Spec-Driven Lead Gen Pipeline
Zero-guesswork lead generation: scrapes websites & Instagram, applies hard ICP filters, tiers prospects (D/A/B/C/F), enriches & verifies emails, and creates CRM contacts—only if qualified.
The Challenge
A B2B agency was manually building lead lists from Instagram profiles and websites, then enriching them with sketchy data and importing everyone into HubSpot—resulting in a CRM full of junk contacts and low reply rates.
Their problems:
- No consistent ICP filtering—VAs were guessing which leads to add
- Duplicate contacts (same business scraped from multiple sources)
- Bad email data (40-50% bounce rate)
- No tiering system—sales reps wasted time on D-tier leads
- Manual enrichment taking 5-10 minutes per lead
They needed a deterministic, repeatable pipeline that only added high-quality, verified leads to their CRM.
Case Video Coming Soon
Watch the full pipeline in action: sheet input → scraping → tier assignment → enrichment → HubSpot sync
The Solution
AutoFlux built a fully spec-driven lead gen pipeline with hard ICP filters, deterministic tier logic, and multi-step email enrichment + verification:
Pipeline Steps:
1. Input (Google Sheet)
User adds target domains or Instagram handles with notes about vertical, geography, and special criteria.
2. Scraping (Apify + Custom)
Extract company data, Instagram metrics, keywords, tech stack, and contact information from multiple sources.
3. Hard ICP Filtering
Rules defined in spec document reject leads that don't meet minimum criteria before enrichment.
4. Tier Assignment (D/A/B/C/F)
Tier A: 50K+ followers + e-commerce. Tier B: 20-50K followers. Tier C: 10-20K. Tier D: Low engagement. Tier F: Fails ICP.
1. Input (Google Sheet)
User adds target domains or Instagram handles with notes about vertical, geography, and special criteria.
2. Scraping (Apify + Custom)
Extract company data, Instagram metrics, keywords, tech stack, and contact information from multiple sources.
3. Hard ICP Filtering
Rules defined in spec document reject leads that don't meet minimum criteria before enrichment.
4. Tier Assignment (D/A/B/C/F)
Tier A: 50K+ followers + e-commerce. Tier B: 20-50K followers. Tier C: 10-20K. Tier D: Low engagement. Tier F: Fails ICP.
- Input (Google Sheet):
- User adds target domains or Instagram handles to sheet
- Can include notes (vertical, geography, special criteria)
- Scraping (Apify + Custom Scrapers):
- Websites: Extract company name, description, contact page data, keywords, tech stack
- Instagram: Pull follower count, bio, website link, engagement rate, post topics
- Regex email extraction from website contact pages
- Hard ICP Filtering:
- Rules defined in spec document (e.g., "Must have 10K+ Instagram followers OR website with Shopify")
- Rejects leads that don't meet minimum criteria before any enrichment (saves API costs)
- Tier Assignment (D/A/B/C/F Logic):
- Tier A: 50K+ followers + e-commerce site + high engagement
- Tier B: 20-50K followers OR clear purchase intent signals
- Tier C: 10-20K followers, meets ICP but lower priority
- Tier D: Meets minimum ICP, low engagement
- Tier F: Fails ICP → rejected, not enriched
- Email Enrichment (Multi-Step):
- Primary: Snov.io domain search for business emails
- Fallback: Dropcontact enrichment if Snov fails
- Verification: Hunter.io + Snov email verification (catch-all detection, syntax check, MX record)
- Only proceeds if email is "verified" or "risky-accept"
- HubSpot Sync:
- Creates contact in HubSpot only if:
- Tier A/B/C (D-tier goes to separate list)
- Email is verified
- Not a duplicate (checked via domain + name fuzzy match)
- Sets custom properties: tier, follower count, engagement rate, source
- Tags for segmentation (e.g., "Instagram-sourced", "E-commerce", "High-priority")
- Creates contact in HubSpot only if:
Tech Stack:
Results
85%
Email deliverability (vs 50% manual)
100%
ICP compliance (hard filters)
Zero
Duplicate contacts in CRM
10x
Lead throughput (same team size)
"Before AutoFlux, our CRM was a mess—full of bad emails and leads our reps would never close. Now we only get tier A/B/C prospects with verified emails, and our reply rates tripled. The tiering system alone changed how we prioritize outreach."
– Founder, B2B Lead Gen Agency
Key Takeaways
Deterministic Logic > Guesswork
Hard-coded ICP filters and tier rules eliminated subjective decision-making and ensured consistent quality.
Multi-Step Enrichment Works
Using Snov + Dropcontact fallback + Hunter verification resulted in 85% deliverability vs 50% with single-source data.
Spec-Driven = Scalable
Because every rule was documented upfront, the client can now adjust ICP criteria or add new sources without rebuilding the entire pipeline.
👆 Swipe to explore each takeaway
- Deterministic Logic > Guesswork: Hard-coded ICP filters and tier rules eliminated subjective decision-making and ensured consistent quality.
- Multi-Step Enrichment Works: Using Snov + Dropcontact fallback + Hunter verification resulted in 85% deliverability vs 50% with single-source data.
- Spec-Driven = Scalable: Because every rule was documented upfront, the client can now adjust ICP criteria or add new sources without rebuilding the entire pipeline.
Need a clean, spec-driven lead pipeline?
Let's build your custom lead gen system.
Book a free blueprint call to see how we can automate your lead sourcing, enrichment, and CRM sync.