The numbers worth watching in your own conversations
Conversation analytics arrived with inside sales and brought one number everybody knows: talk-to-listen ratio. It is real and it is genuinely predictive in aggregate, and taken alone as a personal target it produces some of the worst sales behaviour there is, because a rep trying to talk less says less without asking more, and silence from a rep is not the same as participation from a customer. The numbers worth watching are the ones that describe what the customer did, not what you did.
What is wrong with talk-to-listen on its own
It measures airtime and infers engagement, and the inference breaks in both directions. A rep who talks for seventy percent of a technical diagnosis may be doing exactly the right thing, because module 3 is about narrating out loud and that is airtime with a purpose. A rep who talks for thirty percent while the customer explains at length about their neighbour has a great ratio and no information.
It is also the easiest number to game and the gaming is harmful. Told to talk less, reps go quiet rather than becoming curious, and appointments get shorter and worse. Nothing about the ratio distinguishes a well-placed silence from a rep who has stopped engaging.
In aggregate across many appointments it is a reasonable signal, which is why it works as a coaching input for a team. As a personal target for the next appointment it is close to useless, and it is where most conversation-intelligence tooling stops.
Count what the customer did
The single most useful number in a field-sales appointment is how many questions the customer asked you. It is a direct measure of engagement and it cannot be gamed by the rep going quiet. A customer asking eight questions is participating; a customer asking none has either understood everything or stopped following, and module 3 explains why the second is more likely.
The second is how many questions you asked before the price came out. Not total, before, because questions after a number are usually defence. Reps who close well tend to ask more than they expect and much more than they remember.
The third is the seconds of silence after you say the number, which module 5.2 is entirely about, and which is the one number that most reliably surprises people when they measure it.
And the fourth is whether the customer ever stated the conclusion themselves. Not a count, a yes or no: did they say some version of "so the real problem is not the ceiling" before you did. That is the strongest single predictor in this whole course of an appointment going well, and it is entirely absent from every analytics product we are aware of.
The four, and what a bad number means
- Customer questions asked
- Low is the warning sign. Under about three in a ninety-minute appointment usually means they stopped following rather than that everything was clear, and the likeliest cause is trade vocabulary used without being defined.
- Your questions before the price
- Low means you presented rather than diagnosed. Reps consistently overestimate this one, which is why it is worth counting rather than recalling.
- Seconds of silence after the number
- Under about three means you filled it, and the real objection probably never surfaced. This is the cheapest number on the list to move and the one with the largest effect.
- Did they state the conclusion
- A yes or no. If it is consistently no, the diagnosis is being delivered as a verdict rather than narrated, and that is module 3.1.
What it sounds like
Two appointments by the same rep, both ninety minutes, both ending without a sale. The counts are what separate them.
What to notice. By talk-to-listen alone, A is the better appointment and B is the rep talking too much. Every other number says the opposite: in B the customer asked nine questions, reached the conclusion themselves, and had six seconds to react to the price. A is a rep who stayed quiet and learned nothing. This is the whole argument for not managing to the ratio, and note that the rep's own sense of the two appointments was exactly backwards.
The mistake: managing yourself to a ratio
A rep with a talk-time target starts monitoring their own airtime mid-appointment, which is a genuinely bad thing to be doing while someone is telling you about their roof. Attention that should be on the customer goes to a self-imposed metric, and the appointment gets worse in a way the metric cannot see.
It also produces a specific failure: the rep who stops narrating the diagnosis. Module 3 is about talking through your reasoning while you inspect, and that is the highest-value airtime in the appointment. Cutting it to hit a ratio removes the most persuasive thing you do.
The related error is treating any of these numbers as a target rather than as a diagnostic. Counting is for noticing, and the action is always a specific behaviour change, not the number. Nobody improved by deciding to have more customer questions; they improved by defining a trade word once and then finding out what happened.
And the last one is measuring in aggregate only. A monthly average across forty appointments hides everything. These are per-appointment numbers, reviewed on one appointment a week, which is the previous lesson.
Counting one appointment
- How many questions did the customer ask me?
- How many did I ask before the price came out?
- How many seconds of silence after I said the number?
- Did the customer state the conclusion themselves, or did I?
- Did I use a trade word without defining it once?
- Which single one of these am I trying to move next week?
- Am I about to set myself a talk-time target instead of a behaviour?
Read next
- Listening back to your own appointment The ten-minute loop these numbers belong inside.
- Diagnosing out loud Why high talk time during the diagnosis is the right kind of talking.
So is talk-to-listen useless?
Not useless, misused. Across a team over many appointments it is a reasonable signal and worth looking at. As a target handed to an individual for their next appointment it produces reps who go quiet rather than curious, which is worse than where they started.
What is a good number of customer questions?
We would be inventing precision to give you one. What is safe to say is that very few, under about three in a long appointment, is a warning rather than a sign of clarity, and that it usually traces to vocabulary or to a diagnosis delivered as a verdict.
Can I count these without a recording?
Roughly, for two of them, if you note it immediately afterwards. Customer questions and whether they stated the conclusion are recoverable from memory within ten minutes. The silence after the number is not, and it is the one where memory is most flattering.
Should the business track these?
Some of them, carefully, and it is a different exercise from personal review. Module 11.2 is about what happens when the wrong thing gets measured across a team, which is a bigger risk than measuring nothing.