Use last quarter's demos to fix next quarter's
Average a quarter of recorded demos of one deck and the same moments go quiet. Score the deck, not the rep, and turn the drops into an edit list with owners.
Jared Hawkins-Young · Founder, Lost Star TechnologyOctober 8, 20265 min readPart 9 of 13Key takeaways
- A quarter of recorded demos of the same deck, averaged minute by minute, shows where the room goes quiet regardless of which rep presented.
- Engagement can be read from the transcript the recorder already produces: buyer talk share, buyer questions, rep monologue length and the tone of buyer replies.
- The output is an edit list with owners, not a rep scorecard: cut the architecture section, move pricing after setup, move the live run before pricing, rewrite the contract language.
- It is retention analysis pointed at a sales deck, and it needs at least 20 demos on one deck version before a section is cut.
To improve a sales demo deck, read your own recorded demos of it, all of them, side by side. Across a quarter the same moments go quiet in most rooms, and that turns deck changes from a matter of taste into an edit list with owners. Score the deck, not the rep. This part shows how, using a quarter of demos that includes Northwind's.
Where does the room go quiet?
Engagement here does not need a camera or a new tool. It comes from the transcript your recorder already produces, with speakers labelled, read minute by minute for four things:
- how much of the minute the buyer is talking,
- how many questions the buyer asks,
- how long the rep has been talking without interruption,
- and the tone of what the buyer says back.
One demo on its own is noise. A strong rep carries a weak slide, and a distracted buyer sinks a good one. Average 38 demos of the same deck and the individual rep and the individual buyer wash out. What is left is the deck. It is the same extraction that types signals in part 2, pointed at the order things get shown in.
The shape in Figure 1 is typical of what this kind of reading surfaces. The room lifts when the rep opens on the buyer's problem in the buyer's own words, pulled from the discovery call. It lifts again when the rep runs the product on one of the buyer's own recorded calls. It falls in the architecture section, where the questions stop. It falls when the price lands before anyone has framed what setup takes. And it falls furthest on the contract terms.
Why score the deck instead of the rep?
Conversation platforms have scored calls for years: talk ratio, questions asked, monologue length. That is coaching, and it is aimed at a person. This is aimed at an artifact. If 30 of 38 demos go quiet on the same slide, the problem is the slide, and coaching four account executives one at a time on a slide that fails for everybody spends your one-on-ones on the wrong thing.
Reps also take it better. Nobody is being ranked. The output is a list of changes to a shared deck, and the reps are the ones who get to stop presenting the section that loses the room.
What goes on the edit list?
Each row is a section, what the recordings show, the edit, and who owns it. At a company with four account executives, most of the owners are you.
| Section | What the recordings show | The edit | Owner |
|---|---|---|---|
| Architecture | Questions stop. The rep talks alone for about four minutes. Rarely referenced afterwards. | Cut it to one appendix slide | Enablement (you) |
| Pricing | A sharp drop when the number lands before the setup effort is framed. | Move it after setup and integration | Demo flow |
| Live run on their call | The most reliable recovery in the set, sitting at slide 14 where attention is already thin. | Move it before pricing | Demo flow |
| Contract terms | The deepest and most consistent drop. The language is doing damage the offer is not. | Rewrite the language | Legal (founder or counsel) |
| Competitor comparison | Holds when the rep names what the recorder's CRM sync does not write. Drops when the rep is dismissive. | Name the gaps, in writing | Battlecard |
Run it every quarter. In this illustrative quarter, fourteen slides become nine, and the nine are the ones that survived contact with real rooms. The edits arrive as proposals. Whether the contract language actually changes is a decision for whoever owns the contract, which is exactly the line part 13 draws: scoring the demo runs unattended, deciding what a finding means stays with a person.
Where does the idea come from?
From retention analysis. YouTube Studio's report on key moments for audience retention shows a creator four things: the intro, meaning the share of the audience still watching after the first 30 seconds; top moments, where almost nobody dropped off; spikes, the parts that were rewatched or shared; and dips, the parts that were skipped or where viewers stopped watching. Creators cut the dips. They can see the exact second an audience leaves.
Sales decks never had that, because nobody was measuring the room. Once demos are recorded, the same analysis applies, and the deck stops being a matter of taste. The opening on the buyer's own sentence is the deck's version of a strong intro, and it is the same move the proposal makes in part 8.
How many demos do you need before you change a slide?
More than most small teams have in a month, which is why this runs quarterly. Two to six account executives produce somewhere between twenty and fifty demos a quarter. Our working rule, not a statistical threshold, is to wait for at least 20 demos on the same deck version before cutting a section, and to change one section at a time so the next quarter tells you which change worked.
- Keep deck versions apart. Averaging demos across two versions of the deck blurs exactly the moment you are trying to find.
- Check the section per rep before cutting it. If one rep holds the room through pricing and three do not, the edit might be a talk track, not a slide.
- Change one thing per cycle. The outbound series makes the same argument for email, where openings, value lines and asks are scored separately so a result can be traced to the change that caused it.
Who this is not for: if every demo is built from scratch for each prospect, there is no shared deck to score. Score the sections you do repeat, which for most teams are the opening, pricing, security and the contract.
Next in the series
Everything so far happens after the call. Part 10, Hand the rep the answer while the call is still open, uses the same corpus of recordings inside a live call, and sets out the guardrails that make that safe to build.
Common questions
- How do I improve a sales demo deck using call recordings?
- Collect every recorded demo of the same deck version for a quarter, read each transcript minute by minute for buyer talk share, buyer questions, rep monologue length and the tone of buyer replies, and average across demos. The sections where engagement drops in most demos go on an edit list with an owner for each change.
- Is demo scoring a way to grade my reps?
- No. Averaging many demos removes the individual rep and buyer, so what remains describes the deck. The output is a list of changes to a shared artifact. If one rep holds a section that loses the room for everyone else, that is a talk track to share, not a score.
- How many recorded demos do I need?
- As a working rule, at least 20 demos on the same deck version before cutting a section, which for a team of two to six account executives usually means a quarter. Change one section per cycle so the next quarter shows which change worked.
Sources
Written by Jared Hawkins-Young from the system he builds and runs. Diagrams are drawn from the same specs as the software. Published October 8, 2026. Illustrative numbers are labelled as such; every other figure carries its source.