Most published cold email benchmarks put a good open rate somewhere between 20 and 50 percent. Treat all of them with suspicion. Since Apple Mail Privacy Protection launched in 2021, a meaningful share of recorded opens are Apple fetching images on the recipient’s behalf, with no person involved.
Published by emailcampaign.ai, which runs cold sending infrastructure. Our product can report open rates and this page argues you should not rely on them, which is worth weighing when reading it.
How an open is recorded, and why that broke
There is no such thing as an open event in email. The whole mechanism is a one by one transparent image embedded in the message. When a mail client loads that image, the server hosting it records a request and calls it an open.
That inference held while image loading meant a person had the message on screen. Three things broke it:
| Cause | What happens | Effect |
|---|---|---|
| Apple Mail Privacy Protection | Apple pre-fetches images through a proxy whether or not the person reads the message | Large inflation on any list with Apple Mail users |
| Corporate security appliances | Scanners fetch every image and follow every link before delivery | Inflation concentrated on enterprise recipients |
| Image blocking by default | Outlook and others block remote images until the reader allows them | Deflation: real reads that never register |
| Your own testing | Team members opening seed sends | Small inflation, larger on small campaigns |
Note that two of those push the number up and one pushes it down. The result is not a consistent bias you can correct for. It is noise that varies with the composition of each list.
How to tell whether yours is inflated
Compare open rate to reply rate. Those two normally move together, because reading a message is a precondition for answering it.
A campaign showing 65 percent opens and 0.4 percent replies is not a campaign that was widely read and widely ignored. It is a campaign where most of the recorded opens were machines. A campaign showing 30 percent opens and 6 percent replies is behaving normally.
Segmenting by recipient domain makes this visible: consumer domains and Apple-heavy audiences will show markedly higher apparent opens than the same message sent to a Google Workspace audience.
The benchmarks, for what they are worth
If you need a number to compare against, these are the ranges commonly reported across vendor studies, with the caveat above applying to all of them:
| Metric | Typical published range | Reliability |
|---|---|---|
| Open rate | 20% to 50% | Low. Inflated by proxy fetches |
| Reply rate | 1% to 5% for broad campaigns, 8% to 15% for tightly targeted | High. Requires a human |
| Positive reply rate | 0.5% to 3% | High |
| Meetings booked per 1,000 sends | 5 to 20 for well targeted outbound | High. The number that pays |
| Hard bounce rate | Under 2% | High |
The spread on reply rate is the interesting one. The difference between one percent and twelve percent is almost entirely targeting rather than copy, which is the opposite of where most teams spend their time.
What to measure instead
Reply rate is the primary metric. A person read it and typed something back.
Positive reply rate separates interest from rejection. Both are useful, and a campaign generating many polite declines is a targeting problem rather than a copy problem.
Meetings booked is the one to report upward, because it is the outcome the spend is justified by.
UTM parameters on plain links give you campaign attribution in analytics without rewriting URLs through a redirect. You lose per-recipient click data and keep the part that informs decisions.
The cost of keeping open tracking on
Every tracked message carries a remote image loading from an external host. On an established domain with history, that is a small cost. On a cold outreach domain with no history, the structure of the message is a larger share of the evidence a filter has, and a tracking pixel makes it resemble bulk marketing rather than personal correspondence.
So the trade is a real deliverability cost on every send, in exchange for a number that Apple partly generates. For cold outbound that is a bad trade, and turning open tracking off is one of the few changes that costs nothing and removes a risk.
Questions, answered straight
- What is a good open rate for cold email?
- Commonly quoted ranges run from 20 to 50 percent, but the number is no longer trustworthy. Apple Mail Privacy Protection pre-loads images on behalf of recipients, registering opens that no human performed. If a large share of your list reads mail in Apple Mail, your open rate is partly a measure of Apple's servers.
- Why is my open rate 70 percent?
- Almost certainly inflation rather than performance. Apple proxy fetches, corporate security appliances that follow every link and image in a message before delivery, and your own team opening test sends all register as opens. An unusually high open rate paired with a normal reply rate is the signature.
- Should I still track opens?
- For cold outbound, usually not. Open tracking embeds a remote image in every message, which is a pattern filters have scored for two decades, and it buys you a number you cannot rely on. The cost is real and the data is not.
- What should I measure instead of open rate?
- Reply rate and meetings booked. Both require a human to act, so no proxy can inflate them, and both connect directly to revenue. If you want traffic attribution, put UTM parameters on plain links rather than rewriting them.
- Is a low open rate a deliverability problem?
- It can be, but check placement directly rather than inferring it. A low open rate with a normal reply rate usually means image blocking, not spam foldering. A low open rate and a near zero reply rate together is the pattern worth investigating.
Published 20 September 2026. Updated 20 September 2026. Written by the team that runs the infrastructure; numbers come from the platform's own provisioning and sending, and from the providers' published documentation at the time of writing.