DirectoryWorkflowsCase studiesPodcastsWho's Noah Take the stack assessment
← All case studies
SaaS · Outbound

1,000 emails a day at a 2% reply rate

Clay records → enrich, validate, personalize → Smartlead campaigns

Industry Visitor-attribution SaaS (US)
Buyer Growth / founder
Stack n8n (self-hosted) · Clay · Smartlead · Supabase
Volume Up to 1,000 emails/day
1,000/day emails sent
2% reply rate
100% governance automated

A visitor-attribution SaaS had the pieces - Clay for enrichment, Smartlead for delivery - but nothing reliable connecting them. We built a self-hosted n8n pipeline that runs the whole flow with governance baked in: normalized Clay intake, dedup and suppression, segmented AI personalization, campaign assignment with mailbox caps and inbox rotation, reply classification by webhook, and Supabase logging with daily digests.

Results at a glance

The problem

The organization had the components but no reliable mechanism connecting them: no automated sync for validated records at scale, no data governance (dedup, suppression management), personalization that risked deliverability, and no feedback loop for classifying and routing replies. CSV wrangling slowed every launch and hurt sender reputation.

What was built

01

Intake

Extracts “ready” rows from Clay with field normalization and verification, so only clean records enter the flow.

02

Hygiene

Deduplicates by email and domain; applies global suppressions for customers, partners, competitors, and unsubscribers.

03

Personalization

AI-generated templates segmented by industry, company size, and role, with fallback logic when a signal is missing.

04

Sending

Campaign assignment by persona, daily mailbox caps, warmup protocols, and inbox rotation to protect deliverability.

05

Replies

Webhook-triggered classification (positive / neutral / negative / OOO / bounce) with instant alerts and automated reschedules.

06

Observability

Supabase logging, daily digests, and error alerts so the whole machine is visible and never silently drifts.

How the approach worked

The hard part was never sending an email - it was governance at volume: dedup, suppression, deliverability, and reply handling running reliably without a human in the loop. Build the guardrails first, then let it run.

The results

The pipeline achieved full automation with governance throughout. Reply rate lifted to 2% with persona-specific messaging, and the system processes up to 1,000 emails daily while staying compliant with CAN-SPAM and protecting sender reputation.

Tools & technologies

n8nClaySmartleadSupabaseSPF/DKIM/DMARC

Have a process like this slowing your business down? Tell me your stack and goals and I'll send back a build plan.

Take the stack assessment →