Applicant Intake & Re-Engagement at Production Scale
Regional Home-Appliance Distributor · Tier 3 Production System · Delivered
The Situation
A regional home-appliance distributor was losing hiring opportunities every day — not because applicants weren't applying, but because the team couldn't keep up with them.
Applicant data sat unreviewed in spreadsheets. Follow-up depended on whoever had time to send a text. Nobody had a clear picture of who had been contacted, who had replied, or who had already moved on. The pipeline wasn't moving because the manual layer between "applicant applies" and "applicant gets contacted" was the bottleneck.
Every day of delay was a candidate lost to a competitor or to silence.
The Real Problem
This wasn't a hiring volume problem. The applicants were there.
The business was losing candidates between intake and first contact — a gap that no amount of job posting spend could fix. The pipeline needed to move on its own, without depending on manual bandwidth that was already maxed out.
The System Brealle Built
Regional Distributor Applicant Pipeline v5 PROD — a three-chain automation that ran continuously on a hardened production VPS, processing incoming applicant records from raw intake through outreach, reply tracking, and pipeline state management.
- Scope: Production · Tier 3 governance applied · delivered build
- Scale: 82 nodes across three independent scheduled chains plus a built-in error subworkflow
System Architecture
Chain A — Intake & Qualification
Pulled raw applicant rows from the intake sheet, deduped against the master pipeline using generated keys, normalized fields, applied decision logic, and wrote both to the master pipeline and to a human-review queue. Duplicates weren't silently dropped — they were routed through a dedicated logging branch and recorded before the chain merged back to the main path.
Chain B — Outreach (Re-Engagement)
Pulled approved candidates from the review queue, built personalized re-engagement emails, enforced a daily send cap so the system never exceeded Gmail throughput limits, sent through Gmail with 3-minute pacing between messages, captured the thread ID for downstream reply matching, and logged every send.
Chain C — Reply Handling
Ran on a 15-minute schedule. Pulled active threads from the pipeline, fetched new replies from Gmail by thread ID, matched each reply back to the original candidate, decided whether to respond, sent a threaded reply (also under daily cap), updated pipeline stage and review queue, and logged the chain.
Error Subworkflow — Failure Logging & Halt Handling
Built-in error trigger subworkflow. Any failure in any chain routed to a structured error log entry with chain identifier, timestamp, and error context.
Operational Controls
These were the controls that distinguished a production system from a demo:
- Daily send cap enforcement (Chains B and C) — the system read its own LOGS sheet to count sends-so-far-today and skipped if at cap, marking the row as queued for tomorrow rather than dropping it
- Halt-on-error circuit breaker — every Gmail send was wrapped in error classification logic; classified failures halted the chain and wrote a halt-log entry instead of cascading
- 3-minute pacing waits between sends to respect rate limits, reduce burst behavior, and protect deliverability
- Field Preservation Gate — explicit candidate-payload restoration through transforming nodes, addressing the n8n pattern where Code/AI/HTTP nodes silently drop upstream fields
- Deduplication with full audit trail — every duplicate was logged before being merged out
- Audit logging at every chain — RAW intake, pipeline writes, duplicates, sends, halts, replies, and errors all landed in a single LOGS sheet
Results — Production Outcomes
Directly observed metrics from the production batch:
- Volume processed: 10,580 applicants
- Date range covered: 8.5 months of historical batch
- Classification rate: 100%
- Auto-filter rate: 71% (7,544 of 10,580 filtered as out-of-territory before reaching the operator's queue)
- Throughput: full batch processed in a single session
- Reply taxonomy: 7-bucket classifier with automated routing
Governance
The system was built and governed under Brealle's Workflow Governance Standard (WGS) v1 at Tier 3 — Brealle's highest-risk workflow category, used for external communications with low reversibility.
Tier 3 controls applied:
- Field Preservation Gate (required)
- Audit logging required at every external touch
- Manual approval gate (REVIEW_QUEUE) before any outbound send
- Documented rollback procedure
- Halt-on-error behavior with classified failure modes
Production Footprint
| Layer | Implementation |
|---|---|
| Orchestration | n8n on hardened VPS |
| Reverse proxy + TLS | Caddy with managed TLS |
| State | Postgres-backed n8n instance |
| Hardening | Hardened Linux server, key-based access, firewall enforced |
What This Replaced
- Manual checking of the applicant sheet each day
- Memory-based decisions about who to contact next
- Manual text-message follow-up without centralized state tracking
- Untracked duplicate handling
- Scattered visibility into who replied, who needed review, and who needed next-step action
What It Made Possible
- Continuous, unattended applicant capture
- Consistent re-engagement cadence with cap-respecting throughput
- Complete audit trail of every action — no silent drops
- Reply matching without manual thread tracking
- Production-grade failure handling with documented halt behavior
Key Insight
The constraint was never applicant volume.
It was the manual layer between intake and engagement.
Positioning Statement
Most hiring pipelines don't fail because of bad candidates. They fail because the gap between application and first contact is too slow, too manual, and too dependent on someone remembering to follow up.
Brealle builds the system that closes that gap — so the work runs whether or not someone is working the queue that day.
About Brealle
Brealle designs and builds reliable support systems for trade and home-service businesses — taking the repetitive, time-draining work off the owner's plate so the business stops depending on someone being available to keep it moving. The applicant pipeline above is one example of that work delivered at production scale.
Start with The Time Audit — a free 20-minute call about where your time is actually going.
Losing your evenings to work that could run itself?
Start with The Time Audit — a free 20-minute call about where your time is actually going, and what's worth building relief for first.
Book The Time Audit →Brandon Robinson
Founder, Brealle