Build a Practical Sales Research AI Prompt Library

A sales research prompt library should help the team produce consistent, reviewable work. A folder of impressive-sounding instructions is less useful than a small set of tested prompts with clear inputs, known limitations and an owner. Start with the repeatable task, not a promise to make the AI act like a world-class salesperson.

For Sales Navigator research, the library should process authorized notes and support decisions. It should not claim to collect everyone's data, reveal private contact details or bypass the boundaries of the source platform.

Build each prompt as a small work specification

Every entry needs a name, purpose, required input, expected output and review checklist. Add a version date and an example of an unacceptable result. That last item helps new team members understand what the prompt must avoid.

• Purpose: the single decision or document the prompt supports.

• Inputs: source-labeled notes, account context and relevant constraints.

• Output: a specific structure and sensible length limit.

• Evidence rules: how to distinguish facts, inferences and unknowns.

• Review: who checks the result and what must be verified.

OpenAI's prompt engineering documentation discusses clear instructions and relevant context. Apply that principle to a narrow sales task rather than combining research, scoring, drafting and sending in one instruction.

Start with three useful library entries

Account brief organizer

“Using only the supplied notes, summarize the account in four sections: verified context, possible relevance, evidence against relevance and unanswered questions. Preserve source IDs. Do not invent technologies, budgets, stakeholders or purchase plans. Mark conflicting information for review.”

Missing-evidence finder

“Compare the research notes with our stated qualification criteria. For each criterion, return supported, contradicted or unknown and cite the relevant note. Suggest one specific research question for each important unknown. Do not fill gaps from general assumptions about the industry.”

Call-question planner

“Propose five open questions for a first conversation using the verified notes. Each question should test one assumption. Do not imply that the recipient has confirmed a problem, budget or deadline. Include one question that could reveal poor fit.”

If your team needs research access, explore Leads Automation’s independent Sales Navigator offer at US$15/month on WhatsApp. Confirm the available plan, eligibility and terms before ordering.

Test the prompt with difficult inputs

Create a small test pack containing a complete account note, a sparse note, conflicting dates, an uncertain company match and an irrelevant announcement. A useful prompt should handle all five without inventing missing information. The exact number is a practical starting point, not a quality certification.

For example, a fictional note says that a procurement manager joined a company two years ago, while another says the person recently moved elsewhere. The expected response is to flag a role conflict. Choosing whichever entry supports a sales opportunity is a failure.

Check whether the output preserves uncertainty in the final sentence, not just in a hidden reasoning paragraph. If the summary says “confirmed decision-maker,” a cautious disclaimer elsewhere does not repair the error.

Version the prompt and its evaluation together

When a change improves one case but breaks another, keep the earlier version available. Record what changed and why. Re-run the test pack when the AI model, input source or output format changes. Do not assume a prompt that once worked will always behave the same way.

Keep company-specific instructions out of shared public prompt examples. Use fictional names and sanitized notes when training colleagues or evaluating a new tool.

Document product dependencies separately

The prompt should say what information it needs, not assume every user has the same Sales Navigator subscription. LinkedIn's plan comparison can help identify feature differences, but verify availability for the actual account before relying on native AI or integrations.

Connect the library to a shared lead data dictionary so inputs mean the same thing, and use workflow evaluation to decide which prompts deserve wider use.

For the independent Sales Navigator offer at US$15/month, message Leads Automation on WhatsApp. Check the available plan, eligibility and terms before ordering; a prompt library does not include or guarantee any particular AI feature.