Deliverability

Metric

The share of sent email that reaches the inbox instead of bouncing or landing in spam, used to check that everything before "did they read it" is working.

Why it exists

For years, "nobody opens our email" was treated as a subject-line problem. Teams rewrote the subject and sent again, moved the send time to right after people get to work, and put the recipient's first name in the opening line. The number refused to move, because the repairs were aimed at the wrong place.

A single message passes several gates between the send button and the inbox. Is the address still in use? Has the sending server been behaving itself lately? Has the receiving mail provider already filed this sender under advertising? Fail any one of those and the message goes to spam or comes straight back. Yet the only numbers sending tools showed were messages sent and messages opened. The entire middle was missing, so there was no way to tell an unread message from one that never arrived.

That gap produced a familiar scene: someone running email marketing spending months on subject lines, testing two versions against a 4 percent open rate, while 30,000 of the 120,000 messages sent were sitting in spam folders from the start. Deliverability is the name given to that missing middle so it shows up on the dashboard.

How to calculate it

Subtract bounced messages from sent messages, then divide by sent messages. Send 10,000 and get 250 back and you're at 9,750 over 10,000, or 97.5 percent.

The calculation has three steps.

  1. Count what actually went out. Addresses that unsubscribed or bounced before are filtered by the sending tool, so the denominator is messages dispatched, not the size of the list.
  2. Subtract the bounces. There are two kinds. A hard bounce means the address doesn't exist and never will, and it should be removed from the list. A soft bounce means a full mailbox or a receiving server that was briefly down, and a retry a few days later usually gets through.
  3. Divide what's left by the dispatched count and read it as a percentage.

That formula has a hole in it. Spam folders don't return anything, so a message filtered into spam counts as delivered. Practitioners therefore split the idea in two. The number above is closer to a send success rate, and whether the message reached the inbox has to be measured separately.

Inbox placement is measured with seed accounts. You create about twenty test addresses across the major mail providers, include them in the real campaign, and then look at each one by hand to see how many landed in the inbox and how many went to spam or a promotions tab. Seventeen out of twenty is 85 percent. It's a sample, so it carries error, but it catches losses that a bounce rate alone can't show. Most teams run the seed check on the first send of any new campaign rather than on every send.

What a good number looks like

Above 98 percent for the bounce-based number and above 95 percent for inbox placement is the normal range. A bounce rate over 2 percent says the list has gone stale, and over 5 percent the receiving providers start treating the sending server itself as suspect.

Three causes account for most collapses.

  1. An old list. Fifty thousand addresses collected at an event two years ago, sent to untouched, will produce hard bounces around 8 percent as people change jobs and close accounts.
  2. Spam complaints. One report per thousand messages — 0.1 percent — is the line most providers work with. Cross it and the share of messages filtered into spam climbs noticeably for everything sent from that address afterward.
  3. Missing authentication. There are DNS records that declare in advance which servers may send mail for your domain. Without them, the receiving side has no way to separate you from someone borrowing your name, so it assumes the worst.

Trend beats absolute value. Holding at 98 percent for six months and then dropping to 91 percent in a single week means something changed that week: a purchased list was bolted on, a server used to 5,000 messages a day was handed 80,000, or the sending domain was swapped. A brand-new sending domain needs two or three weeks of gradually increasing volume before it has any history to be judged on. Start at a few hundred a day, roughly double it each day the bounce and complaint rates stay clean, and only then point the full list at it.

How it sounds at work

"Check deliverability before you look at open rate. We're bouncing at 6 percent."

Fix the list first; the subject line isn't the problem.

"New domain, so we're still warming up. Nothing over 2,000 a day this week."

Sending a lot at once from a domain with no track record gets it flagged, so volume is going up slowly.

"Three of our twenty seed accounts went to the promotions tab."

Nothing bounced, but the message isn't showing up where people actually look.

At Pagedeck, Ravi once sent a product update to every dormant customer after six months of silence. All 42,000 addresses went out untouched, and the campaign came back with a 4.8 percent bounce rate and 0.3 percent spam complaints. The damage didn't stop at that campaign. The regular newsletter the following week opened at half its usual rate, and the seed accounts showed half of it sitting in spam. One send had narrowed the path every later message would travel.

Recovery took a month. The list was cut to 14,000 by keeping only people who had opened something in the last six months, and daily volume was capped at 3,000 and raised slowly from there. That's also when click-to-open rate and unsubscribe rate started getting read alongside it, since the cost of repairing this number runs far above the cost of protecting it.

What people get wrong

The most common mistake is reading 99 percent as proof the mail is being seen. The sending tool prints that figure in large type, which makes the assumption easy. But it only excludes what came back, and spam folders send nothing back — every message could be filtered and the dashboard would still read 99 percent. A team that misses this spends months varying subject lines and send times while the list quietly degrades. High deliverability paired with an unusually low open rate is a signal to measure inbox placement, not to rewrite the subject.

Trusting the sending tool to handle it has the same shape. The tool helps with authentication records and drops bounced addresses automatically, but what the receiving side evaluates is your domain's reputation, not the tool's. Another company on the same platform can send impeccably clean mail and it won't help you if your complaint rate is high.

Easily confused with

TermWhat it countsHow it differs
Open rateShare of arrived mail that gets openedThe next gate down. If deliverability is broken, open rate is calculated over an already filtered audience and can't be interpreted

Where to go next

  • CRM marketing

    see the activity this number belongs to

  • Open rate

    the next number, covering whether delivered mail actually got read

Related terms

Last updated 2026-09-16