decision-maker-finder

Decision Maker Finder

Overview

Input: a list of companies (domains preferred, names accepted). Output: a contact file with the number of people per account the user sets, each tagged with a buying role, a verified work email, an email status, and a mobile number only when the user asked for one for a planned call block. When the user's past deals pulled in procurement, security or finance, it also produces a committee file: the procurement, security and finance names the AE will meet later, mapped from the Lead Database at no wallet cost and never sequenced. The report shows coverage per account and per role, which tool found each contact, and what it cost.

The mistake it prevents: emailing whoever has the biggest title. At a 2,000-person company, the VP Operations buys a scheduling tool and the operations manager uses it. The skill picks roles from what the product is and the deal size, then searches for those roles at each domain, cheapest source first.

When to Use

• "Here are 300 target accounts, find who I should email."
• "Get the CFO and Controller at these domains."
• "Who runs IT at these hospitals?"
• "Map the buying committee for our 40 enterprise accounts."
• After tam-builder tiered the account list, or a CRM export produced it.

Use a different skill when:

Prerequisites

Authoritative Foundations

Details in references/framework-notes.md.

Workflow

1. Clean the account list

Normalize every domain to its registrable form, drop duplicates and platform hosts, and remove suppressed accounts (customers, open deals). Accounts with only a name need a domain: query_lead_database with companyName and limit: 5, accept the domain only when name, state and size agree with the source row; mark the rest unresolved and continue without them.

2. Size the committee and choose roles

Pick the product category in references/buying-role-map.md and set the slots from two user inputs:

Adjust by size with the map's size table: at the smallest companies slot 1 is the founder or owner and slot 2 is often the same person; at the largest, slot 1 moves to the SVP or business-unit head with a region in the title.

3. Lead Database pass (included in plan)

Batch domains in companyDomain (halve the batch on a 503), one call per slot, with the slot's seniority, title and notTitle. Count with limit: 1, then page with after until next_after is null. Rank people per slot (title match, seniority, email present, HQ country) and keep the best as primary and the next as backup. Map-only roles use the same pass and cost nothing from the wallet.

4. Website Finder pass (pay per contact found)

scrape_website_finder takes many domains but one jobTitle per request. Group empty-slot domains by the title needed and send one run per title; misses get one more run with the fallback title. Leave autoValidateEmails and autoFindMobiles off so every email goes through one validation run and mobiles are bought only for the slots the user asked for.

5. Email Finder pass (pay per email found)

For people with a name and domain but no email, send find_emails with contacts: [{first_name, last_name, domain}] (or full_name). Found emails are validated at lookup per the tool description; record them with email_status = valid and found_by = email_finder.

6. Validate the rest

Every Lead Database and Website Finder email goes through validate_emails in one run after dedupe and suppression. Map email_result into the standard's buckets (valid_catch_all becomes catch_all). Only valid rows reach the main file; invalid names go back to Email Finder once.

7. Mobiles (only if asked)

Run the decision table in references/mobile-and-phone-handling.md: a named rep and a dated call block exist, and one do-not-contact list is shared across tools. Send valid work emails or profile URLs for the agreed slots to find_mobiles. Store results in mobile as E.164, never in phone.

8. Assemble and report

One row per person with account_domain, slot, buying_role, found_by, email_status, plus the person schema. Cap at the agreed contacts per account; the committee file holds map-only names. Fill the report from templates/output-template.md and run scripts/check-output.py.

9. After launch: role audit

Before launch, the user sets the number of replies or connected calls to review and the wrong-person share that triggers a title change. At that count, sort answers into wrong person, referred, not now, not interested and meeting. Count wrong-person and referral answers per slot title; "not now" is timing and never counts against the title. A slot over the user's threshold means the role map picked the wrong title for this product: promote the referred-to title into the slot and re-run step 3 for accounts not yet contacted. Change the slot title or the copy, never both in the same round (Berman: test one variable at a time), so the next review shows which change worked.

ScraperCity Implementation

Lead Database pass

{"companyDomain": "northbayfoods.example, pinecrestmfg.example, harborlinelogistics.example",
 "seniority": ["vp", "c_suite"],
 "title": "VP Operations, COO, Chief Operating Officer, VP Supply Chain, Director of Operations, Plant Manager",
 "notTitle": "assistant, intern, former, coordinator",
 "limit": 100, "after": "0"}

Run a second call for slot 2 with "seniority": ["director"] and the champion titles. The tool documents vp, director, c_suite for seniority and sales, engineering for department; test other values with a limit: 1 call.

Committee map when the user's past deals pulled in procurement and security (same batch):

{"companyDomain": "northbayfoods.example, pinecrestmfg.example",
 "title": "Procurement, Purchasing, Sourcing, Information Security, IT Security, Controller",
 "notTitle": "assistant, intern, former, clerk, analyst i",
 "limit": 100, "after": "0"}

Website Finder pass

{"domains": ["pinecrestmfg.example", "ridgefieldpackaging.example"],
 "jobTitle": "VP Operations",
 "autoValidateEmails": false,
 "autoFindMobiles": false}

Email Finder pass

{"contacts": [
  {"first_name": "Lena", "last_name": "Okafor", "domain": "northbayfoods.example"},
  {"full_name": "Tomas Ruiz", "domain": "harborlinelogistics.example"}
]}

Use company_name only when no domain exists; the tool's schema marks domain as preferred.

Mobile Finder pass

{"inputs": ["[email protected]", "https://www.linkedin.com/in/tomas-ruiz-ops"]}

Apollo for large accounts

Build the people search on Apollo.io (Filters 1.0 URL) for the large accounts the database pass missed. First try it free: query_lead_database with url set to that search and limit: 1; read search_translation and pagination.total. Only if that falls short:

{"url": "https://app.apollo.io/#/people?organizationDomains[]=bigco.example&personSeniorities[]=vp", "count": 500, "fileName": "dmf-large-accounts"}

Apollo runs usually take 15 to 30 minutes; register a webhook at app.scrapercity.com/dashboard/webhooks or poll every few minutes.

Worked cost estimate (ceiling)

300 accounts, 2 contact slots each (the user's input), 600 slots, plus slot 3 mapped because evaluators joined the user's past deals. Prices are per result found, so the table is the ceiling if every lookup hits.

Run a test batch of each paid step, at a size the user picks, to turn the ceiling into an expected cost (hit rates are unknown until measured). Show this table, run check_wallet, and get approval before any paid step.

Output Format

1. decision-makers-[list]-[date].csv: one row per contact, valid emails only, with account_domain, slot, buying_role, found_by, email_status, email_validated_at, mobile (if requested), plus the person schema from the quality standard.
2. decision-makers-[list]-[date]-committee.csv (when map-only roles exist): map-only names with buying_role, never loaded into a sequencer.
3. decision-makers-[list]-[date]-gaps.csv: accounts with an empty slot, the titles tried, and the tools tried.
4. A report from templates/output-template.md: account list funnel, role plan, waterfall results by step, coverage by role, validation, mobiles, cost, sample rows, gaps and next steps.

What the reader receives is the deliverable only: no internal file paths, reference section numbers, tool ids or skill names (name the product, such as Lead Database, Email Validator or Website Finder, and describe the next action instead of naming a skill); no sales-framework acronyms (keep the content: who signs, who champions); each figure stated once; every outside number with its source and date on first mention; no value left as "you set" when a sourced default or a dated step can carry it; no table of checks that were not run (one "Not checked:" line instead); no repeated identical rows where one template plus a list will do. Next steps are a table with Step, Owner, Date and Done when.

Quality Check

• Role plan written before any search, tied to the product category, the contacts per account and the roles from the user's past deals.
• Every contact row has a buying role, an email status and a found_by tool.
• Main file is 100% email_status = valid; map-only names are in the committee file, not the main file.
• No account exceeds the agreed contacts per account.
• Waterfall table shows input and found counts per step, and found never exceeds input.
• Coverage by role adds up to the accounts searched.
• Mobile numbers only appear if requested, in E.164, with the call block and the do-not-contact check.
• Cost table uses catalog prices and states the total.
• scripts/check-output.py report.md prints PASS.

Common Pitfalls

Execution Artifacts

Related Skills

• account-researcher for a brief on the top accounts before outreach.
• cold-email-list to take this file to a sequencer.
• cold-calling for scripts and call blocks once mobiles are in hand.
• mobile-appender to add mobiles to an existing contact file.
• icp-list-builder when there is no account list yet.
• tam-builder to produce and tier the account list.
• crm-enricher to write contacts and the committee file back to the CRM.
• multi-channel-outreach to sequence email, phone and LinkedIn across the roles.