| name | lead-research-assistant |
| description | Build evidence-backed B2B account and prospect research for sales, partnerships, pilots, or customer discovery. Use whenever the user wants an ICP, target-account list, qualified leads, account scoring, buying signals, decision-maker roles, outreach angles, CRM-ready research, or deeper analysis of named companies. |
| category | business |
| license | MIT |
| compatibility | Live web access is required for current company facts, public contacts, consent rules, and source verification. |
Lead Research Assistant
Find organizations with a credible, timely reason to care. Optimize for fit,
evidence, and a respectful next step—not for the largest possible list.
Safety and Data Rules
- Use public, lawful, business-relevant sources and respect site terms, robots
controls, access restrictions, and rate limits.
- Do not obtain or infer sensitive personal traits, private contact details,
passwords, hidden data, or personal email addresses. Do not enrich people
through data breaches or deceptive pretexts.
- Do not invent employee counts, funding, technology, budgets, contacts, or
buying intent.
- Record a source URL and verification date for every material lead claim.
- Prefer role-level targeting until a named person is verified from a current
public source.
- Follow applicable marketing, privacy, telemarketing, and platform rules.
Provide research and compliant strategy; do not promise that outreach is
lawful in every jurisdiction.
- Exclude obvious non-fit accounts and explain why.
Inputs and Defaults
Inspect the repository, homepage, docs, pricing, and existing customers first.
Establish:
- product and measurable outcome;
- target user, buyer, champion, and blocker;
- company type, size, geography, maturity, and technology constraints;
- triggering events or pain signals;
- disqualifiers;
- desired motion: discovery, pilot, self-serve sale, enterprise sale, or
partnership;
- list size and required output format.
When the ICP is vague, produce an initial hypothesis and a small calibration set
before scaling the search.
Workflow
1. Define the ICP and Negative ICP
Create a one-page hypothesis:
- firmographic fit;
- workflow/problem fit;
- technical and compliance fit;
- urgency/trigger fit;
- ability to buy and implement;
- reachable channel;
- explicit exclusions.
Separate must-have, strong signal, and nice-to-have criteria. A narrow
segment with repeatable evidence is more useful than a broad category label.
2. Create a Source Plan
Use sources appropriate to the signal:
- company website, documentation, pricing, trust/security pages;
- official job posts and careers pages;
- filings, procurement notices, public budgets, and registries;
- product changelogs, engineering blogs, repositories, and status pages;
- current news and official announcements;
- conference talks, podcasts, public interviews, and permissioned directories;
- professional profiles for current roles, subject to platform terms.
Search results and aggregators are leads to verify, not final evidence. Prefer
primary sources for company facts.
3. Build Account Evidence Cards
For each account capture:
| Field | Requirement |
|---|
| Company | Legal/common name and official website |
| Fit evidence | Specific match to ICP with source |
| Trigger | Current event or observable need with date |
| Current approach | Known tool/workaround, or unknown |
| Likely user/buyer | Role, not an invented person |
| Value hypothesis | Outcome tied to observed context |
| Risk/disqualifier | Reason it may not fit |
| Confidence | High, medium, or low with explanation |
| Next research | Missing fact that changes priority |
Do not claim a company “needs” the product merely because it belongs to an
industry.
4. Score Transparently
Score 0 to 5 and cite the basis:
| Criterion | Weight |
|---|
| Problem/workflow fit | 3 |
| Trigger strength and recency | 3 |
| Product/technical fit | 2 |
| Economic value | 2 |
| Buying/implementation plausibility | 2 |
| Reachability | 1 |
| Competitive lock-in | -2 |
| Legal, reputation, or data risk | -2 |
Show raw scores, confidence, and missing data. Do not use a precise score to
disguise weak research.
5. Identify the Buying Group
Map likely roles:
- end user;
- operational champion;
- budget owner;
- technical/security reviewer;
- procurement/legal approver;
- executive sponsor where relevant.
Only name individuals when their current role is verified publicly. Include the
source and date, and avoid unnecessary personal details.
6. Design the Contact Strategy
For top accounts, create a concise, evidence-grounded approach:
- why this account now;
- the observed context;
- a cautious value hypothesis;
- one low-friction question or offer;
- the most appropriate channel;
- follow-up cap and opt-out handling.
Personalization should use relevant business facts, not surveillance. Never
fabricate familiarity, referrals, customer logos, or results. Avoid automated
high-volume sequences until the segment and message are validated.
7. Calibrate Before Scaling
Review a small batch for:
- false positives and missing disqualifiers;
- source freshness;
- response quality, not only response rate;
- conversion to qualified conversation;
- objections and segment differences;
- complaints, opt-outs, blocks, or reputation damage.
Update the ICP and scoring model before researching a larger list.
Output Contract
Return:
- product and ICP summary;
- positive and negative ICP criteria;
- source methodology and research timestamp;
- ranked account table;
- evidence card for each high-priority account;
- buying-group role map;
- account-specific outreach hypothesis;
- CRM-ready CSV columns when requested;
- uncertainties, exclusions, and next research steps.
Sources
