How fast does a message actually arrive?
Every service in this category says “instantly”. The word means nothing on its own, and the honest answer is that most of the delay is not ours to control. Here is what the wait is actually made of, and how to measure it for yourself in about a minute.
What the delay consists of
From pressing “send” on the other site to seeing the message here, four separate things happen — and they are not equally slow.
| Stage | Who controls it | Typical share of the wait |
|---|---|---|
| The sender's queue | the site you signed up to | usually most of it |
| DNS lookup of the domain | your domain's name servers | milliseconds, unless misconfigured |
| SMTP delivery | both mail servers | under a second on a healthy path |
| Appearing on this page | us | about a second, no refresh needed |
Measure it yourself
You do not have to trust a table someone else published. The measurement is simple and takes a minute:
- Open a mailbox here and leave the tab visible.
- From any account you already have, send a message to that address — note the time you press send.
- Watch. The message appears by itself; nothing needs refreshing.
Repeat it two or three times. A single measurement tells you about one moment; three tell you whether that moment was typical. Then do the same on any other service you are comparing, from the same account, within the same few minutes — otherwise you are comparing two different networks at two different times and calling the difference a result.
If it is slow, this is usually why
- The sending service is throttling. Large senders batch confirmation mail. Nothing on the receiving end changes that.
- The domain is misconfigured. A missing or wrong MX record makes delivery fail slowly rather than quickly, because the sender retries. The domain debugger answers this in seconds.
- The sender refused the domain outright. Then nothing is coming at all, and waiting is the wrong thing to be doing — how to tell the difference.
- Greylisting. Some paths deliberately reject the first attempt and accept the retry a few minutes later. It looks exactly like slowness and resolves on its own.
Why we do not publish a league table
This page used to carry one, measured once in 2017 and never re-measured. It is gone, and the reason is worth keeping:
A speed table is true for the hour it was made. Services change hosting, add queues, add captchas; the number stays on the page and keeps being read as current. Worse, it invites the reader to trust a measurement they cannot check against a service they can measure themselves in a minute. The procedure above is more useful than any number we could print, because it produces your number, today, on your connection.