest-impact failure mode.
• No customer-in-driver-seat. The operator runs a fixed flow. The buyer is passive audience. Conversion sits at the lower band regardless of product fit.
• Demo-as-pitch. One-way feature parade, no routing questions, "any questions?" at the end. Silence reads as agreement; it is disengagement, and surfaces as the unreturned follow-up.
• No per-feature decision based on discovery. The operator shows all thirty features because they have not decided which three to show. The buyer leaves with no narrative.
• No recorded handoff. The champion is asked to re-explain the demo to four colleagues from memory. They do not commit. Multi-thread engagement stalls.
• Single demo for mixed audience. Three attendees, three different questions, one generic walkthrough. The unaddressed attendee kills the deal in the review the operator does not attend.
• Sandbox-doesn't-reflect-production. Demo runs cleanly, buyer signs, implementation fails. Renewal is a churn conversation. A 12-month-delayed failure that traces back to demo discipline.
• Demo length above 45 minutes. Recall drops, advancement drops. The operator's sense of having "covered everything" is inversely correlated with the buyer's sense of having understood anything.

Pre-demo checklist

• Discovery notes reviewed within 24 hours; the three named problems written down explicitly
• Demo flow chosen — which three to five feature areas, in what order
• Attendee list confirmed; per-attendee role and expected question identified
• Production vs sandbox decision made; chosen environment walked end-to-end one hour before the call
• Notifications hidden, unrelated tabs closed, screen-share pane maximized
• Recording enabled with the buyer's explicit consent in the opening
• Named next step drafted before the call begins, with two alternates in reserve
• Time-box agreed in the opening; operator's stop-watch visible

Where this fits

The demo is the second of three call types: discovery (Chapter 1), demo (this chapter), multi-thread expansion (Chapter 4). A discovery without demo discipline produces a demo with no anchor. A demo without multi-thread discipline produces a single-thread deal, which has a 30-50% probability of dying in internal politics regardless of how well the demo went.

The conversion-rate differential between 15-25% and 50-65% is the largest single operator-controlled lever in the sales motion. Not product fit. Not pricing. The discipline of running the call as a structured conversation against the buyer's specific named problems, with the buyer in the driver's seat, on a 25-minute demo body, with a recorded handoff sent within 24 hours and a named next step closed before the call ends.