AI Prompts for Sales: Prepare, Draft, and Review Your Work

Useful AI prompts for sales help you prepare questions, draft accurate follow-ups, and check what you can promise. Give the assistant approved sources, the purpose of the conversation, and explicit limits on speculation and commitments. Review the result before using it with a customer.

The reusable part is your method: research instructions, approved messaging, and review checks. fenbs can keep that work context available to connected assistants while customer records remain in your approved customer system. Start with one repeatable task that currently needs too much correction.

What context should you include without copying a customer record?

Use only the information your organization permits in the chosen AI tool. A prompt can work with a fictional company, public sources, approved product facts, and a general description of the sales stage. It does not need an exported contact list or a complete customer relationship management record.

Separate the inputs by purpose. Public information can support preparation; an approved product sheet can support a capability statement; a customer’s confirmed words can establish an agreed next step when you are authorized to use them. Label interpretations so a plausible guess does not become an alleged customer requirement.

• Conversation goal: the useful decision or understanding you want to reach.
• Approved offer: current capabilities, exclusions, and commitments a salesperson may make.
• Sources: labeled public references or other material authorized for this use.
• Known limits: what you do not know and what must be confirmed by a person.
• Output checks: factual support, respectful wording, and a realistic next action.

Keep lasting instructions separate from the current request, as illustrated in system prompt examples. Update the request when the stage or approved offer changes; a familiar account name does not make old assumptions current.

Which prompt makes pre-call research useful?

Ask for a concise preparation note with evidence and questions. Avoid a request to discover a company’s hidden pain points. Public information may suggest a topic to ask about, but it rarely tells you how a particular team works or whether it has budget and authority to act.

Prepare a call outline using only the public sources supplied below.
Our approved offer: [paste].
Call goal: [goal].
Return: confirmed public facts with source IDs, possible relevant topics labeled as hypotheses, five neutral discovery questions, and information we still need.
For each question, explain which assumption it tests.
Do not infer budget, buying intent, private systems, or personal characteristics. Do not invent recent events or customer relationships.
Sources and retrieval dates: [paste approved public sources]

Open each cited page and confirm both the fact and its date. A careers page mentioning a tool does not establish that the whole business uses it, and an old expansion announcement does not establish an active purchasing project. Keep the distinction visible in your preparation note.

Suppose a fictional regional manufacturer lists several locations. You could ask how teams coordinate requests across locations. You should not state that its current process is broken. The prompt should help you arrive curious and prepared, with room for the customer to correct your starting picture.

How can you generate discovery questions without steering the answers?

Ask about the work before proposing your solution. Good discovery questions invite a description of current steps, consequences, and desired outcomes. Require follow-up questions for different answers so the assistant does not produce a script that assumes every conversation follows the same path.

Create a discovery guide for [type of work] in [general business context].
Goal: understand the current process before proposing a solution.
Return six open questions covering current steps, people involved, exceptions, consequences, desired outcomes, and decision process.
For each, include a neutral follow-up and a note about what the answer would help us understand.
Remove leading wording and assumptions that a problem exists.
Include one question that could show our offer is not a suitable fit.
Do not script pressure tactics or fabricate urgency.

Read the questions aloud. If one sounds like a disguised product pitch, rewrite it in the customer’s language. “How do you handle approvals today?” leaves space for a useful answer; a question built around your product’s terminology may teach you more about your script than their process.

Choose a few questions rather than reading the entire guide. Listen for the answer that changes the next question. Save the general improvements to your discovery method in fenbs Skills and prompts, without turning the shared prompt library into a second customer database.

What prompt produces a follow-up you can actually send?

Use a follow-up prompt that distinguishes what was said, what was agreed, and what remains open. Supply only notes authorized for the tool. Ask the assistant to leave a visible placeholder when an owner, date, or commitment is missing instead of filling it with something plausible.

Draft a concise follow-up using only these authorized notes and approved product facts.
Structure: what we understood, agreed next steps, open questions, and a polite closing.
For each next step, include the stated owner and date. Use [confirm owner] or [confirm date] when absent.
Do not add promises, discounts, delivery dates, customer quotations, or implied agreement.
Keep tentative ideas labeled as tentative.
Return the draft plus a private review list of every commitment and its supporting note. Do not send anything.
Notes: [paste authorized notes]
Approved facts: [paste]

Compare the draft with your notes line by line. “We discussed a trial” does not mean “we agreed to begin a trial.” A proposed next meeting is not a confirmed appointment. Confirm the recipients and any dates separately, and remove internal review notes before sending the final message.

For commercial email in the US, review the FTC’s CAN-SPAM compliance guide↗ (opens in a new tab). It covers business-to-business commercial email too, including truthful headers and subjects, required identification and address information, and a way to opt out. A well-written draft does not complete those sending checks.

How should an assistant help with objections and promises?

Use objection practice to identify missing information rather than to force a rebuttal. Ask the assistant to distinguish a misunderstanding, a genuine mismatch, and a question you need someone else to answer. Sometimes the accurate response is that your offer does not meet the requirement.

Review this fictional objection against our approved offer.
Objection: [fictional wording].
Approved capabilities and limits: [paste].
Return: what the concern might mean, one clarifying question, an evidence-supported response, and where we should acknowledge a limitation.
Label interpretations as hypotheses. Do not invent a roadmap item, integration, customer result, or guarantee.
List questions requiring a product specialist or manager before we respond.

Check any performance or outcome claim against the evidence it requires. The FTC’s advertising guidance↗ (opens in a new tab) explains that factual advertising claims need support, including claims conveyed indirectly. Keep comparisons and promises inside the scope of what your sources actually establish.

Use the broader sales AI workflow guide, sales and marketing workflow guide, prospecting tools guide, copywriting prompts, or research prompts for adjacent tasks; these templates focus on preparing and reviewing your own work.

How do you know a sales prompt is ready to reuse?

Try it with a fictional account and an intentionally incomplete input. A good prompt should surface the gap instead of hiding it. Check whether it invents commitments, treats assumptions as facts, or produces generic wording when evidence is missing. Save the corrected version with a short explanation of when to use it.

Keep the prompt separate from an autonomous AI sales agent. A reusable draft instruction does not authorize outreach or customer-system updates. Decide explicitly who reviews the work and which actions the assistant may take before connecting it to any sending workflow.

Record those approved boundaries in fenbs Decisions and rules so connected assistants can read them before reusing the prompt.

How to do this with fenbs

1. Create a free account and add a project for your reusable sales instructions.
2. Save one tested preparation or follow-up prompt in Skills and prompts, with fictional inputs and review criteria.
3. Record approved messaging boundaries in Decisions and rules, excluding customer records and private deal details.
4. Connect a supported assistant and ask it to read the context before preparing a draft.
5. Save a Lesson learned when you correct a recurring mistake, and save progress with the next improvement.

fenbs is free for one person, with no credit card. It fits reusable work instructions; it is not the right fit for maintaining your pipeline or replacing the customer system your team already approves. Start with one fictional follow-up and a strict commitment check.

Find more patterns in the AI prompts collection. Your first useful library entry can be a single prompt that reliably admits when a fact or commitment is missing.