Warmup pool Opt-in · cross-customer · tier-banded · auto-quarantine

Reputation as infrastructure,
not vibes.

Single-account warmup is a closed loop talking to itself. NuMail's warmup pool is an opt-in network of real mailboxes across thousands of customers — tier-banded by sending history, paired by reputation, and policed continuously. Your inbox placement stops being a guess.

Start a 14-day trial →Read the engineering write-up
▸ Why a shared network

Warming alone is warming nothing.

Mailbox providers don't reward volume. They reward conversation — genuine sends that land in the primary inbox, get opened, get replied to, get pulled out of spam. The math only works at network scale.

Single-account warmup

A closed loop with no signal

Classic warmup tools spin up a handful of throwaway addresses and have them email each other. Gmail and Microsoft have seen that pattern for a decade. Identical send graphs, no real engagement, recycled IPs — it reads as synthetic, because it is.

Worse: every account warms in isolation. One mailbox's bad behaviour teaches you nothing about the next, and a single throwaway sender ramping into your domain pool can drag the whole reputation down with it.

NuMail warmup pool

Reputation, measured against the field

Thousands of real, actively-sending business mailboxes opt in. The pool pairs them so warmup traffic looks like what it is: human inboxes corresponding across domains, ESPs, and geographies — never the same two addresses twice in a row.

Because every mailbox is scored against the whole network, we can see a problem forming before a provider does — and isolate it before it touches your campaigns or your pool-mates.

▸ Gold · Silver · Bronze

Mailboxes are banded, not pooled blindly.

Every mailbox carries a reputation score from 0–100, recomputed continuously from delivery, engagement, complaint, and bounce history. That score sorts it into a tier — and tiers only warm with tiers near them.

★ Gold

Established senders

Score 80–100 · 90+ days history

Aged mailboxes with a clean complaint record and steady primary-inbox placement. They anchor the pool — Gold interactions carry the most reputational weight.

  • Pairs with Gold · Silver
  • Daily warmup volume up to 40
  • Reply-back rate ~92%
  • Ramp ceiling Full send limit
◆ Silver

Maturing senders

Score 55–79 · 21–90 days

Mailboxes past the fragile first weeks but still building trust. They warm against a mix of Gold and Silver to keep lifting placement without over-exposing fresh accounts.

  • Pairs with Gold · Silver · Bronze
  • Daily warmup volume up to 25
  • Reply-back rate ~85%
  • Ramp ceiling Stepped +15%/wk
● Bronze

New & recovering

Score 0–54 · < 21 days or flagged

Brand-new connections and mailboxes recovering from a complaint spike. They start slow, pair mostly with Silver, and have to earn their way up by sending clean.

  • Pairs with Silver · Bronze
  • Daily warmup volume up to 10
  • Reply-back rate ~78%
  • Ramp ceiling Conservative
▸ Reputation pairing

The pool decides who talks to whom.

Every interval, the pairing engine matches your mailboxes with pool-mates whose tier, ESP, and geography make the conversation look natural. A Gmail mailbox in the US gets paired across Microsoft and Google, across regions, against senders at a compatible tier.

No pair repeats inside a rolling window. Warmup mail is opened, marked important, replied to, and dragged out of spam if it lands there — the exact engagement signals providers weight, generated by mailboxes that are themselves real senders.

The result is a send graph with no detectable fingerprint, because there isn't one to find.

PAIRING ENGINE · CYCLE 4,213live
sarah@mercer-group.com
Gold · Gmail · US-East
tomas@brightpath.io
Gold · M365 · EU-West
dev@bluebirdlog.com
Silver · Gmail · US-West
amir@foxtrotcap.com
Gold · M365 · UK
new@helix.bio
Bronze · Gmail · US
ops@saltstone.co
Silver · SMTP · CA
This cycle · across 11,400 opted-in mailboxes38,900 pairings
▸ What runs underneath

Five mechanisms, one outcome.

Inbox placement you can rely on — because the system is watching every mailbox, all the time, and acting on what it sees.

Cross-customer pool

An opt-in network of real, actively-sending mailboxes across thousands of customers — Gmail API, Microsoft Graph, and SMTP. Opt out anytime; your mailboxes simply warm against a smaller internal set.

Gold / Silver / Bronze tiers

Mailboxes are banded by a live 0–100 reputation score and only warm with adjacent tiers — so a fresh Bronze account can't lean on Gold trust it hasn't earned, and Gold senders aren't dragged down.

Reputation pairing

The pairing engine matches by tier, ESP, and geography, with no pair repeating inside a rolling window. Warmup traffic is opened, replied to, and rescued from spam — the signals providers actually reward.

Auto-quarantine triggers

A spam-complaint or bounce spike past a per-mailbox threshold pulls that mailbox out of the pool and your campaigns within minutes — before a provider escalates. It re-enters only after it sends clean again.

Continuous reputation scoring

Delivery, open, reply, complaint, bounce, and spam-placement data feed a score that recomputes continuously — not nightly. Tier moves, ramp ceilings, and pairing eligibility all follow the score in real time.

Per-mailbox transparency

Every mailbox shows its current tier, score trend, today's warmup volume, and spam-placement rate in the dashboard — and over the API at /v1/mailboxes/:id.

▸ Auto-quarantine

One bad day shouldn't cost a domain.

Reputation damage compounds. The pool's job is to catch a spike in minutes and contain it to the one mailbox — never the pool, never your other campaigns.

Contained, then recovered.

Quarantine isn't a punishment — it's a circuit breaker. The moment a mailbox crosses a complaint or bounce threshold, it stops sending campaign mail and leaves the warmup pool. It keeps a quiet, Bronze-tier recovery rhythm until its score climbs back.

You get a mailbox.error webhook and a dashboard alert the instant it trips — no surprise blacklist a week later.

  • Spam-complaint trigger > 0.3% rolling
  • Hard-bounce trigger > 4% / 1h
  • Quarantine latency < 5 min
  • Blast radius 1 mailbox
  • Re-entry Auto, on clean score
09:14:02

Complaint rate crosses threshold

Rolling spam-complaint rate on dev@bluebirdlog.com hits 0.34% — past the 0.30% guardrail.

09:14:48

Pulled from active campaigns

In-flight sends pause; queued sends reroute to healthy mailboxes in the same campaign pool. No leads stall.

09:15:10

Removed from warmup pool

Pairing engine drops the mailbox from this cycle. Tier reset toward Bronze; pool-mates unaffected.

09:15:11

Webhook + alert fired

mailbox.error delivered with reason quarantine.complaint_spike; owner notified in-app.

+ 6 days

Clean recovery → re-admitted

Score recovers above 55 on a quiet send pattern. Mailbox rejoins the pool at Silver and resumes warmup.

▸ Continuous scoring

The score never stops moving.

Reputation isn't a badge you earn once. It's a live number, and every send either compounds it or chips at it. We surface it so you can act on it.

6
signals feed the score — delivery, opens, replies, complaints, bounces, spam placement
~2m
median recompute interval — not a nightly batch job
0–100
every mailbox carries a live reputation score, visible in-app and via API
11.4k
opted-in mailboxes in the pool today, across Gmail, M365 & SMTP
Figures are illustrative of a live pool and move continuously. Thresholds are tuned per ESP.
▸ Deep dive

How we built a warmup pool that providers can't fingerprint

The engineering write-up: the pairing graph, why we tier on a continuous score instead of static age, the quarantine guardrails, and the privacy model that keeps cross-customer warmup safe. No marketing — the actual design.

Read the post →

Stop guessing at
inbox placement.

Connect a mailbox, opt into the pool, and let reputation become something you measure — not something you hope for. Every plan includes the warmup pool.