Your warm-up score says 92. Your campaign is in spam.
By Chris Aylott
The score was not wrong. The problem is what it was measuring against.
What most warm-up tools actually test against
Every warm-up tool measures inbox placement against a seed network — a pool of mailboxes that receive your warm-up emails, open them, move them to inbox, and signal positive engagement back to ISPs. The premise is sound. The execution is where it breaks down.
Most seed networks are built from synthetic addresses: mailboxes created specifically to receive warm-up traffic, owned and operated by the warm-up provider. Some networks contain tens of thousands of these addresses. They are hammered — receiving hundreds or thousands of warm-up emails every single day from thousands of different senders.
"A mailbox that receives 500 warm-up emails a day, opens every one, never reports spam, and has no other email activity has an anomalous engagement profile. It looks like what it is: a warm-up farm."
Gmail and Outlook build machine learning models on traffic patterns. Within weeks of a large synthetic network launching, ISPs have a model for it. Mail going to those addresses gets evaluated differently — or excluded from reputation signals entirely.
Your 92 score is accurate. You placed well against addresses that ISPs no longer treat as representative of real recipients.
The problem scales with network size
The more senders a seed network supports, the worse this gets. A synthetic mailbox shared by 50 senders receiving 50 warm-up emails a day each is receiving 2,500 emails a day. No real business inbox works like that. The engagement signals it generates — all positive, all from warm-up traffic, with no negative signals — are statistically impossible from a genuine inbox.
This is why warm-up scores inflate over time while real inbox placement does not improve proportionally. The score reflects performance against a known synthetic environment. The real world has different rules.
What a seed mailbox needs to actually provide a trust signal
A seed mailbox has to be indistinguishable from a real business inbox — because the ISP model is trained on real business inboxes. That means:
- It receives normal business email alongside warm-up traffic, not only warm-up traffic
- Its daily volume is within the range of a normal active mailbox — not hundreds of warm-up emails per day
- It has genuine, varied engagement patterns across different senders and message types
- It has been active long enough to have a real sending and receiving history
A mailbox capped at receiving a handful of warm-up emails per week, used by a real business that also receives and sends normal email daily, looks exactly like a real inbox to an ISP model. Because it is one.
How InboxPlease builds its network differently
Every seed mailbox in the InboxPlease network is a real customer mailbox — an active business inbox, not a synthetic address.
Real customer mailboxes
Participating inboxes belong to active InboxPlease customers — companies with real email traffic. Invitation-only, with a 50% subscription discount for members.
Send-only permissions
InboxPlease requires only permission to send to the mailbox — not full read/write access. No competitor in the market operates this way.
200-email daily cap
Every seed mailbox is hard-capped at 200 warm-up emails per day, keeping the engagement profile within normal business ranges.
~3 emails per week
Members set their own cadence. Typical is around 3 warm-up emails per week alongside normal business mail — invisible in the noise of a real inbox.
High score, still in spam? Check these first
Before re-running the warm-up, verify each of these.
Authentication
SPF, DKIM, and DMARC aligned and passing on real delivered messages — not just present in DNS.
List quality
Unverified addresses and disengaged contacts generate negative signals no warm-up can offset.
Volume ramp
Jumping from warm-up volumes to campaign volumes in one step registers as a suspicious spike.
Seed network
Ask your provider: are the seed mailboxes real customer inboxes or synthetic addresses? If they won't say, the score is unmeasurable.
A 92 from a genuine network means something. A 92 from a synthetic farm that ISPs have already modelled means you passed a test the real world does not administer.
Test your inbox placement against real mailboxes
InboxPlease's seed network is made entirely of real customer inboxes — capped at 200 emails/day, invite-only, with send-only permissions.
Start free