Delivered vs Sent vs Opened: Which Denominator Is Yours?

Sent is what your platform handed to the mail system. Delivered is sent minus bounces. Opened is inferred from an image load. Every rate your newsletter reports divides by one of these, and which one it picks changes the number more than anything you write.

The four numbers

There are three in the title and a fourth that gets used by accident, which is the one that causes the most damage.

List size. Every address on your list, including unsubscribed, suppressed, and long-dead ones. It is a database count, not a description of anything that happened during a send. It is never a correct denominator for an engagement rate.

Sent. How many messages your platform handed off for this edition. List size minus anyone unsubscribed, suppressed, or excluded by segment.

Delivered. Sent minus bounces. The receiving mail server accepted the message. This is the correct denominator for open rate and click rate, and it is what a well-behaved dashboard uses.

Opened. How many of those deliveries registered an open, which is inferred from a tracking image being requested. This is the softest number in the set and the only one that is not directly observed by the sender.

Why the same send produces four different rates

Take one edition, and count the clicks once.

List size          5,200
Sent               5,000   (200 unsubscribed or suppressed)
Delivered          4,850   (150 bounced)
Opened             1,940
Unique clickers      145

Now compute a click rate four ways:

145 / 5,200 (list size)  = 2.8%
145 / 5,000 (sent)       = 2.9%
145 / 4,850 (delivered)  = 3.0%   <- the standard
145 / 1,940 (opened)     = 7.5%   <- this is CTOR, a different metric

Same send. Same 145 people. A number between 2.8% and 7.5%, all four arithmetically correct, and only one of them comparable to anything published elsewhere.

That spread is why two operators can argue about whether 5% is good and never resolve it. They are not disagreeing about performance. They are dividing by different things.

Sent: nearly always the wrong choice

Sent looks like the honest denominator because it counts what you actually attempted. The problem is that it includes messages that never arrived anywhere.

A send with meaningful bounce is being penalized twice if you divide by sent: once because those recipients could not engage, and again because they inflate the denominator. Bounces are a list hygiene signal worth watching on their own, and folding them into your engagement rate blurs two separate problems into one number.

Delivered: the right default, with one large caveat

Delivered is the correct denominator, and it means less than most people assume.

It means the receiving mail server accepted the message. It does not mean the message reached an inbox. A message routed to spam is delivered. A message filed into a promotions tab is delivered. No sender can see the difference, on any platform, at any price.

So delivered is the right thing to divide by, and it is not evidence that anyone had the chance to read anything. Those two facts sit together uncomfortably, and any tool that implies otherwise is overreaching.

Worth knowing about what sits behind the number:

• Hard bounces are permanent, usually a dead address. Most platforms suppress these automatically after one.
• Soft bounces are temporary: a full mailbox, a server having a bad day. Platforms differ on how many soft bounces they tolerate before suppressing, which means delivered is not perfectly comparable between providers even when the label matches.
• Suppression removes an address from sent, not from delivered, so a growing suppression list quietly shrinks both.

Opened: inferred, not observed

Opened is the only number in the set the sender does not directly witness.

An open is recorded when a tracking image loads. Apple Mail Privacy Protection and similar tooling request that image on the reader's behalf, whether or not a person ever looked. Those count. So a share of your open count describes software, and the share is not knowable from your dashboard.

This matters most when opened becomes a denominator. Clicks divided by opens is click-to-open rate, and an inflated open count deflates it without any reader behaving differently. A CTOR that declined over a year may be reporting nothing but a shift in which mail clients your audience uses.

Which rate uses which denominator

Rate	Numerator	Denominator
Open rate	Measured opens	Delivered
Click-through rate	Unique clickers	Delivered
Click-to-open rate	Unique clickers	Measured opens
Bounce rate	Bounces	Sent
Unsubscribe rate	Unsubscribes	Delivered

Bounce rate is the one exception worth remembering: it divides by sent, because the whole point is measuring what failed to be delivered.

How to check what yours is doing

Two minutes, once, and you never have to wonder again.

1. Open any recent edition in your platform and write down sent, delivered, opens, and clicks as raw counts, not percentages.
2. Compute the click rate yourself as unique clickers over delivered.
3. Compare it against the percentage the dashboard displays.

If they match, the dashboard uses delivered and you can trust it. If the dashboard's figure is noticeably higher, it is probably dividing by opens. If it is lower, it may be dividing by list size. Either way you now know what you are reading, which is the whole point.

Beehiiv and Kit both report the raw counts, so this check works on either without an export.

Why this is worth caring about

Denominator drift is invisible and it accumulates.

An operator who divides by list size will watch their rates decline steadily for a year while their writing gets better, because the denominator grows with every subscriber who stops engaging but never unsubscribes. They will conclude the newsletter is failing and change the wrong thing. Nothing in the dashboard will contradict them, because the arithmetic was never wrong.

Fixing this costs nothing and makes every other metric page you read usable. It is the cheapest correction available in newsletter analytics.

Where this goes next

Your newsletter platform owns all four numbers and reports them per send. For understanding the denominators, it is entirely sufficient and you do not need anything else.

Distinctful becomes relevant one step later, when the rate is understood and the question turns into what to do about it: which topics preceded the editions that performed, and whether anything moved on the offer side in the same week. It does not recompute your rates and does not replace your provider's reporting.