Best Prompts for Sales

Account research is easy to start and hard to stop. A salesperson can open a company website, annual report, job postings, leadership pages, and recent news, then still struggle to answer the practical question: what should I know before the next conversation?

The problem is not a lack of information. It is the distance between public facts and a useful sales hypothesis. A good account brief should reduce that distance without pretending to know the buyer’s priorities.

Why ordinary prompting fails

A request such as “research this account and find sales opportunities” invites speculation. It can produce invented initiatives, stale facts, generic pain points, and confident statements about people the model has never heard from. It may also collect so much background that the next action disappears.

A better prompt requires dated sources, distinguishes fact from inference, and connects each hypothesis to a question the salesperson can ask. It should help prepare a conversation, not manufacture certainty about the account.

Reusable prompt

Copy
ROLE
You are a careful B2B account research analyst. Prepare a concise brief for a salesperson before a first or next meeting. Prioritize relevance and verifiability over volume.

REQUIRED INPUTS
1. Account name, website, and relevant geography or business unit
2. Our product, the problems it addresses, and where it is not a fit
3. Meeting type, known attendees, and stage of the relationship
4. Research sources with URL, title, publisher, and publication date
5. CRM notes or prior correspondence, clearly labeled as internal
6. Research cutoff date

STEPS
1. Confirm the account identity and flag namesakes or ambiguous entities.
2. Extract verified facts about business model, stated priorities, leadership changes, relevant hiring, major initiatives, and material recent events.
3. For every fact, record the source and date.
4. Identify up to five signals that may relate to the problems our product addresses.
5. Turn each signal into a hypothesis, not a claim. Explain the reasoning and provide one discovery question that could confirm or reject it.
6. Summarize known stakeholders and their publicly stated responsibilities. Do not infer personal motives.
7. Review internal notes for commitments, objections, open questions, and relationship history.
8. Recommend the next conversation objective and a short meeting plan.

OUTPUT FORMAT
A. Executive snapshot, maximum 120 words
B. Verified facts table: fact | relevance | source | date
C. Signals and hypotheses table: signal | hypothesis | confidence | discovery question
D. Stakeholder context: person | verified role | known relevance | unknowns
E. Relationship history and open commitments
F. Suggested meeting objective and five-question agenda
G. Risks, contradictions, and missing research
H. Source list

EVIDENCE AND UNCERTAINTY RULES
- Do not use facts that lack a supplied source.
- Distinguish “Verified fact,” “Internal note,” and “Hypothesis.”
- Never invent revenue, technology use, budgets, priorities, or personal views.
- Treat undated pages as potentially stale and label them.
- Use exact dates rather than “recently.”
- If sources disagree, include both and explain the conflict.
- Confidence reflects evidence quality, not sales potential.

What to provide

Give the model a bounded source pack rather than asking it to roam without limits. Useful inputs include the account’s official pages, filings or reports where relevant, credible announcements, selected job postings, CRM notes, and correspondence your team is allowed to use. Record URLs and dates. Add your own product boundaries so the model does not force a fit.

State the meeting stage. Research for a first introduction differs from preparation for a technical follow-up or renewal discussion. Include known attendees and open commitments, but remove sensitive personal data that is not needed for the business purpose.

How to review the output

Check account identity first. A correct-looking brief about the wrong subsidiary is still wrong. Open the sources behind the three most important facts and confirm that the wording, date, and business unit match. Look closely at causal language. A hiring pattern may suggest an initiative, but it does not prove a budget or an executive priority.

Next, review the discovery questions. They should be neutral enough to disprove the hypothesis. Replace questions that merely steer the buyer toward your preferred answer. Confirm that the meeting objective fits the relationship stage and that previous commitments are represented accurately.

Where it fails

Public information can be incomplete, outdated, or written for investors rather than buyers. Private companies and small business units may leave few useful traces. The prompt also cannot tell whether an internal note remains accurate unless someone updates it.

Research can create false confidence when a signal is only loosely related to your product. It can also become intrusive if the source pack includes unnecessary personal information. Keep the scope business-relevant, comply with your company’s policies, and let the conversation validate the hypotheses.

Practical takeaway

A useful account brief is not a dossier. It is a short set of verified facts, testable hypotheses, and better questions. The evidence rules matter more than the amount of material collected because they prevent preparation from turning into fiction.

Try this prompt in Agentic Workers before one real meeting, then remove any statement you could not defend by opening its cited source.

<!-- X derivative: A useful account research prompt produces verified facts, testable hypotheses, and neutral discovery questions, not a confident story about a buyer you have not met. -->