Project
B2B Partnerships Engine
Founder / Head of Growth & Ops, Dansu
I built an end-to-end outreach workflow that managed the full partnership pipeline — from first contact through inbound triage to active opportunity — eliminating the dropped leads and missed follow-ups that were killing deals.
- I designed pipeline stages, standardised inbound reply reason codes, and built sequence variants by partner type and score tier, with pacing controls and dedupe logic to keep deliverability clean.
- Reply rates of 3–8% and meeting rates of 1–3% on cold outreach, with A-tier partners consistently converting at 2–3x the rate of lower-tier targets.
Overview
Partnerships were a high-leverage growth channel for Dansu, but execution kept failing. Even with good targets, deals stalled because the pipeline lived in inbox threads, mental notes, and inconsistent follow-up. I couldn't answer basic questions like "who do I need to chase today?" or "which sequences are working?" — so I built a CRM and outreach workflow that fixed that.
The problem
The failure wasn't lead generation. It was pipeline hygiene. Leads were going cold between "sent" and "meeting" because:
- No single source of truth on partner status
- Follow-ups missed or inconsistent
- Inbound replies handled manually, without structure
- No stage definitions, so nothing was measurable
I needed the pipeline to be operable weekly without heroic effort.
What I built
I designed the entire workflow around one principle: every lead has a stage, an owner, and a next action.
CRM object model. Each partner record held the org details, segment, score tier (fed from the discovery engine), contact routes, last touched date, next action, follow-up date, sequence version, status, and reason codes. No loose notes — everything structured.
Pipeline stages. Discovered → Enriched → Scored → Queued → Contacted → Replied → Meeting → Active Opportunity. Each stage implied an action, not a feeling.
Outreach sequences. Variants by org type (festival vs retailer vs venue) and score band (A-tier got more personalised effort). Personalisation tokens came from enrichment — location, upcoming events, sponsor cues — so messages felt relevant rather than blasted.
Inbound triage. I standardised reply buckets: interested, not now, wrong contact, send deck, no budget, not relevant. Each reply triggered a stage move, a next action, and notes for context. This gave me repeatability and analytics instead of an inbox.
Follow-up discipline. Pacing controls and domain throttling to protect deliverability. Tier-based persistence — A-tier partners got more follow-up attempts than C-tier. The point was sending consistently, not sending more.
Systems I built
- n8n workflows for outreach sequencing, follow-up scheduling, and reply routing
- Supabase as the single source of truth for all partner records, statuses, and history
- Gmail API integration for send management and reply detection
- Dedupe and domain-throttle logic to protect sender reputation
Impact
- Reply rate: 3–8% on cold outreach
- Meeting rate: 1–3% from cold
- A-tier replies converted at 2–3x the rate of C-tier targets
- Partnership pipeline went from unmeasurable to trackable by stage and conversion rate
- Partnerships became operable — I could run the full cycle weekly without dropping leads or losing context