Sending Servers and Public Servers

Updated November 1, 2026 · 3 min read

In this article
  1. How emails reach the provider
  2. If our address is bad, not your email
  3. If everyone rejects

When a provider accepts an email, it looks at two things: which domain it came from and which internet address it was sent from. The domain is yours. The address is ours: it is the address of one of Estismail's public servers. It is worth knowing, because a reputation at a provider is built from both, and we answer for our half.

How emails reach the provider

A campaign is sent by several of our sending servers at once. They do not go to providers themselves: every email is passed through one of the public servers, and that server hands it to the provider. Any sending server can send an email through any public one. We call this a swarm.

A public server is chosen anew for every email. Emails to a provider are split between public servers in turn, and each has its own share: the better a server's rating at this provider, the more emails per minute it can be trusted with. A new public server is warmed up first: only a few emails to each provider go through it, then more. And if at some minute there is no room left on any public server, the email simply waits in the queue for the next minute.

As a result, a campaign's emails are spread over many addresses, and no address gets more emails from a provider than the provider is ready to accept from it.

If our address is bad, not your email

Sometimes a provider rejects emails only through one public server or a few, while the very same emails get through the others. That means the provider did not like our address, and it would be unfair to punish your campaign for it.

Estismail spots this picture from the provider's answers: on one or a few public servers there are many rejections, and on the others almost none. Then those public servers simply stop sending your campaign to this provider, and the campaign itself goes on, without a pause and at the same speed.

Here is when we decide that: "spam" rejections came through 1–3 public servers, each of them has 10% or more, and that is 10 times more than on the others, while through the others emails get through and "spam" rejections there are no more than 1%.

If this happened during the test batch, or fewer than 5 emails got through the other servers, the picture is less clear. Then sending to the provider still goes into a normal pause, and the bad servers no longer take part in the campaign after it.

It happens that the same public server gets "spam" rejections within 6 hours in the campaigns of two different customers. Then it is clear that the problem is the address, not one customer's emails. Such a server stops sending emails to this provider for 6 hours in all campaigns, then for 2 hours gets only a small share of emails and, if all is calm, returns to work.

If everyone rejects

When a provider rejects through all public servers at once, the problem is no longer the address. Most likely it dislikes the email itself or the address list. Then the usual pauses, retry and, if rejections do not stop, a stop come into play: see the article on pauses, retries and stopping.

That is why your half matters even with our servers: SPF, DKIM and DMARC domain signatures and a clean list. How to connect a domain is explained in domains and setup.

Was this article helpful?
Next articleThe speed chart in the campaign report and what to do →