You can usually find out how many of your messages failed to reach guests. Finding out which ones is harder, and most venues have never looked. Here is why a small failure rate matters more than it sounds, and what to fix first.

You send booking confirmations. Reminders. The occasional note to a regular.

Most of them arrive. A few do not — a typo in an address, a number that changed, a mailbox that no longer exists.

⚠️ You will almost never hear about those. The guest who did not get their confirmation does not ring to say so. They assume the booking is fine, or they assume it is not, and either way the first you know is when something goes wrong at the door.

The short version

A small failure rate is normal; not knowing which ones failed is the problem
Most systems count failures but do not record who they were to
The fix is cheap: capture the failed addresses, then clean them
A rising failure rate can get your whole sending ability suspended
When that happens, confirmations stop too — not just marketing

Why "a few percent" is not small

What it sounds like	What it is
"3% failed"	3 in 100 guests got nothing
"A handful bounced"	A handful of people turned up to a surprise, or did not turn up at all
"It's within tolerance"	Tolerance is about the sender, not the guest

⚠️ The rate is a sending metric. The failure is a guest experience. Those are two different things and only one of them shows up on a dashboard.

One: the two questions, and only one has an answer

How many failed?   - almost always answerable
Which ones failed? - usually not recorded anywhere

⇒ The first is a number that makes you worry. The second is the one you can act on.

Why the second is missing

Most sending systems record counts by default and addresses only if you switch it on. Nobody switches it on, because nobody looks until there is a problem, and by then the failures that caused it are weeks old and gone.

⚠️ We had exactly this. Our own sending setup counted every failure and stored none of the addresses. We could say "31 failed this week" and could not say to whom.

What we had     a count
What we needed  a list
Difference      one setting, never switched on

Two: what actually causes them

In this trade the causes are boringly consistent.

Typed wrong at the door   - the most common by far
Changed and never updated - phone numbers especially
Full or abandoned mailbox - free addresses that stopped being read
Blocked by the provider   - looks like spam, or your sender is unfamiliar

⚠️ The first one is a process problem, not a technology problem. Somebody read an address aloud in a noisy room and somebody else typed it.

The cheap fix for the first one

Read it back. Out loud. Once.

⇒ Thirty seconds at the point of capture removes most of them. No system involved.

Three: the failure that spreads

Here is the part most venues do not know.

If your failure rate climbs past a threshold, the service that sends your mail can suspend your ability to send at all.

Suspended sending stops
  marketing            - you expected this
  booking confirmations - you did not
  password resets       - you really did not
  invoices              - nor this

⚠️ The failures come mostly from stale guest addresses. The consequence lands on your operational mail. They share a sender, so they share a reputation.

⇒ That is why a slowly rising failure rate is worth watching even when nothing feels wrong.

The number to know

Most providers act somewhere around 5%. Under that, nothing happens. Over it, sending can stop with little warning.

Comfortable   under 2%
Watch it      2 to 4%
Act now       above 4%

⇒ If you do not know yours, that is the first thing to find out. It is one question to whoever sends your mail.

Four: the list you cannot clean

Without addresses, a rising rate is untreatable. You know it is going up and you cannot touch it.

With the list     remove or correct 31 addresses, rate falls
Without the list  wait, and hope the old failures age out

⚠️ Waiting does work, because most providers measure over a moving window and old failures drop out of it. But it is not a fix. It is the same rate coming back next month.

What to ask for

"Can we get the actual addresses that failed, not just the count?"

⇒ For most systems this is a setting, not a project. Ask before you need it, because the ones that failed last month are already gone.

Five: hard and soft are different problems

Not all failures mean the same thing.

Hard   the address does not exist - delete or correct it
Soft   temporary - mailbox full, server busy - retry later

⚠️ Treating soft failures as hard loses you guests who were reachable. Treating hard failures as soft keeps the rate high forever.

⇒ If your system reports a reason, use it. If it only reports failure, assume hard after it fails twice in a row.

Six: the number nobody keeps

You do not need a dashboard. You need one figure per month.

Messages sent
Messages failed
⇒ the rate, and the trend over three months

⚠️ The trend matters more than the number. 2% stable is fine. 2% that was 1% last month and 0.5% before that is going somewhere.

Where the trend usually comes from

A new source of guests  - a listing, an agency, a promotion
A bulk import           - somebody loaded an old spreadsheet
A staff change at the door - a new person capturing details

⇒ All three show up as a step change, not a drift. If the rate jumped, ask what changed that month.

Seven: the guest side of it

Everything above is about your sending. The guest experience is simpler and worse.

They booked
They got nothing
They either turn up unsure, or do not turn up at all

⚠️ They do not blame the email. They conclude you are disorganised, or that the booking did not take.

The one mitigation that always works

Two channels for anything that matters

⇒ If the confirmation is important, send it and say it. A message plus a spoken "you're booked for Friday at nine" survives any delivery failure.

⇒ On the reverse direction — messages coming in that nobody sees — see the enquiry nobody knows you missed.

The address that was never yours to begin with

One category deserves separating: addresses given reluctantly.

A guest who did not want to give one     - gives a fake
A guest booking for somebody else        - gives their own
A guest who gave one years ago           - it was real then

⚠️ The first two will fail forever and no amount of cleaning helps, because there was never a working address. Counting them alongside genuine failures makes your rate look like a system problem when it is a capture problem.

Why the setting is usually off

Not neglect — it is off by default in most systems, because storing addresses has a cost and most senders never ask. It stays off until somebody has the specific problem this article is about.

⇒ Which is to say: assume it is off until you have confirmed otherwise, and confirming is one question.

Numbers worth keeping

One   Messages sent per month
Two   Messages failed per month, and the rate
Three The trend over three months
Four  How many failed addresses you have actually corrected

⚠️ Four is usually zero, and it is the only one that changes anything.

The card for the office

────────────────────────
  Messages that did not arrive

  1. What is our failure rate? (ask whoever sends)
  2. Can we get the addresses, not just the count?
  3. Read every address back at the door, out loud
  4. Anything that matters goes on two channels

  A rising rate can stop ALL your mail,
  including confirmations.
────────────────────────

Common objections

"Our system would tell us"

⚠️ It will tell you a count if you look. It will not come and find you, and it will usually not tell you who — the shape described in the automatic message that never sent.

"We hardly send any email"

⇒ Then a handful of failures is a high rate. Small senders hit thresholds faster, because the percentage moves on a few messages.

"Guests would tell us if they didn't get it"

⇒ Some do. Most assume it is fine, or assume it is not, and quietly go elsewhere. Silence is the normal response to a missing message.

"We'll deal with it if it becomes a problem"

⚠️ The problem arrives as "none of our mail is sending," usually on a Friday. By then the failures that caused it are weeks old and unrecoverable.

"We changed the address and it's fine now"

⚠️ Worth confirming the change actually stuck — a screen saying Saved is not proof the record changed, as covered in the edit that said saved. Reopen the record and read it.

"It's only a few addresses"

⇒ It is a few addresses and it is every message you send. The addresses are the cause; the sending is what stops.

What to do this week

One   Ask what your failure rate is. Write it down with the date.
Two   Ask whether the failed addresses are recorded anywhere.
Three If not, ask for that to be switched on. It is usually a setting.
Four  Start reading addresses back at the door.
Five  Put anything that matters on two channels.

Summary

You can usually get how many failed; which ones takes a setting nobody switched on
Without the list, a rising rate is untreatable — you can only wait
A rate above roughly 5% can suspend all your sending, confirmations included
Hard and soft failures need opposite responses
The trend matters more than the number; jumps have causes you can name
Two channels for anything that matters survives any delivery failure

Nobody tells you their confirmation did not arrive. They just show up unsure, or not at all.

The Messages That Never Arrived