A long list of companies is not much help if nobody can explain why they are on it. AI-assisted lead research becomes useful when each entry includes a specific fit, a recent reason to look closer, and a source you can open yourself.
Define fit in terms of your service
Suppose you manage retail installations and store refreshes. You might look for specialty retailers announcing several new locations in regions your crews can cover. A company’s expansion plan is relevant to your work; a generic description such as fast-growing brand is much less useful.
Write down your service, geographic limits, practical project size, and exclusions. Include existing clients, companies you have already contacted, and businesses that clearly fall outside your capacity. Decide whether the unit is a parent company, brand, or individual location so the list does not count one opportunity several times.
Ask for evidence with dates
Useful signals might include an official store-opening announcement, a facilities role being advertised, or a procurement notice relevant to your service. Ask for the exact page supporting each signal and record both the publication date and the event date when available. A newly indexed article about an old opening is not a new opportunity.
Company websites, official careers pages, and public procurement documents are good starting points. AI can help discover and summarize them, but you should be able to verify the qualification without rerunning the entire research task.
- Give the assistant your criteria and exclusion list.
- Request a small sample of qualified companies with supporting URLs.
- Reject weak examples and refine the criteria before expanding the list.
- Ask for separate fields for verified facts, inferred relevance, and missing information.
Make the output easy to review
Use a spreadsheet with company, website, location, fit reason, signal, source URL, relevant dates, suggested role to contact, and research status. A fit score is only useful if its rules are visible. You could assign one point each for geography, service match, project scale, and a current signal, then treat missing evidence as unknown.
Keep contact research tied to professional information. If a current facilities lead cannot be verified, name the role you need rather than inventing a person or guessing an email address. A job title found on an old biography may no longer be current.
Turn research into a thoughtful next step
Before outreach, open the strongest source for each shortlisted company. Check whether the announced project is still upcoming, already completed, or being delivered through an existing partner. A relevant event suggests a conversation may be worthwhile; it does not prove a company has budget or wants outside help.
For the retail example, your next step could be identifying the vendor qualification process. For another service it might be a short question to the relevant team. Keep outreach separate from the research request, and approve the actual recipient and message before anything is sent.
Review the first batch for usefulness, not just size. If five well-supported entries produce better conversations than fifty broad matches, make the next run more selective.
A prompt to make it your own
Research [number] companies that may need [specific service] in [geography]. Required fit: [criteria]. Exclude [existing clients and previous leads]. Look for [signals] during [time window], prioritizing official company and procurement sources. For each entry, provide the exact supporting URL, publication date, event date if different, and a concise fit explanation. Separate verified facts from inferred opportunity. Identify a relevant professional role and include a named person only when their current role is supported. Do not invent contact details. Deduplicate parent companies and brands using [rule]. Return a reviewable spreadsheet and send no outreach.
A useful lead is a documented reason to investigate, not a prediction that someone is ready to buy.
Use-case inspiration: 9x’s ChatGPT Work video and companion article. Examples, instructions, and prompts written for AI Prompts.
Use the prompts with the tools and permissions available in your AI environment. Some workflows require connected accounts or uploaded files.