B2B Outbound Database Distribution Pipeline

Python pipeline that takes raw CRM exports and produces segmented, channel-ready contact lists for a B2B outbound sales team. Built for a team of 4 BDRs covering multiple territories (LATAM, EMEA, Spain, Italy).

What it does

Takes a unified contact database and distributes it into actionable weekly lists per sales identity, applying:

• Territory-based assignment — each company goes to exactly one identity based on country/region
• Channel routing — contacts are routed to the right outreach channel based on buyer presence, contact count, and language callability
• Contact caps — prevents over-prospecting a single company (max 4 buyers / 6 influencers / 6 referrers per company)
• Deduplication — by domain (company level) and LinkedIn URL (contact level)
• Weekly prioritization — selects the top 15 companies per identity for the current week's email campaign
• 3-month backup reserves — generates a second batch from leftover capped contacts for future re-prospecting cycles

Channel logic

Condition	Channel
Has buyer + language match	Call list this week → email if no close
No buyer + ≥5 contacts + language match	Email campaign now
No buyer + <5 contacts + language match	Cold call list
Language not callable by BDR + ≥5 contacts	Email campaign (all contacts together)
Language not callable by BDR + <5 contacts	Non-prospectable
No contacts at all	Cold call (complement)

Contact cap logic

Applied to companies with extreme contact counts (≥10 buyers or ≥15 influencers/referrers):

TOC	Cap	Priority
BUYER	4	VP/EVP/Chief > Director Sales/Mktg > Manager
INFLUENCER	6	GM/CEO/COO > Director (non-sales) > VP/Head of
REFERRER	6	Sales/Mktg/Revenue/Reservations > others (exclude juniors)
LOW QUALITY	3	No scoring, first N kept

Scripts

Script	Description
distribuir_bbdd.py	Initial distribution of contacts by territory and identity
reorganizar_todo.py	Rebuilds all output files after territory changes
generar_exclusiones_y_reservadas.py	Manages exclusions (replied contacts, active campaigns)
generar_llamadas.py	Generates weekly call lists per BDR with phone enrichment
generar_campana_semana.py	Selects top 15 companies per identity for this week's email campaign
limpiar_extremos.py	Applies 4/6/6 caps to extreme companies in campaign lists
generar_reserva_3meses.py	Generates 3-month backup contacts from capped-out companies
generar_tam_aerolineas.py	Dedicated pipeline for airline vertical (geo-based split)
generar_tam_empresas.py	Company-level TAM summary with TOC breakdown
generar_xlsx_semana.py	Formatted Excel summary for weekly BDR briefing
generar_resumen_ejecutivo.py	Executive Excel dashboard with metrics per BDR and identity

Exclusion logic

Before any contact reaches an output file, three exclusion layers are applied:

Exclusion type	Scope	Source
Companies with existing meetings	Entire company excluded globally	exclusion por reunion pactada/ — CRM export of accounts with a booked meeting
Contacts in active campaigns	Individual contact excluded (company stays)	exclusion en campana/ — one file per identity with current campaign enrollments
Contacts who already replied	Individual contact excluded across all identities	output/CONTACTOS_YA_RESPONDIERON.csv — built from campaign reply data

The distinction matters: a company with one replied contact is not fully excluded — only that specific person is removed. The rest of the company's contacts remain available for prospecting.

Expected input structure

project/
├── bases originales/        # Raw CRM exports (one CSV per region)
├── exclusion por reunion pactada/   # Companies with existing meetings (excluded globally)
├── exclusion en campana/    # Active campaign matches per identity
│   └── match con bbdd/      # Pre-matched: companies in active campaigns already in our database
└── output/                  # Working files (generated by scripts)

All input files are standard CRM exports (HubSpot format) with columns including: First Name, Last Name, Job Title, LinkedIn URL, Company, Domain, Country, TOC (Type of Contact)

Output structure

ENTREGABLE/
├── 1_TAM_POR_IDENTIDAD/      # Full contact base per identity
├── 2_LLAMAR_ESTA_SEMANA/     # Weekly call lists per BDR (with phone numbers)
├── CAMPAÑA MAIL/
│   ├── CAMPAÑA_SEMANA/       # Top 15 companies for email this week (capped)
│   ├── CAMPAÑA_OLAF/         # Remaining pipeline (capped, for future weeks)
│   ├── BACKUP_CAMPAÑA_SEMANA/# 3-month reserve from this week's companies
│   └── BACKUP_OLAF/          # 3-month reserve from OLAF pipeline
├── 5_COLD_CALL_HOTELES/      # Hotels without buyer for cold outreach
├── 6_NO_PROSPECTABLE/        # Contacts that don't fit any channel
└── AEROLINEAS/               # Airline vertical (split by contact geo region)
    ├── DIEGO_AEROLINEAS_GEO.csv    # EMEA + APAC contacts
    └── RODRIGO_AEROLINEAS_GEO.csv  # LATAM + USA contacts

Setup

pip install -r requirements.txt

Scripts use absolute paths configured at the top of each file — update the OUTPUT, ENTREGABLE, and BASES variables to match your local directory structure before running.

Key design decisions

Domain normalization — all company matching is done via normalized domain (strips https://, www., paths and query strings) to handle inconsistent CRM data.

Airline geo split — unlike hotels (split by company country), airline contacts are split by the contact's own geo region (Geo region field from LinkedIn enrichment). A single airline company can appear in both lists with different contacts, enabling parallel outreach by timezone.

Soft deduplication — contacts are deduplicated by LinkedIn URL within and across identities. Companies are deduplicated by domain. No contact appears in more than one identity's hotel base; airline contacts may appear in both Diego-Aero and Rodrigo-Aero only if they are different people.