Why deals go quiet after the demo, and how to hear it on the call

Almost every guide on post-demo silence treats it as a follow-up problem. The advice that follows is about cadences, sequences and break-up emails, which quietly assumes the call itself was fine and the failure happened afterwards in the inbox.

That assumption is where the guides go wrong, and it is the reason the advice does not work. The deal did not go quiet after the call. It went quiet during it, in a specific thirty-second stretch you can go and listen to.

Forrester surveyed more than 16,000 business buyers in 2024 and found that 86% of purchases stall somewhere in the buying process. Gartner puts "no decision" ahead of any individual competitor as a cause of lost deals. You lose to silence more often than you lose to a rival, and silence is the loss nobody reviews, because there is nothing to review. Nobody said no. Nobody pushed back on price. The thread stops, and three weeks later somebody moves the deal to closed lost and writes "went dark" in the notes.

This article is about the part that happens before that: what the last few minutes of a demo contain, why the usual way of measuring it produces numbers that cannot help you, and what to check on the next call.

A deal goes quiet in the last three minutes, not in the inbox

The shape is always the same. The session runs well. The buyer asks real questions, says "interesting" at least twice, mentions a colleague who should see this. Around minute fifty someone says "I'll send over a proposal and we'll take it from there." The buyer says "perfect, thanks." Both sides hang up satisfied.

Nothing was agreed. No date, no named person, no action.

From that moment the deal depends entirely on the buyer pushing it forward inside their own company, against everything else on their plate, with no external pressure and no commitment they made out loud. Most of the time that does not happen, and it has very little to do with how much they liked the product.

The practical change is small. Treat "no date agreed" as the end of the deal's active life rather than the start of a follow-up period. It changes what happens next, and it stops the pipeline carrying deals that are not moving.

A warm call and an agreed call sound almost the same

The words buyers use to be polite and the words they use to commit are largely the same words. Written down in a table the difference is obvious. Heard live, in the last two minutes of a call that went well, it is close to invisible.

Only the last row commits anyone to anything.

The consequence is not that somebody misread the room. It is that the deal now sits in the pipeline at a stage it did not earn. Forecast reviews start from a number that includes it, and the calls that most need a second look are the ones nobody flags, because nothing went wrong on them.

This is also the decision we had to make explicit when we built the call scoring model. Warmth is not one of the inputs. A call where the buyer never named a date fails on that criterion regardless of how well the rest of it went.

A default worth starting from: check the last 90 seconds of the recording. If the buyer did not say a date in that window, log the call as no next step, whatever the tone was. Do not argue about it in the review. The argument over whether someone "seemed committed" is exactly what makes the metric useless.

That argument is also expensive in a way that is easy to miss: it happens in the review, after the deal is already cold, and it produces a discussion instead of a correction. We can show you where that gap opens up on your own recordings.

The marker is who said the date out loud

A next step counts as agreed only when the buyer said the date and the action. Not the rep. The buyer.

It sounds like a small distinction and it is not. When the rep proposes the time, the buyer has given up nothing. When the buyer proposes it, they have spent something: a slot in their own calendar, named in front of their own colleagues. Backing out now costs them a little face.

Compare two endings.

Rep: I'll send over some times for next week. Buyer: Sounds good.

Buyer: Thursday afternoon works. I'll bring our security lead, she'll want to see the data handling part.

The first produces silence. The second produces a meeting, a stakeholder and a topic for the next call, all from letting the buyer say the sentence.

The operational difference shows up in what you can do afterwards. With the first ending there is nothing to act on: no date to protect, no person to prepare for, no subject to research. With the second, the next call has an agenda before anyone has opened their calendar.

And unlike most things people argue about in call reviews, this one is not a judgement about tone or rapport. It is a question of who spoke the words, and anyone reviewing the recording can answer it in thirty seconds.

Ranking the next step against other factors tells you nothing

Here is the part most write-ups skip, and the reason a number of sales dashboards quietly mislead the people reading them.

Rank the factors that correlate with won deals and "next step agreed" comes out at or near the top, every time, on any dataset. That result is worthless. The next step is not one cause sitting alongside others, it is part of the outcome. A deal that progressed has a next step because it progressed. The correlation is guaranteed by the definition rather than discovered in the data.

We ran into this directly. The first version of our criteria ranking put "next step fixed" at the top and kept it there, and the chart was useless in a particularly convincing way: it looked authoritative, it was technically correct, and it told a sales leader nothing they could act on. In Spellit Actions these criteria are now stripped out of the influence ranking by a dedicated filter, with a note in the interface explaining why.

The operational cost of leaving them in is specific. The circular criterion occupies the top of the ranking, the criteria that actually vary between reps get pushed down where nobody reads them, and the weekly coaching conversation ends up being about the thing everybody already agreed on.

So what belongs in the ranking? Keep two separate lists. One for the things you measure to understand what varies across the team. One for the things you check on every call because they are non-negotiable. The next step belongs on the second list, and putting it on the first is what makes the first list useless.

A dashboard that confirms what you already knew feels like management, which is what makes it worse than no dashboard. We can run your last month of calls and show you what actually varies.

Your objection-handling rate is probably computed against the wrong denominator

The same error shows up in the number most teams quote about objections, and it takes about five minutes to check.

Objection handling is usually computed as calls where the objection was handled, divided by all calls. That denominator is wrong. Calls where no objection was raised do not belong in it. Include them and the rate drifts upward on nothing more than a quiet week, because objection-free calls mechanically improve a number that is supposed to measure what happens under pressure.

The correct denominator is calls where an objection actually occurred, identified by its own marker rather than inferred from whether somebody filled in a field.

We got this wrong in our first version and it took a while to spot, because a broken metric of this kind does not look broken. It produced numbers that were plausible, moved week to week, and moved for reasons nobody could explain. That last part is the tell: if a rate changes and no one can name what changed underneath it, check the denominator before you check the team.

There is a small habit that prevents the whole class of error. For any rate you report, write the denominator next to it in words rather than in formula. "Out of calls where an objection came up" is a sentence somebody can disagree with. "Objection handling: 78%" is not.

Run the same check on your next-step metric. Computed across every call in the system, it includes calls where a next step makes no sense, and describes your call mix rather than your team.

Three questions that close a demo

Each one is built to make the buyer state the commitment instead of accepting yours.

1. "What would need to happen on your side for this to move forward?" Asked before you propose anything. A vague answer means a vague deal, and you have learned that while they are still on the line rather than three weeks later.
2. "Who else needs to see this, and what will they want to know?" Produces the stakeholder and the topic in one answer. A name without a topic is a forward that goes nowhere.
3. "What does the next step look like from here?" Then stop talking. The next four seconds are uncomfortable and that is the entire point. Fill the silence and you have proposed the next step yourself, and the commitment is gone.

Listen for the walk-around. "I'll come back to you once I've had a think" is a refusal to answer question three. Treat it as information rather than a soft yes, and ask what specifically they want to think about. The answer to that is usually the real objection, arriving late.

For deals that already went quiet, the move is not a fifth email. Go back to the last real conversation and find what was left unresolved, because that is what the buyer is sitting with.

What others say

The cadence advice is not wrong about the symptom. Deals do go quiet, and sequences do sometimes restart them. It is wrong about where the problem lives. A sequence cannot repair a commitment that was never made, it can only make the absence less visible to you for another two weeks.

There is a second body of writing on this topic that needs a warning label. Several widely shared figures on post-demo ghosting come from studies of a few hundred calls with no disclosed method, no period and no company mix. At least one now circulates with two entirely different sets of percentages, depending on whether you read the original page or a summary of it.

We ran into this while researching this article. Of six sources with usable-looking numbers, three had no primary publication behind them, and two of those quoted suspiciously similar sample sizes. We dropped all three.

Before quoting a statistic about your own funnel, find the sample size and the definition behind it. If either is missing, what you have is a slogan, and slogans are a poor foundation for a quarterly plan.

What to do next

Pick one recording from last week where the deal has since gone quiet. Skip to the last 90 seconds. Note who said the date, if anyone did.

That is a thirty-second check and it will usually settle the question on its own. If it turns out most of your quiet deals ended the same way, the fix is not a new sequence. It is question three, asked on every demo, starting with the next one.

See it across a month instead of one call. Bring us your recordings and we will show you how many ended without an agreed next step, and which of those deals are still worth reopening. Book a 45-minute demo.