A campaign of ten thousand messages looks uniform from your side and very uneven from the receivers' side: a large share of the list sits on three or four providers, each of which sees a sudden spike from a domain it has never heard of.
Shuffle, then pace per domain
Interleave the list so consecutive sends do not target the same domain, then apply a per-domain pace on top. The shuffle alone does not help if the queue drains faster than the provider's tolerance; the pace alone does not help if the top of the queue happens to be the largest domain. Together they turn a burst into a stream.
Warm up the domain, not the list
Reputation attaches to the sending domain and IP, not to the campaign. A brand-new domain sending ten thousand emails in an hour is the exact pattern receivers are trained to distrust. Start with the most engaged segment, tens of messages, then grow daily. The engagement signals from those first recipients are what teach the receiver that your mail is wanted.
The measurement that tells you to stop
Watch the ratio of deferred to accepted responses per domain. A rising 4xx rate on one domain means it is throttling you; continuing at the same pace turns a temporary deferral into a block. Pause that domain, keep the rest moving.
One concrete default
Ten concurrent connections per receiving domain, one message per connection, with a backoff that doubles on 4xx. It is slow enough to be polite and fast enough that a ten-thousand message campaign finishes in a couple of hours.
Volanea applies per-domain pacing inside the send queue and exposes the deferral rate per domain, so the decision to pause is data instead of guesswork.
Top comments (0)