Acquisition OS
Automated discovery, verification, scoring and outreach — from lead to qualified pipeline without manual follow-ups.
What It Is
Acquisition OS is an operating system for customer acquisition. It solves a specific problem: discovering, verifying, and reaching qualified prospects requires manual research, verification across multiple sources, scoring decisions, and approval workflows—all before a single outreach email is sent.
The system automates discovery and verification. It runs recurring research cycles (every 2 days), cross-references leads against multiple sources to eliminate duplicates, scores prospects based on brand strength and digital weakness, enriches each lead with AI-generated outreach angles, and routes candidates for human review before campaign deployment.
Acquisition OS is for founder-led companies, agencies, and service businesses that acquire customers through direct outreach. It eliminates the 80% of acquisition work that is research and verification, leaving humans to focus on strategy and relationship-building.
Lead Corpus
Acquisition OS maintains a persistent lead database. Every 2 days, it runs a discovery sweep across configured geographies, deduplicates against the existing database using fuzzy matching, and ranks all candidates (new and existing) by lead score.
The current production database contains 92 leads across three tiers:
Generation Pipeline
Every production cycle follows this verified pipeline, running every 2 days at 09:30 UTC:
If any step fails, the entire run is logged as failed and no incomplete report is generated. Failure alerts are sent via Telegram.
Lead Qualification & Scoring
Scoring Model
Brand Strength (0–50):
• Rating × 4 (normalized)
• Log-scaled review count
• Signals: premium location, mall presence, multi-branch
Digital Weakness (0–50):
• No site: +46
• Aggregator-only: ~42
• Placeholder site: +41
• Basic site: ~28
• Good site: ~14
• Excellent site: ~5
Tier Assignments
Tier A
Lead Score ≥ 65
High-intent, strong fit, ready for outreach
Tier B
Lead Score ≥ 40 and < 65
Medium-fit, exploratory outreach
Tier C
Lead Score < 40
Lower priority, research-needed candidates
Score Calculation: lead_score = 0.55 × brand_strength + 0.45 × digital_weakness ± AI adjustment (±10)
Automatic Exclusions: Corporate chains and franchises are automatically excluded to focus on independent operators.
Segmentation & Filtering
Acquisition OS supports segmentation across multiple facets for targeted campaign deployment:
Geography
- • Navi Mumbai
- • Mumbai (Central, South, West)
- • Pune
- • Configurable regional zones
Business Category
- • Cafes & restaurants
- • Retail & boutiques
- • Services & salons
- • Gaming & entertainment
- • Gyms & fitness
Export Format: Leads are exported as CSV and XLSX with all segmentation facets as filterable columns. Campaign managers can slice by tier, geography, category, lead score, or enrichment data.
Campaign Workflow
Acquisition OS automates candidate selection but requires human approval before outreach deployment. The complete workflow is:
Key constraint: Acquisition OS does not automatically send emails. It qualifies and routes candidates. Human review happens before any outreach is deployed.
Architecture
System Components
Web Application
Vercel (Next.js) — manages UI, campaign config, approval queue
Lead Engine
Python service — handles discovery, research, scoring (runs independently)
Data Persistence
SQLite database — maintains lead corpus, run history, tier assignments
AI Integration
GPT-4o-mini via OpenAI API — enrichment only, scores adjusted per urgency
Research APIs
Tavily (primary) + DuckDuckGo (fallback) — web search and business discovery
Data Flow & Integration
Production Mode
LEAD_ENGINE_MODE=remote
Vercel app → HTTP request → Lead Engine → Database
Execution
Task Scheduler runs every 2 days at 09:30 UTC
Outputs
PDF reports, XLSX exports, CSV mirrors, batch metadata
Architecture principle: The Lead Engine runs independently from the web application. Vercel does not make direct API calls to research services or LLMs. All research logic, scoring, and enrichment happens within the isolated Python service.
This separation prevents rate-limiting issues, keeps credentials secure, and allows the system to handle long-running research tasks without web request timeouts.
Key Capabilities
→Custom scrapers for discovery across configured geographies
→Lead scoring and tiering (A/B/C based on real scoring model)
→Verification (email status, website classification)
→AI-driven outreach (personalized angles per prospect)
→Multi-facet segmentation (geography, category, tier, score)
→Human approval gate before any outreach deployment