The Problem
Most 'AI lead scoring' tools score implicitly — the model 'vibes' the lead — leaving sales teams unable to audit why a lead got 78 vs 42, and locking the score into a vendor's black box that can't be ported.
The Scavio Solution
n8n + GPT/Claude + 12-line weighted rubric in the system prompt + Scavio enrichment for missing form fields. The rubric IS the product; treat it like code.
Before
220-person SaaS, 120 leads/week, two AEs spending 15 hrs/week triaging, median response on hot leads 9 hours.
After
Lead → score within 60 seconds, routed to right channel, reason logged in CRM. Replaces ~15 hrs/week manual triage. Per-lead cost $0.01-0.04.
Who It Is For
Inbound-heavy SaaS teams with 50-500 leads/week; growth engineers tired of opaque ML models; agencies productizing lead scoring.
Key Benefits
- Auditable: every score has a reason rooted in the rubric
- Per-lead cost <$0.05
- No vendor lock-in (rubric ports across LLMs)
- Scavio enrichment fills firmographic gaps the form misses
- Setup takes 1 day, not 2-4 weeks
Python Example
import requests, os
H = {'x-api-key': os.environ['SCAVIO_API_KEY']}
RUBRIC = '''Score 0-100 using ONLY this rubric. Return {"score":int,"reason":"..."}.
Title fit:30 Industry:25 Company size:20 Intent:15 Fit notes:10'''
def enrich(company):
r = requests.post('https://api.scavio.dev/api/v1/search', headers=H,
json={'query': f'site:linkedin.com/company {company}'}).json()
return r.get('organic_results', [])[:3]
def score(lead, llm_call):
e = enrich(lead['company'])
return llm_call(RUBRIC + f'\nLead: {lead}\nEnrichment: {e}')JavaScript Example
// Same shape in TS via fetch + LLM SDK. n8n is the deliverable.Platforms Used
Web search with knowledge graph, PAA, and AI overviews