Skip to content
Waseit
← All articles
Local dataContact

One address in twenty is dead

We wrote to 2,122 businesses across sixteen markets, at the address published on their own website. 104 mailboxes no longer existed. In Japan, one in eight.

4.9%

of published addresses no longer exist

September 8, 20266 min read
In this article
  1. What we measured
  2. One in twenty, and the country changes everything
  3. The generic address dies first
  4. And 116 domains receive nothing at all
  5. What this measurement does not prove
  6. What it changes for you

Short answer: published contact addresses die quietly, and nobody notices. We wrote to 2,122 businesses across sixteen markets, each at the address it publishes on its own website. 112 bounced, 104 of them permanently: the mailbox no longer exists. A customer writing to that address gets nothing either, and unlike a mail server, a customer does not send a report.

12.7%

in Japan, against 0% in Poland

116

domains with no mail server left at all

A dead address warns nobody. The customer believes they wrote, the business believes nobody wrote, and both are right.

What we measured

Between 26 July and 7 September 2026 we sent 2,122 messages to 2,122 distinct addresses, one per business, across sixteen markets. Every address comes from the company's own website: we guess none and we buy none. Then we read what the servers sent back, keeping the distinction the protocol itself makes: a permanent refusal (the mailbox does not exist) is not a temporary one (mailbox full, server unavailable).

These are not businesses drawn at random: they are the ones we contacted, overwhelmingly small local outfits - practices, clinics, agencies, shops. We say so before the numbers rather than in a footnote.

One in twenty, and the country changes everything

104 permanent refusals out of 2,122 addresses, or 4.9%. Eight more were temporary and do not count: a full mailbox empties. The spread between countries runs from one to thirteen.

Published addresses that refused the message, by market. Sends from 26 July to 7 September 2026, 2,122 addresses.
MarketAddresses written toBouncesShare
Japan1021312.7%
United Arab Emirates119108.4%
Saudi Arabia119108.4%
Qatar12997.0%
United States675405.9%
Brazil14685.5%
France202115.4%
Singapore10343.9%
United Kingdom12643.2%
Belgium14421.4%
Switzerland11710.9%
Poland10000%

Japan and the Gulf lose between one address in eight and one in twelve; Poland, Switzerland and Belgium lose almost none. We have no measured explanation for that gap. The obvious hypothesis is the age of the page: an address published then forgotten survives as long as the mailbox does, and dies the day the business changes provider without touching the site. We name it as a hypothesis; our measurement does not test it.

The generic address dies first

Of the 112 addresses that bounced, 60 were generic - `info@`, `contact@`, `hello@`, `office@` - or 54%. That is more than their share of what we sent. The reason fits in one sentence: an address in a person's name dies when that person leaves, which gets noticed; a generic address dies when nobody reads it any more, which does not.

And 116 domains receive nothing at all

Before sending anything, you can ask the DNS whether a domain has a mail server. Across 7,407 addresses in our base, 125 can receive nothing, or 1.7%, and 116 of those because the domain has no MX record left. The site answers, the contact page renders, the form looks fine: the mail arrives nowhere. Japan holds 29 of them, Belgium 18, the Emirates 17.

What this measurement does not prove

  • The sample is not random: these are the businesses we contacted, in sixteen markets we chose.
  • A permanent refusal says this server refused this message. Almost all of them name a mailbox that does not exist, but a very strict spam filter can produce the same code.
  • We measure the address PUBLISHED on the site. A business unreachable at that address is often still reachable by phone, by form or on a social network.
  • A missing MX record says the domain receives no mail today. It does not say the business has closed.
  • All our sends leave from a single sending domain. Another sender might see different rates.

What it changes for you

Write to yourself, from outside.

Once a quarter, send a message to the address shown on your site from a mailbox that is not yours, and check it lands.

An address nobody reads is worse than no address.

The customer believes they reached you and waits for an answer that will not come.

The day you change mail provider

, check that the domain shown on the site still has a mail server. It is the quietest failure in this measurement.

Beware the orphan generic.

If `info@` has no named human behind it, give someone's address, or forward it.

Keep a second visible channel

- phone, form - so that one broken channel does not cost the contact.

You can see in a minute, free and without an account, what AI assistants answer when someone looks for your trade in your city.

Frequently asked questions

What share of published contact addresses no longer work?
4.9% in our measurement: out of 2,122 addresses published by the businesses themselves and written to between 26 July and 7 September 2026, 104 permanently refused the message because the mailbox no longer exists. Eight further refusals were temporary and do not count.
Why do generic addresses die more often?
Of 112 addresses that bounced, 60 were generic (info@, contact@, hello@, office@), or 54%. An address in a person's name dies when that person leaves, which gets noticed. A generic address dies when nobody reads it any more, and nothing signals it.
How can I tell whether my address still works?
Send yourself a message from a mailbox outside your company, once a quarter. Also check, after any change of mail provider, that the domain shown on your site still has a mail server: 116 domains in our base have none left while their website answers normally.
Which countries have the worst problem?
Japan (12.7% of addresses written to) and the Gulf (8.4% in the Emirates and Saudi Arabia, 7.0% in Qatar). At the other end, Poland produced no bounce at all across 100 addresses, Switzerland one out of 117. We have no measured explanation for that gap.