# Sales Roleplay

> Sales Roleplay

- Skill: `ingridleiria/sales-roleplay` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add ingridleiria/sales-roleplay`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ingridleiria/sales-roleplay/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: ingridleiria (https://skillmd.com/u/ingridleiria)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/ingridleiria/sales-roleplay

---


# Sales Roleplay

Sellers rehearse against people who want them to do well. A colleague plays the buyer, agrees with the value proposition, asks the objection the seller already has an answer for, and everyone leaves feeling prepared. The first genuine resistance arrives in front of a real buyer with real revenue attached, and the seller discovers under load that they have never said the difficult sentence out loud.

The cost sits in a small number of expensive moments: the first meeting that does not earn a second, the discovery call that produced a demo request and no understanding, the price conversation where a discount was given before it was asked for. Those moments are cheap to rehearse and expensive to lose. A rehearsal only helps if the buyer in it behaves like a buyer, which means withholding, unhurried, and unimpressed by anything the seller has not earned. A buyer who agrees quickly teaches nothing and quietly certifies a bad habit.

## When to use this, and when not to

Use it before a specific call that matters, particularly a first meeting with a named account or a price conversation with a live deal attached. Use it to drill one habit the seller has already named, which is the highest-yield version of this skill. Use it to onboard a new seller before they touch a real prospect, to pressure-test new positioning or a new price, and to rehearse a recovery conversation with an account that went quiet.

Do not use it to review a call that already happened; that is `sales-call-analysis`, which works from the transcript rather than from memory. Do not use it to rehearse an internal conversation with an executive; `principal-simulator` models a specific known individual rather than an invented buyer. Do not use it to produce written material for testing a scoring tool, which is `demo-call-transcript-generator`. Do not use it to judge whether a seller is good, since a rehearsal under observation measures nerve as much as skill; capability judgements belong in `sales-team-competency-assessment`. Do not use it to build the deliverable that follows discovery, which is `discovery-to-proposal-deck`.

## What you need before starting

**What is being sold, at roughly what price and in what currency, to whom.** The price band changes buyer behaviour more than anything else: a decision at 4,000 US dollars a year is made by one person in one call, and a decision at 400,000 involves a committee, procurement and a security review. Missing: ask for a band rather than a figure, and failing that assume a mid-market annual contract near 40,000 US dollars and say you have assumed both the band and the currency.

**Who the buyer is.** Title, company type, industry, and their relationship to the problem. Missing: pick from the persona reference and announce the choice in one line, so the seller can correct it rather than discovering mid-call that they are selling to the wrong person.

**The stage.** Cold outreach, discovery, demo, proposal walk-through, objection gauntlet, negotiation, closing, or a full run. Missing: default to discovery, because it is where most deals are actually lost and where most sellers are weakest.

**What the seller wants to work on, in their own words.** "I fold on price" and "I talk too much in discovery" are far better session inputs than any stage label, because they name the habit the debrief will be judged against. Missing: ask for the last call that did not go well and what they would change; that answer is the goal.

**How experienced the seller is, and how long you have.** Experience sets the difficulty; a twelve-minute drill and a forty-minute simulation are different sessions with different coaching. Missing: assume twenty minutes, infer experience from the first two exchanges, and adjust the buyer's resistance silently rather than asking again mid-session.

**For a rehearsal tied to a real account, whatever is actually known.** Notes, an email thread, the last call summary. Missing: run a generic persona and say plainly that the buyer is invented, so nothing from the rehearsal is later recorded as fact about the account.

## The method

1. **Fix the session contract in four lines before any dialogue.** What is being sold, who the buyer is, which stage, and the habit being worked on, written where both parties can see it. Sessions that skip this drift into a general conversation about selling, which is comfortable and produces nothing. If the seller wants to start now, state the four assumptions in one line and begin.

2. **Build the buyer before speaking as them.** Fix five things privately: the opening mood, one competing priority that makes this less urgent than the seller assumes, one hidden truth explaining their real motivation, one stakeholder they will not mention unless asked directly, and one past disappointment with a similar purchase. These make the buyer resist coherently rather than randomly. A buyer improvised line by line contradicts itself by minute ten, and once the seller notices, the rehearsal is over even if the dialogue continues.

3. **Set difficulty deliberately, and say nothing about it.** First three months in the job: a buyer who is distracted but fair and rewards a decent question visibly. Two years or more: polite, evasive, and gives ground only to a question they have not been asked before. Rehearsing a specific high-value call: the hardest plausible version of that buyer, because the point is to find the crack now rather than on the day.

4. **Open in character and stay there.** The buyer speaks first only in inbound and negotiation scenarios. Do not narrate, do not label what you are doing, and do not soften a hard response with an aside. Break character only at a stop word, at a planned coaching point, or when the seller has visibly lost the thread for two consecutive turns.

5. **Move the buyer along the unlock ladder in response to question quality, never in response to time elapsed.** The ladder is below. This is the core mechanic: the buyer's openness is the feedback signal, delivered inside the fiction, and it teaches faster than commentary afterwards. When a question earns a level, mark it in the dialogue: "nobody here has asked me that."

6. **Release objections at a controlled density.** Early stages get roughly one objection every five minutes, always in the presentable form rather than the true one. Later stages get them in clusters, because a real proposal conversation delivers three at once. The escalation rule: if the seller answers the presentable objection well and then asks what else is on the buyer's mind, give them the true one; if they answer well and move on, keep it, let the deal stall on it, and raise it in the debrief as what they left in the room.

7. **Step out to coach at four points.** After the first three minutes, because the opening is fixable and sets everything after it; after the first real objection; after the moment the seller earns or loses access to the real problem; and at the end. More often than this fragments the conversation, less often and the seller repeats one error for twenty minutes.

8. **Coach in four fixed lines, always with language they can reuse.**

```
WORKING:     what landed, and why it worked
DEVELOPING:  what was close but blunt, and the sharper version
MISSED:      what was skipped, and what it costs later in the deal
TRY THIS:    the exact words, in quotation marks, ready to say
```

The TRY THIS line is the only one that reliably changes behaviour. A note with no sentence the seller can say out loud is an observation, not coaching.

9. **Rerun the hardest thirty seconds, at least twice.** Rewind to the line before the failure and play it again, with the buyer responding differently each time so the seller cannot memorise a path. Three attempts at one moment beats one attempt at six, and the improvement is visible inside the session, which is what makes the seller believe the rest.

10. **Debrief against the stated goal, not against your favourite fundamentals.** If the seller came to work on price, the debrief leads with price even when their discovery was worse. Score out of ten with one sentence of rationale, and name one pattern rather than a list of moments.

11. **Set the next session before ending.** One scenario, one habit, and what would count as fixed. A rehearsal programme with no next scenario becomes an occasional novelty.

## The unlock ladder

The buyer occupies one of five levels of openness. Movement up requires the seller to earn it; movement down is fast and should be used.

| Level | What the buyer gives | What earns the next level |
| --- | --- | --- |
| 0. Closed | Short answers, checks the time, offers to receive a deck | A reason to spend ten more minutes that is about them, not the product |
| 1. Transactional | Answers factual questions, volunteers nothing | An open question about how the problem shows up in their week |
| 2. Descriptive | Describes the problem in their own words | A second level on the same pain, and a question about consequence |
| 3. Consequential | Names what it costs in their numbers, and who else is affected | A question about process, decision and timing that does not sound like qualification |
| 4. Collaborative | Names stakeholders, timeline and money, and starts problem-solving with the seller | Nothing; hold here and let the seller advance the deal |

Drop a level for a leading question, a pitch before level 2, a feature answer to a business question, or ninety seconds of talking without a question. Drop two for a discount offered unprompted, and let the buyer take it and then ask for more, which is what actually happens.

## Objections, and what is usually behind them

Play the left column. Know the middle column, which is what the seller has to reach. The right column is what a good answer looks like, and is what the coaching hands over.

| Objection as spoken | What is usually behind it | The move that works |
| --- | --- | --- |
| We are happy with what we have | Change is risky and nobody is blamed for the current state | Quantify what staying the same costs this year, in their numbers |
| There is no budget | No urgency, or no value established with whoever holds it | Find the compelling event, then ask what it would take to be funded |
| We are evaluating other options | Nothing has differentiated yet | Ask what criteria they are using, and offer one they had not considered |
| Send me some information | No trust, and a polite exit | Qualify the interest, offer a shorter and more specific next step |
| It is too expensive | Compared against the wrong thing, usually against zero | Reframe against the cost of the problem and of a failed alternative |
| We need to think about it | Something specific is unresolved and unsaid | Ask what they would need to know to decide either way |
| Now is not a good time | No trigger, and a competing priority that is winning | Price the delay and attach it to a date they already own |
| I need to talk to someone internally | The decision map was never established | Welcome it, and use it to map who else, in what order, on what criteria |

Two rules for the buyer. The presentable version always comes out first. And never accept an answer that addresses the words rather than the concern, because that is the failure the seller needs to feel.

## Stage notes

| Stage | The buyer's posture, the seller's job, and what the coaching watches | Passed when |
| --- | --- | --- |
| Cold outreach | Mildly annoyed, thirty seconds of patience. Earn two more minutes, do not qualify and do not sell. Did they open in the buyer's world and ask for a specific next step rather than fold at the first pushback | The first sentence contains no product noun, and a dated next step is asked for once, after at most one pushback |
| Discovery | Cautious, not hostile. Understand the problem well enough that the buyer feels heard, and map the decision. Watch the second-level questions and whether they know who else has to agree | At least three follow-up questions go a level deeper than the answer they follow, and every name required to approve is on paper by the end |
| Demo | Judging whether this was worth the hour. Show only what connects to a pain already heard. Count the features shown that nobody asked about | No more than one feature shown that traces to nothing said in discovery, and every other screen is introduced by the pain it answers |
| Proposal | Looking for a reason to decide either way. Defend value rather than read the document. Was value anchored before the price, and what happened in the silence after the number | Value is stated before the number, and after the number the seller says nothing until the buyer speaks |
| Negotiation | Testing limits, watching for desperation. Trade rather than give. Did the concessions buy anything, and do they know their walk-away | Every concession is paired with something received in the same sentence, and the walk-away is stated aloud before the call, unprompted |
| Closing | Near a decision and still hesitant. Surface the remaining concern. Did they ask for the business at all, and was the ask apologetic | The business is asked for in one direct sentence with no hedge in front of it, and the remaining concern is named by the buyer rather than guessed at |

Each threshold is observable from the transcript by somebody who was not in the room, which is the test a coaching note has to meet. A stage where the threshold is met twice running comes off the practice list.

Pre-built buyers across common industries, each with an opening mood, real pains, a hidden truth, a withheld stakeholder and a set of objections, are in `references/personas-and-objections.md`. Invent the names and company details; keep the pressure structure.

## Worked example

**Situation.** Marta Feld sells route-planning software to distributors at an average annual contract of about 28,000 US dollars. All figures in this example are in US dollars. In eleven months she had closed four deals and lost six, and in five of the six she had discounted before the buyer asked, ending at an average of 19 percent off list. Her own account was blunt: "I fold on price." A live deal with a regional distributor had a proposal call eight days out.

**Task.** Get her through a proposal walk-through and a price conversation without conceding unprompted. Success was defined before starting: she states the price, stops talking, and offers nothing until the buyer has asked for something specific.

**Action.** The first session ran as a full simulation, hello to next step, forty minutes, coaching held to the end. That format was abandoned after one run: the debrief carried eleven observations, four of them about discovery, and Marta left with a general sense of having done badly and nothing she could practise. The wrong turn was structural. A full simulation demonstrates the problem rather than treating it, and running it first spent the session on a diagnosis she had already supplied for free.

The rebuild was narrow. The buyer was a Director of Transport at a four-depot distributor, twenty-two minutes of availability, a failed telematics rollout two years earlier, and one hidden truth held in reserve: a depot consolidation already approved for the following year, which made a twelve-month payback far more attractive than he would admit. The withheld stakeholder was a finance director who had rejected two capital requests that year.

The session then ran the same ninety seconds six times: the seller states the annual price, the buyer says "that is a lot more than I was expecting", and then says nothing at all.

Attempt one: Marta filled the silence in four seconds and offered a phased rollout; the buyer took the phasing and then asked for a discount on top, which is what a real buyer does with an unprompted concession. Attempt two: two seconds of silence, then a feature list, and the buyer asked for the deck to share internally. Attempt three: seven seconds of silence, and the buyer spoke first, naming the real constraint, that anything above 25,000 needed the finance director she had never asked about. Attempts four and five drilled what to ask next. Attempt six released the hidden truth: asked what was changing next year, the buyer mentioned the consolidation and the payback framing became available.

Two other notes came out of it. Her value sentence ran fifty-five words and lost the buyer halfway; the rewrite was seventeen words and used the buyer's phrase "empty running". And she had been introducing price at the end of the walk-through, so the number arrived when attention was lowest and every remaining minute became a negotiation.

**Result.** Three sessions across six days, sixty-five minutes of rehearsal. On the live call she stated the price at minute fourteen rather than minute thirty-one and did not concede in the silence. The buyer named the finance director as approver, which had not surfaced in three previous conversations with that account. The deal closed six weeks later at 26,500 against a list of 28,000, in exchange for a case study and a two-year term, which is a trade rather than a concession. A single closed deal does not prove a habit changed; the defensible claim is narrower and still useful, that in her next four opportunities she introduced price before the final third of the call in all four.

### A second scenario, where it goes differently

Ben Aluko, second week in the job, same product, no sales background. The same buyer at the same difficulty produced a collapse: by minute four he was reading website copy aloud, and by minute six he had apologised twice.

Three things changed. Difficulty dropped to distracted but fair, because a beginner has to feel the mechanism working before they can be made to work for it. The coach demonstrated first, playing the seller for two minutes and then swapping, which gives a beginner a model to imitate rather than a standard to fail against. And the session worked one skill only, the second-level question, as eight short exchanges rather than a continuous call. The debrief carried no score, because a number given to someone in week two is noise they will carry as a verdict; instead, one thing to keep doing, one sentence to memorise, and the same drill in three days.

What did not change is that the buyer never volunteers the real problem. Lowering difficulty means responding more generously to a decent question, not answering a question that was never asked.

## Output

Open every session with the contract block, and close every session with the debrief block. Both are fixed, so that two sessions weeks apart are comparable.

```
SESSION CONTRACT
Selling:      [what, price band, to whom]
Buyer:        [title, company type, industry]
Stage:        [outreach / discovery / demo / proposal / objections / negotiation / closing / full]
Working on:   [the seller's own words]
Difficulty:   [fair / evasive / hardest plausible]
Length:       [minutes]
```

```
DEBRIEF
Practised:            [stage, industry, scenario in one line]
Highest level reached: [0 to 4 on the unlock ladder, and the question that earned it]
Biggest strength:      [the single most effective thing they did, quoted]
Highest leverage fix:  [the one change with the largest effect on the next real call]
Pattern to watch:      [the habit visible more than once, not a single moment]
Left in the room:      [what the buyer was willing to say and was never asked]
Score out of ten:      [honest, with one sentence of rationale]
Say this next time:    [one sentence, in quotation marks, ready to use]
Next practice:         [scenario, habit, and what would count as fixed]
```

Where the session is tied to a live account, add one line naming which parts of the buyer were invented, so nothing from the rehearsal is later mistaken for something the real buyer said.

## Failure modes

**A buyer who is not on the ladder.** The buyer who caves: the seller reaches level 3 without a second-level question, or an objection is dropped after one answer, which teaches that their current questions suffice. The brick wall, commoner among experienced coaches: three good moves earn nothing and the session ends with no usable language. Fix both by honouring the ladder in both directions rather than playing a mood.

**Coaching in adjectives.** "Your discovery was weak" is not usable. Recognise it by scanning for quotation marks: none means no language was handed over. Fix by writing the TRY THIS line first and building the note around it.

**Rehearsing the pitch instead of the questions.** Recognise it when most of the session is the seller talking and most of the coaching is about phrasing. Deals are lost in what was never asked far more often than in how it was described. Fix by disallowing any statement that is not a response to something the buyer just said.

**The session that becomes an appraisal.** Recognise it when a manager is watching and the seller performs for them instead of working on the weakness. Nobody rehearses a weakness in front of the person who writes their review. Fix by running it privately and reporting only the pattern, or by giving the manager the buyer role so they are inside the fiction rather than judging it.

**Score inflation.** Recognise it when everyone scores seven. Anchor the scale: four gets a polite email, six earns a next step, eight is a call where the buyer said something they had not told the previous vendor.

**Persona drift.** Recognise it when the buyer contradicts itself on company size, stakeholders or timeline. Once the seller notices, they stop treating the buyer as real. Fix by rereading the five buyer facts at each coaching break.

**Inventing a real person's psychology.** Recognise it when the session asserts what a specific named individual believes or fears. That is invention presented as intelligence, and it ends up in a customer record. Keep the account context factual and the personality invented.

## Edge cases

**Fifteen minutes before a real call.** Do not run a simulation. Run three exchanges: the opening sentence, the hardest question the buyer will ask, and the next step request. Coach those three and stop. A partial rehearsal of the highest-risk moments beats a rushed full run.

**The problem is the offer, not the skill.** Recognise it when the objections are all true and none has a good answer: the price really is above the market, the integration really does not exist. Stop and say so, since drilling a seller against an unanswerable objection teaches them to bluff, and route the finding to whoever owns pricing or product.

**Practising in a second language.** Slow the buyer's speech and shorten their sentences, and separate the two failure types in the debrief, since language errors and selling errors are different problems and conflating them makes the session about the wrong one. Hand over the TRY THIS line in writing.

**A team session with an audience.** Rotate the seller role at a fixed five-minute cut, and have the audience predict what the buyer will do next before it happens. Prediction is a better group exercise than critique and does not damage the person in the chair.

**A lost-deal recovery or a paused account.** The buyer starts at level 0 with one added fact: they said no once and remember why. The job is to earn a reason to reopen rather than re-pitch. `account-reengagement-plan` covers the written plan that precedes the rehearsal.

**A committee call.** Play two buyers with different criteria, an operational sponsor who wants the problem solved and a finance approver who wants risk contained, and let them disagree in front of the seller. Coach whether the seller noticed and used the disagreement, or talked over it.

**No context and no patience for setup.** Start with a discovery call, a mid-market buyer, and the assumptions stated in one line. An imperfect rehearsal that starts beats a perfect one that gets scheduled.

## Quality bar

- The buyer never volunteers the real problem and never rewards a question that was not earned.
- Every coaching note contains a sentence in quotation marks the seller can say out loud.
- The session was judged against the goal the seller stated, not against a generic model of a good call.
- At least one moment was rerun until it improved, and the improvement is described concretely.
- The buyer's five fixed facts held for the whole session without contradiction.
- The debrief names one pattern, not a list of moments.
- The score is honest, including when it is low, and the scale is anchored to observable outcomes.
- The next session is specified with a scenario, a habit, and what would count as fixed.

## Adapting this to your context

The buyer behaviour, the difficulty settings and the price bands come from business-to-business software and services selling into mid-market accounts. The two-role separation is the method; almost everything else is a dial.

- **The price band.** It drives buyer behaviour more than the industry does, so set it and the currency in the session contract. A four thousand a year decision and a four hundred thousand decision are different rehearsals, and running the wrong one teaches a habit that will not transfer.
- **Who the buyer is.** The personas assume a commercial buyer with discretion. A public sector or grant-funded buyer cannot be sold to outside a process, so the rehearsal becomes a clarification meeting and the withheld truth is about evaluation criteria rather than budget. An agency rehearses objections about taste and about who else is pitching.
- **The stages.** Cold outreach through closing suits a seller-led cycle. Where buying is self-serve and conversation starts after a trial, drop discovery and rehearse expansion and renewal instead.
- **The debrief score.** Useful for an experienced seller working one named habit, harmful in someone's second week. Decide before the session whether this one carries a score.

- **What not to change.** The buyer never volunteers the real problem, and the coach steps fully out of character to debrief.

## Related skills

`account-reengagement-plan` produces the written plan and the objection set for a paused account, which this skill then rehearses. `competitive-battlecard` supplies the competitor claims the buyer should throw. `sales-call-analysis` reads the real call afterwards and names the same habits from evidence rather than memory, which is the honest check on whether the rehearsal worked. `demo-call-transcript-generator` writes synthetic transcripts for testing analysis tooling, a different job from live practice. `sales-team-competency-assessment` is where a capability judgement belongs, not a rehearsal. `discovery-to-proposal-deck` picks up once discovery has landed. `principal-simulator` is the equivalent for an internal executive audience rather than a buyer.