Sales Research examples

These are prompt templates, not customer stories or measured results. Replace bracketed fields with real facts. Leave unknowns explicit. The examples use Codex’s $sales-research selector; use /sales-research in standalone Claude Code.

A prospect wants recommendations before hiring you

Use $sales-research to draft a reply to this prospect: [redacted message].
I sell [service]. We have not agreed to a paid engagement. They want me to
recommend which changes to make. Help me distinguish a fit conversation from
actual diagnosis, and propose a useful next step without sounding defensive.
Give me the copy first. Do not invent a price or promise outcomes.

Look for a useful, respectful next step with room to decline. The answer should adapt to the request instead of insisting every conversation must be paid.

Prepare a discovery call

Use $sales-research to prepare a discovery call with [buyer type].
Their stated problem is [their words]. My offer is [offer]. We know [facts]
and do not yet know [gaps]. Choose the questions that would change whether
we proceed. Suggest an agenda and the decision to reach at the end.

Look for questions about stakes, fit, authority, constraints, and readiness. The agent should use what you already know instead of repeating a generic script.

Respond to a request for a lower price

Use $sales-research to review this request for a discount: [redacted message].
Here is the agreed scope and the terms I can actually offer: [facts].
Compare holding the price, reducing scope, and changing commitment or timing.
Recommend a response. Keep unsupported assumptions about their budget explicit.

Look for a trade tied to value, scope, risk, or commitment. The agent should avoid making up market rates or deciding that every concession is wrong.

Investigate SaaS conversion

Use $sales-research to investigate why people try [product] but do not buy.
Here is our actual funnel evidence and customer feedback: [attach or paste].
Separate observations from hypotheses. Select relevant sales and positioning
sources, then recommend the next customer question or experiment that could
change our decision. Do not assume missing conversion numbers.

Look for product-specific reasoning. A paid consulting diagnostic is not the default answer to a self-serve product’s conversion problem.

Bound an enterprise AI proof of concept

Use $sales-research to help scope a proof of concept for [AI use case].
The prospect says success means [their criteria or unknown]. Available data,
evaluation access, procurement requirements, and sponsor authority are [facts].
Recommend the smallest proof that supports a buying decision, the evidence
needed to evaluate it, and the next commercial step. Cite relevant primary sources.

Look for success criteria, data readiness, a bounded scope, and a decision owner. The agent should flag missing buying criteria before prescribing a large build.

Review a sales call

Use $sales-research to review this redacted call transcript: [attach transcript].
Our intended outcome was [decision]. Identify what I did well, where I missed
important evidence, and what I should change on the next call. Tie each finding
to a passage or timestamp. Draft a follow-up, but do not send it.

Look for feedback grounded in the actual transcript and a short practice plan. Missing transcript evidence should stay missing.

Assess an interview request

Use $sales-research to assess this interview exercise: [redacted request].
They ask me to explain how I would approach [problem]. The expected time and
deliverables are [facts or unknown]. Is this a reasonable evaluation or a
material unpaid deliverable? Help me clarify the scope without accusing them.

Look for a distinction between explaining your approach and producing a usable client deliverable. Consulting boundaries should not override the interview context.