A typical B2B cold email open rate sits around 21% in the US, and strong campaigns reach 40–60%. But before you optimise against those numbers, you need to know that roughly half of all tracked opens in 2026 are not human — which makes open rate the most-measured and least-trustworthy metric in cold email.
Here is what the benchmarks actually mean, what our own subject line tests proved (including a 5x spread between two versions of the same offer), and which number to steer by instead.
What is a good cold email open rate?
| Open rate | Reading |
|---|---|
| Below 15% | Deliverability problem, not a copy problem |
| 15–25% | Typical range; US average sits around 21% |
| 25–40% | Good — targeting and subject line are working |
| 40–60%+ | Strong, and worth checking against the caveat below |
Published B2B averages for 2026 range from 15.1% to 55.71% depending on the source. That 40-point spread is not measurement noise — it is a symptom of the problem in the next section.
One rule holds regardless: an open rate below 15% is almost never a subject line issue. It means your messages are not reaching inboxes. Fix authentication and list quality first — see how to reduce email bounce rate and Gmail bulk sender requirements.
Why open rate is a broken metric
Open tracking works by embedding an invisible image. When the recipient’s client loads it, the open is logged. Since September 2021, Apple’s Mail Privacy Protection pre-loads that image automatically, whether or not anyone reads the message.
- Apple Mail accounts for roughly 47% of all email opens as of January 2026.
- MPP pre-loading drives 49.29% of tracked opens — machines, not people.
- Typical B2B inflation runs 15–20+ percentage points, and up to 35 on lists with heavy Apple usage.
- A campaign reporting 50% may have a real human open rate of 25–30%.
This is why published benchmarks disagree so violently. They are measuring different mixes of humans and pre-fetchers.
The important nuance: inflation applies roughly equally across variants within the same campaign. So absolute open rates are unreliable, but relative comparisons between subject lines tested on the same list at the same time remain valid. A/B testing still works. Reporting the winning number to your board does not.
What our own subject line tests showed
Our cold email team tested a range of approaches against B2B lists. Four results, same offer, very different outcomes:
| Subject line | Characters | Open rate |
|---|---|---|
| Business Insurance | 18 | 67% |
| Boost Your Sales | 16 | 30%+ |
| Boost Your Sales Strategy with ZenBee | 37 | 30%+ |
| Enhance Your Business Development with ZenBee.io | 48 | 13% |
Same product, same team, same period — and a fivefold difference between best and worst.
Why “Business Insurance” won
The campaign targeted B2B insurance companies, and the subject line reads exactly like something a colleague or a client would send. It makes no claim, sells nothing, and carries no branding — it simply names the recipient’s own subject matter.
The body then picked up those same words and led naturally into the offer, asking whether they wanted to increase their business insurance sales. That continuity matters: a native-sounding subject line followed by an obvious sales pitch produces an open and an instant delete.
Why the longest version lost
“Enhance Your Business Development with ZenBee.io” fails on three counts at once. At 48 characters it truncates on mobile. It leads with vendor vocabulary — enhance, business development — that no colleague would ever use. And it names the sender’s brand, which tells the recipient this is a sales email before they have read a word of it.
Notably, “Boost Your Sales” and “Boost Your Sales Strategy with ZenBee” performed about the same. Adding the brand cost little here — but stretching to 48 characters and swapping to formal vocabulary cost more than half the opens.
What the broader data says about subject lines
Industry testing agrees with what we found:
- 3–7 words, 21–40 characters is the performing range; five words tends to peak.
- Subject lines of 21–40 characters open at around 49%; those above 60 characters drop to around 39%.
- 60% of B2B email is opened on mobile, where clients truncate between 30 and 43 characters.
“Business Insurance” is two words and 18 characters. It survives truncation on every device.
What to measure instead
| Metric | Healthy | Why it beats open rate |
|---|---|---|
| Reply rate | 3–5% normal, 5–10% good, 10%+ top | Requires a human decision |
| Bounce rate | Under 2% | Cannot be faked by a pre-fetcher |
| Meetings per 100 contacts | Track your own baseline | The only number that pays |
| Click-to-open rate | Where links are used | Filters out most machine opens |
Two findings that move reply rate far more than any subject line:
- Verified lists reply at roughly twice the rate of unverified ones.
- Campaigns under 50 recipients average 5.8% reply, against 2.1% for large blasts.
How to lift the number that matters
- Verify before every send. The single largest lever, and it doubles reply rate on its own.
- Send smaller batches to tightly defined segments rather than one large blast.
- Write subject lines a colleague could have sent. Under 40 characters, no brand name, no vendor vocabulary.
- Carry the subject line into the first sentence so the open does not feel like a bait-and-switch.
- Personalise on a real signal. Generic cold email replies below 1%; signal-based personalisation reaches well into double digits.
- Keep the body to 50–125 words. Above 200 words, reply rate collapses.
How to run a subject line test properly
- Change one variable. Our 67% versus 13% result is only interpretable because the offer stayed constant.
- Same list, same window. Splitting across weeks measures the week, not the subject line.
- Compare relatively, never absolutely. Machine opens inflate both arms equally, so the gap is real even though the numbers are not.
- Confirm the winner on replies. A subject line can win opens and lose meetings.
Frequently asked questions
What is a good cold email open rate in 2026?
15–25% is typical, 25–40% is good, and above 40% is strong — but treat all of these as inflated by machine opens. Below 15% indicates a deliverability problem rather than a copy problem.
Are open rates still worth tracking?
As a relative signal inside a controlled test, yes. As an absolute measure of performance, no. Never report an open rate without noting that roughly half of tracked opens are automated.
How long should a cold email subject line be?
Three to seven words and under 40 characters. Mobile clients truncate between 30 and 43 characters, and 60% of B2B email is read on mobile.
Should you put your company name in the subject line?
Usually not on first contact. In our tests the branded short version held up, but the branded long version halved. Branding announces a sales email before the recipient has read anything.
Does personalisation in the subject line work?
Inserting a first name does little now — everyone does it and it costs nothing. Referencing something specific and true about the recipient’s situation works considerably better.
Open rate or reply rate — which matters more?
Reply rate, decisively. It requires a human to act, cannot be inflated by pre-fetching, and correlates with meetings. Use open rate only to diagnose whether messages are arriving at all.
The takeaway
Our best subject line was two words long, mentioned no product, made no claim, and named nothing but the recipient’s own business. It outperformed the most polished version we wrote by five times.
That is the whole lesson. Subject lines that sound like marketing get treated like marketing — and open rate will only ever tell you whether you got through the door, never whether it was worth walking through.
Request a demo to see the verified data and signals behind campaigns like these, or read sales prospecting with sales intelligence for the workflow around them. More in the Cold Email & Deliverability hub.