So what is a good open rate?#
A useful one is a number that moves in the direction you expect when you change something you can see. For a small list of people who asked to join, 30 to 45 percent is ordinary. For a list that has been mailed for years and grown by a form on a website, 15 to 25 percent is ordinary. Somebody quoting you a single industry figure is quoting their own customers.
| Where the number came from | What it can tell you | What it cannot |
|---|---|---|
| A sending tool's benchmark page | How their customers performed in their period | Whether your list is anything like theirs |
| A blog post with a table of industries | The shape of an argument | Nothing verifiable, if it names no source |
| Your own last twelve sends | What your readers do about your mail | Only what you were already doing |
What is the average open rate by industry?#
Every published figure comes from a vendor measuring its own customers, and the vendors do not agree with each other. Here are four with their sources attached, so the spread on one activity is visible rather than hidden.
| Who published it | What they publish | What it measures |
|---|---|---|
| Mailchimp, on its own benchmarks page | Around 34 percent for email marketing campaigns | The campaigns its customers send, which leans toward lists that joined deliberately |
| Klaviyo, on its benchmarks page | About 31 percent for campaigns and 32 for automated flows, with its top tenth near 45 | Ecommerce stores mailing people who recently bought, an audience that expects the mail |
| Salesforce, on its benchmarks page | It calls a good click-through rate something between 2 and 5 percent | Its own customer base, and a different metric entirely |
| Independent studies of open rate | Roughly 21 to 27 percent across all industries, varying by source | Whatever that sample held, measured after Apple changed what an open means |
Take the shape of the argument from those and leave the numbers alone. The only figure that describes your list is your own, read against your own previous sends, and the four pages above cannot see your audience.
How is an open rate calculated?#
The tool puts a picture roughly the size of a full stop inside the HTML. When the mail app downloads that picture, the tool records an open and divides it by an address count. Two sends with the same readers can print different percentages from two tools, because the denominators are different and neither is wrong.
| The denominator | What sits in it | How it moves the number |
|---|---|---|
| Sent | Every address you pushed at, including ones that bounced | Pulls the number down |
| Delivered | Only the mailboxes that took the message | The honest one, and the least used |
| Unique opens | Each address counted once however often it was read | Measures reach |
| Total opens | Each download counted again | Measures re-reading, not audience |
Why do two tools print different numbers for the same send?#
- The picture is blocked. Plenty of mail apps refuse remote images until the reader asks for them, so the open never registers however carefully the message got read.
- The picture is prefetched. Apple Mail fetches remote content through its own servers before the reader decides anything, so an open registers on mail nobody looked at.
- One tool counts a re-read and another does not, which makes total opens and unique opens two different numbers from one audience.
- The denominators differ. A tool dividing by delivered will always print a higher figure than a tool dividing by sent, and neither is lying.
- Security scanners and robots fetch the picture too. Some tools filter that traffic out and some leave it in, and the careful ones print a lower number on purpose.
Two of those five pull in opposite directions inside the same report, which is the real reason the metric is unreliable. Blocking pushes the figure down for the readers who care about images, and prefetching pushes it up for the readers who never open the app.
Why did open rates jump and never come back?#
Apple shipped Mail Privacy Protection in the autumn of 2021. It downloads and caches remote message content through Apple's own servers ahead of the reader doing anything, and it hides the reader's IP address. Since the picture is fetched whether or not the message is read, a large share of Apple Mail recipients now register as opening mail they never looked at.
The percentage you see after that is a mix of real readers and an automatic fetch you cannot separate. Anyone still comparing their 2026 open rate to their 2019 open rate is comparing two different measurements.
What should I look at instead?#
- Bounces per send, because that is the list telling you which addresses have stopped existing.
- Unsubscribes per send, which is a reader making a decision rather than a pixel guessing.
- Complaints, because that is the one signal a receiving mailbox punishes.
- The thing you wanted, whether that is a booking, a reply, a shop visit, or someone telling a friend.
Clicks sit between opens and the outcome, and they survive privacy prefetching better than an open does, since a click needs a person to press something. If your report has no clicks, the honest conclusion is that the message asked for nothing.
Is open rate useless then?#
No, it is a trend rather than a score. Subject lines and send times still show up there, and your own numbers compared against your own previous sends will tell you whether a change helped. What it cannot do is tell a stranger what your list is worth, which is what a benchmark table pretends to offer.
Does a low open rate mean my email is going to spam?#
Not on its own, because a message sitting in the promotions tab has an open rate and a message in the trash has one too, and both read low for entirely different reasons. Placement shows up elsewhere first: bounces from addresses that no longer exist, unsubscribes arriving in clusters, and complaints you can see in a report.
If the same message reaches the primary tab at your friend's address and still prints a low open rate on your list, that is an interest problem rather than a filter problem. The two have opposite fixes, and guessing wrong costs a quarter either way. The filter side is at why is my email going to spam.
What is a good click rate, and does it matter more?#
It matters more for nearly every purpose, because a click needs a person to press something. Prefetching cannot invent one and a blocked mailbox cannot produce an accidental one. The vendor range worth knowing is the Salesforce page that calls a good click-through rate something between 2 and 5 percent, and that is its own customers rather than yours.
For a small business the useful version is not a rate but a count: how many people booked, replied or came into the shop, out of how many you mailed. That number cannot be inflated by a picture, and it is the only one anybody outside marketing cares about.
How do I run a fair test on my own list?#
Hold everything except one variable: same day of the week, same audience, same cadence, and two subject lines. A tool that splits the list and reports the two halves separately does this for you; a tool that makes you remember what you changed last month does not.
Judge it on the number of readers, not the percentage. An open rate difference of four points across 90 recipients is a handful of people, and it is noise unless you see it repeat.
How big does my list need to be before an open rate means anything?#
One reader out of 300 is a third of a percentage point. One reader out of 30 is more than three points. Below roughly 100 recipients, an open rate is a story about a handful of people rather than a measurement, and the honest way to report it is the count.
| Addresses you mailed | What one extra reader does to the percentage | How to read it |
|---|---|---|
| 30 | About 3.3 points | Say "12 of 30 opened" and never quote a percentage |
| 100 | About 1 point | Readable, but a five point swing is still five people |
| 500 | About 0.2 points | A trend starts to mean something once it repeats across three sends |
| 2,000 | About 0.05 points | Now you can test one variable at a time and trust the answer |
That is not a reason to wait until you are large. It is a reason to stop quoting a percentage off a send of forty, and to watch unsubscribes and replies instead, since those are decisions you can hear at that size.
How do I fix a list where nobody opens any more?#
- Send one plain issue with no header image and one clear thing in it, to everybody, and treat that number as your new baseline.
- Split the quiet half off and mail them once a month rather than four times, so the tired addresses stop dragging the report.
- Change one thing per send: the subject, the day, or the offer. Never two at once, because then the result belongs to neither.
- After three sends, keep the pattern that moved the reader counts and drop the one that did not.
If the plain issue also fails, the diagnosis is placement rather than interest, and that is a separate page: how often should you send a newsletter covers what to do with a list that has gone quiet.
Do the big mailbox providers publish a threshold?#
Gmail, Yahoo and Microsoft publish sender guidance about authentication, complaints and engagement, and none of them publish the figure at which a domain starts landing in spam or getting blocked. Anyone quoting you an exact complaint percentage as a rule is quoting another blog, and that includes the posts that link here.
Campaign Kicker reports opens, unsubscribes and bounces for each campaign, counts an open once per address so the number means reach, and will run two subject lines against each other when you want a fair test.
See your own numbers