A good sales conversation runs both ways. It ends with one next step and a date on it. In B2B it starts even earlier. The buyer has looked you up before anyone says a word.
A deal moves through six of these conversations. Each one has a different job.
Every dialogue on this page is constructed. The company, the people and the offer are made up. For each one I say why it works, and whether that reason comes from a cited study or from my own campaigns.
Stages 1 to 3 are my team's daily work. Stages 4 to 6 happen inside the client's sales team, so those examples lean on research instead of my campaign numbers.
What does a good sales conversation example look like, stage by stage?
Here is the cast for every example. Ledgerline is a made-up firm that audits payroll and subcontractor payments for construction companies, and Maya Lund is its founder. Tom Reyes is the invented finance director of a regional builder in the UK.
An outbound team writes to Tom in Maya's name and hands over once a call is booked.
| Stage | Who owns it | Job of the conversation | Success signal |
|---|---|---|---|
| 1. First touch | Outbound team | Earn a reply from a stranger | Any human reply |
| 2. Reply | Outbound team | Turn one reply into a real exchange | The prospect writes a second time |
| 3. Booking | Outbound team | Agree a time, an agenda and who joins | Invite accepted |
| 4. Discovery | Client sales team | Learn the problem in the buyer's words | Buyer names a cost or a deadline |
| 5. Doubt | Client sales team | Find out what the hesitation really means | A specific concern, or a clear no |
| 6. Next step | Client sales team | Fix the decision path and the next meeting | Date, attendees, written recap |
Why does each stage need its own conversation? Look at one real month. In one of our campaigns, run from 5 LinkedIn profiles plus email, we contacted 2,187 people in 30 days. The result: 309 replies -> 26 conversations -> 14 booked meetings -> 11 held -> 2 signed contracts.
The steepest drop sits between the reply and the conversation: 309 replies, only 26 real exchanges. So I never treat a reply as the finish line. The order never changes: conversations first, then calls, then clients.
How do you start a conversation with a prospect who has never heard of you?
Constructed example, stage 1. Maya's LinkedIn message goes out after Tom accepts her connection request:
Maya: Thanks for connecting, Tom. A question from someone who spends her days inside builders' payroll: when subcontractor invoices and site timesheets disagree at month end, who on your team settles it?
A few days later, an email follows up on the same question.
Subject: Month-end timesheet gaps
Hi Tom, following up from LinkedIn. Builders usually catch payroll errors on their own site staff. Where they miss errors is the point where subcontractor invoices meet timesheets, because nobody owns that check. Ledgerline audits exactly that point for regional builders. I am not asking you to buy anything. Would 15 minutes next Tuesday or Thursday be useful, to walk through how you check it today? If this is not your area, who should I ask?
Maya
Compare a weak version: "Ledgerline is a leading payroll audit firm helping construction companies save money. Can I send our deck?" Pitching in the first message is the fastest way I know to get ignored.
The strong version works because a call is free for Maya and expensive for Tom: 15 minutes taken from a busy month. So the message sells the call and keeps the service for later.
Tom will also check Maya's profile and Ledgerline's site before he replies, so whatever he finds there is part of this conversation too. Nobody buys from an email. The decision comes after they have vetted you. What Tom finds on his own does a lot of the selling before Maya ever speaks.
That is why I never set the monthly goal at "book 10 calls". I set it at 50+ conversations started. The calls follow.
What do you say when the prospect replies, and when do you follow up?
Replies come in a handful of types. Each type needs a different next message.
| Reply | Next message | Aim | When to stop |
|---|---|---|---|
| A curious question ("Why?") | Answer in two lines, then ask one question back | Keep the exchange going | When they stop asking |
| "Send me some info" | One question that tells you what to send | Turn a brush-off into a stated need | After you send the one item they chose |
| "Not now" | Ask when, and what will have changed by then | Get a date | Never drop it: log the date in your CRM |
| Wrong person | Thank them and ask for the right name | A referral you can mention | Once you have the name |
| Polite no | Thank them and ask one optional question about why | Learn something, leave the door open | Right away: remove them from the sequence |
| Silence | A new angle in each follow-up | A first reply | When the full sequence has run |
Constructed example, stage 2. This is the "send info" reply, turned into a question:
Tom: Interesting. Send me some info.
Maya: Happy to. So I send the right thing: is the bigger headache subcontractor overpayments or overtime on your site crews?
Tom: Subcontractors. We had a bad quarter on it.
Maya: Then I will send our two-page subcontractor reconciliation checklist. If it raises questions, 15 minutes with me will answer them faster than another PDF.
For us, follow-up number three usually lands best, provided there is real breathing room between messages. One or two messages never get you there, so my sequences run to 5 or 6 touches at minimum, ideally across LinkedIn and email.
I also put a person on the inbox, so a "send info" never sits unanswered over a weekend.
In my cold email campaigns the reply rate is usually about 1%, often 0.5% once thousands of emails go out per day, and meetings come from 10% to 50% of replies. The funnel at the top of this page mixes LinkedIn and email, so it is not a cold email benchmark.
Very specific, heavily personalized campaigns can reach 15% to 17% replies. Averages sink as sending volume grows, so treat these figures as ranges with no promise attached.
How do you turn a yes into a booked meeting without pushing?
Constructed example, stage 3:
Tom: Fine, let's talk.
Maya: Great. Tuesday at 10 am or Thursday at 2 pm UK time, 20 minutes on Zoom? I will bring the checklist and a sample reconciliation from a builder your size, with names removed. You will leave knowing whether your month-end check misses anything. Bring your payroll lead if you like.
Tom: Thursday. I'll bring Priya.
Maya: Done. The invite is on its way with the agenda in it.
This works because Maya answers the question she raised in stage 1: is this worth Tom's time? Two slots, a stated agenda and a clear outcome leave Tom nothing to negotiate except the day.
Booked and attended are different numbers. In the funnel at the top of this page, 14 meetings were booked but only 11 were held.
An agenda in the invite is a cheap way to narrow that gap. For the full workflow, see this guide to turning a positive reply into a booked call.
What does a discovery conversation sound like when it goes well?
Constructed example, stage 4 (client team). From here on the examples are research-led, and Maya runs the call herself.
Maya: Walk me through what happens between a subcontractor invoice arriving and the payment going out.
Tom: It hits accounts payable, the site manager signs off the hours, and we pay on the 25th.
Maya: When you say the site manager signs off the hours, what is he checking them against?
Tom: Honestly? His memory, and the gate log when it works.
Priya: The gate log broke on two sites in spring.
Maya: So the check rests on one signature per site, and two sites ran without the log for a while. Is that right?
Tom: Right. We only found one overpayment because a subcontractor flagged it himself.
Maya: Is there anything we haven't covered that would make this a priority, or rule it out?
Tom: Audit season. The auditors arrive in February.
Maya uses four question types in that exchange. They come from Edward Miles's work in the Negotiation Journal, as summarized by Harvard Law School's PON negotiation program:
- An open question to start
- An information probe on "signs off"
- A summary probe that Tom can correct
- A clearinghouse question at the end, which asks what is still missing
The same summary warns against mixing an open and a closed question in one turn. Ask "How do you handle approvals? Is it working?" and you may get only the yes or no. One question per turn.
Follow-up questions also make people like you. Across three studies, people who asked more questions, especially follow-ups, were better liked by their partners (Huang and colleagues, Journal of Personality and Social Psychology, 2017). Those studies covered get-to-know-you chats and speed dating, not sales, so treat the finding as a sensible habit with no deal data behind it.
There is also vendor data on talk time. Gong's help center puts the best-converting split on B2B calls at roughly 43:57, rep talking 43% and listening 57%. It also recommends monologues of about two and a half minutes at most, and 18 or more questions per hour (Gong).
How should a seller handle doubt or disinterest on a live call?
Constructed example, stage 5 (client team).
Tom: Look, this is interesting, but February is close and I'm not sure we can take on a project right now.
Maya: When you say you're not sure you can take it on, is that your team's time or the budget?
Tom: Time. Priya is already stretched.
Maya: Then the real question is how much of Priya's time this needs. Most of the work sits on our side. From her, it is about two hours to pull the files. Does that change things, or is the timing still wrong?
Tom: Two hours we can find.
Maya skips the rebuttal and uses an information probe instead. "Not sure" turned out to mean two hours of Priya's time, a problem she can solve.
Sellers often misread moments like this. A Harvard Business School working paper found that salespeople give too much weight to a handful of vivid signs of disinterest and misjudge which calls will fail (Harvard Business School working paper). In the same research, a language-model "stopping agent" cut the time spent on failed calls by 54% and kept nearly all sales. Moving the saved time to new calls raised expected sales by up to 37%.
The preprint and the HBS AI Institute summary give the scale: 11,627 calls, over 94% of them failed, 169 seconds on average, and the model could predict failure from about the 60-second point.
- A single sharp "no" tells you less than it feels like it does. Probe it once.
- Repeated, specific signals deserve respect (my reading, not the paper's), and then you end the call politely: "Sounds like the timing is wrong. Can I check back after the audit?"
Objections that arrive by email are a different case, covered in these answers to cold email objections. For reusable phrasing across stages, see sales talk tracks.
How do you end a sales conversation with a real next step?
Constructed example, stage 6 (client team).
Maya: Here is what I heard: two sites ran without a working gate log, you found one overpayment by luck, and the auditors arrive in February. Who else signs off a project like this?
Tom: Our MD, Alan, for anything above a few thousand.
Maya: Then let's book 30 minutes with Alan next Wednesday: you, Priya and me. I will bring a one-page scope with Priya's two hours written in, and send you a recap tonight with that date. Fair?
Tom: Fair. Wednesday works.
Maya does three things here:
- She summarizes the problem in Tom's own words.
- She asks who else decides.
- She sets a dated meeting with named attendees and puts it in writing the same day.
"Let's stay in touch" is a polite exit with no date attached.
In the funnel at the top of this page, the step from held meetings to signed contracts is closing work, and no outbound team can do it for you. For more on that gap, read why a booked call is worth little before a closer takes it.
Which popular sales conversation rules hold up against evidence?
| Popular rule | What the evidence says | Rating |
|---|---|---|
| "Ask lots of questions" | Follow-up questions raised liking in social chats and speed dating; sales outcomes were not tested | Research-backed, outside sales |
| "Hit the ideal talk ratio" | The figure in the discovery section comes from Gong's own call data, with no published method | Vendor data |
| "Never quit a call early" | In the HBS stopping research, cutting failing calls kept nearly all sales | Contradicted by research (consumer telecom) |
| "Pitch your value in the first line" | Pitch-first openers get ignored in the outbound campaigns behind stage 1 | Experience |
| "Pause three seconds before you answer" | Gong suggests waiting 0.6 to 1 second after the customer stops talking | Vendor guidance, not a three-second rule |
Check any example you adapt against these five rules:
- One main ask per message.
- One question per turn.
- Summarize before you move forward.
- Name the next step, with a date.
- Log which stage the conversation reached, so you can see where people drop out.
A reply is not a meeting, and a meeting is not a deal. Give every stage its own conversation.
Curious where your six conversations lose people? Book a call and share a recording of one. We will look at the next step you asked for.
Questions people ask about sales conversations
What is the 3-3-3 rule in sales?
There is no single agreed 3-3-3 rule. Different trainers attach the name to different habits, so ask whoever quotes it which version they mean and what result backs it. My filter for any numbered rule: does it come with a sample, a setting and a date? If not, test it on a small batch of your own conversations before adopting it.
The rule I actually run my campaigns on is simpler. The more versions you test, the sooner one of them wins. And no copy rule saves a bad list or a weak sender.
What are the 5 C's of sales?
There is no standard set of five C's; each author lists different ones. Use whichever version you meet as a preparation checklist, then test each item against one question: does it help a specific conversation, such as the reply or the booking, do its job? If it does not, drop it.
Should you use a word-for-word script?
Use the dialogues for structure and keep your own words. A script read aloud sounds read, and prospects rarely answer the way the script expects. What helps more is deciding three things before the call: the one question you open with, the point where the offer comes up, and the next step you will ask for.
How long should a first sales call last?
I found no published study that gives one right length for a B2B first call. The rule that matters is to put the length in the invite and end on time. If the conversation needs more, ask for a second meeting with the people the decision depends on, which doubles as your next step.