What is a good cold email reply rate?
Measured the strictest way, as replies divided by every email sent, a typical cold email campaign in 2026 gets about 0.5% to 1%. Belkins measured 0.45% across 7.5 million emails in 2025; Smartlead's median sender gets 0.74% of the people they contact; QuickMail's median campaign gets 0.48%. Above about 3% is top-quartile work on any denominator, and 10% is the top few percent.
The figures you see quoted most, 3.43% (Instantly), 3.7% (Saleshandy) and 4.5% (Hunter), are real, but each divides by something different and each is an average, which a few very good senders pull upward. Before comparing your number with any of them, check what it was divided by. The rest of this page does that for every benchmark that ranks.
What counts as a reply?
A reply rate is replies divided by something, and both halves vary. The top line can include or exclude auto-replies, out-of-office notices, "unsubscribe" and "wrong person". The bottom line can be emails sent, emails delivered, people contacted, or people who opened. Those four denominators can differ by a factor of ten on the same campaign.
A Hacker News commenter put the numerator problem in one line in 2022: "stop emailing me or unsubscribe. wrong email address. these are things that could make it look like a good reply rate". Only two publishers below say they exclude auto-replies, and only one reports interested replies separately.
Every published benchmark, side by side
| Publisher (what it sells) | Reply rate | Divided by | Auto-replies excluded? | Sample | Period |
|---|---|---|---|---|---|
| Belkins (lead-gen agency) | 0.45% | Emails sent | Yes | 7,530,489 emails, 34,393 replies | Jan to Dec 2025 |
| Belkins, previous study | 5.8% | Recipients who responded (per its 2024 methodology) | Yes | 16.5 million emails, 93 domains | Jan to Dec 2024 |
| Belkins, 2023 study | 7% (open rate 36%) | Unique recipients contacted | Not stated | 11 million+ emails, 93 domains | 2021 to 2023 |
| Instantly (sending tool) | 3.43% average; top performers above 10% | Total emails sent, including follow-ups | Not stated | "billions" of interactions, no count | Jan to Dec 2025 |
| Saleshandy (sending tool) | 3.7% | Emails delivered | Not stated | 53.1 million emails, 60,000 sequences | Jan to Jun 2026 |
| Hunter (finder and sequences) | 4.5%; sales use case 3% | Recipients (45 replies per 1,000) | Not stated | 31 million emails | 2025 |
| Smartlead (sending tool) | 0.74% median; top 5% 3.85% | Contacts | Not stated | 850 million+ emails | Not stated |
| QuickMail (sending tool) | 0.48% median; top 25% 5.98% | Not stated (per campaign) | Not stated | Campaigns with 1,000+ leads over three months | Not stated |
| Snov.io (finder and sequences) | 5.1%, credited to Belkins | Not stated | Not stated | Not stated | Not stated |
Two things jump out. First, medians sit far below averages: QuickMail's top-quartile threshold (5.98%) is 12.5 times its median (0.48%), so a few excellent campaigns drag any average up. Second, the only figures computed on every email sent, Belkins 2025 and Instantly, differ by a factor of 7.6, partly because Instantly counts follow-up replies and does not say it removes auto-replies.
Why Belkins went from 5.8% to 0.45%
Belkins publishes the most-cited reply-rate study in this category, updated every year at the same URL. Its figure fell from 7% (2021 to 2023) to 6.8% (2023) to 5.8% (2024) to 0.45% (2025). Its current page explains the drop: "In prior studies, reply rates were calculated against the number of unique recipients who opened an email. This year, we're reporting against total sends." It adds: "A 5% reply rate against openers and a 0.45% reply rate against total sends can describe the same campaign."
The archived copies do not say that. The 2023 methodology reads: "The reply rate was calculated as the number of unique recipients who responded to the outreach email, divided by the number of unique recipients contacted." The 2024 version: "Reply rate reflects the share of unique recipients who responded to a message, excluding auto-replies and bounces." Neither divides by openers.
What changed for certain is the move from recipients to sends. A sequence of three or four emails sends three or four times as many emails as it reaches people, so the same replies divided by sends come out three or four times smaller. At four emails per person, 0.45% of sends is up to 1.8% of people contacted, which is a long way closer to 5.8% than the headline suggests. Belkins does not publish emails per person, so the exact conversion is not possible.
Belkins also stopped measuring opens. Its 2025 methodology says "Open rate tracking was disabled across all campaigns in this study period", because "the rate tracking pixel was hurting deliverability throughout the industry in 2024." Its 2024 study recorded open rates rising to 46% early in the year and falling to 31% to 32% before tracking stopped.
How to convert between the benchmarks
| Conversion | Working | Result |
|---|---|---|
| Belkins 2025, check | 34,393 replies ÷ 7,530,489 emails | 0.457% |
| Belkins 2025 per person, 3 emails each | 0.45% × 3 | Up to 1.35% |
| Belkins 2025 per person, 4 emails each | 0.45% × 4 | Up to 1.8% |
| Smartlead per send, 3-email sequence | 25 replies per 1,000 contacts ÷ 3,000 emails | About 0.83% |
| Hunter reply-to-open | 4.5% ÷ 30% open rate | 15% |
| Saleshandy reply-to-open | 3.7% ÷ 21% open rate | 17.6% |
| QuickMail spread | 5.98% ÷ 0.48% | 12.5× |
| Instantly average against Belkins 2025 | 3.43% ÷ 0.45% | 7.6× |
The practical rule: before comparing, convert your own number to the same denominator as the benchmark. If your sequencer reports replies per person and you compare against Belkins' per-send 0.45%, you will think you are doing three or four times better than you are.
Why open rates can't be trusted any more
Apple's Mail Privacy Protection, launched in 2021, is the reason most of these publishers now ignore opens. Apple's own support page says it "prevents senders from seeing if you've opened the email message they sent you." Apple's announcement said it "stops senders from using invisible pixels to collect information about the user." Litmus, which sells email testing, puts Apple at 62.26% of all email opens, mostly marketing email rather than cold B2B.
The effect cuts both ways. An email marketer on Hacker News described open rates that "would have naturally trended from 20% to 60%" after the change, "because Apple 'opens' the emails itself". Another put the uncertainty plainly: "You can't really say anything authoritatively about open rates if 5-10-30-50% of your recipients are protected". EmailBison notes that tools showing 40% to 55% open rates "typically reflect MPP-inflated figures".
Tracking opens may also cost replies. Snov.io reports reply rates rising "from 1.08% to 2.36%" when open tracking was switched off, and Hunter reports 7.4% against 4.4% for campaigns without tracking. Both are observational: senders who switch tracking off may simply be more careful senders. Still, two publishers found the same direction.
| Publisher | Reply rate, tracking on | Reply rate, tracking off | Difference | Sample |
|---|---|---|---|---|
| Snov.io | 1.08% | 2.36% | 2.2× | Not stated for this figure |
| Hunter | 4.4% | 7.4% | 1.7× | 31 million emails, 2025 |
What is the 30/30/50 rule?
It is a phrase with no agreed meaning. EmailBison defines it as "30% open rate, 30% reply-to-open rate, and 50% positive reply rate". ProfitOutreach says it means "30% of your success comes from targeting, 30% from your offer, and 50% from the quality of your list", which adds up to 110%. At least five other vendor pages give different splits across personalisation, value, subject lines, calls to action and follow-up, and none names who coined it.
Taken as EmailBison defines it, the rule implies 9 replies per 100 emails sent (30% × 30%) and 4.5 positive replies. That is 20 times Belkins' 0.45% and close to Instantly's top performers. It describes an excellent campaign, not a normal one, so do not use it to judge whether yours is broken.
Where the numbers get copied from
Several ranking pages repeat each other's figures under the wrong name. EmailBison presents the finding that campaigns under 50 recipients reply best as "Woodpecker's data"; Woodpecker credits the underlying figure, 5.8% against 2.1% for large campaigns, to "Belkins' analysis of 16.5 million emails". Snov.io credits a 5.1% average to Belkins, a figure that appears in none of the Belkins studies above. And the LinkedIn post ranking second for this query gives a US reply rate of 3.43%, which is Instantly's worldwide average with no country breakdown.
None of this means the numbers are invented. It means that by the time a figure reaches a listicle it has often lost its denominator, its sample and its date, which are the three things you need to use it.
How to measure your own reply rate
- Pick the denominator and write it down: replies ÷ emails sent is the strictest and the easiest to compare with Belkins.
- Remove auto-replies, out-of-office notices and bounces from the replies before dividing.
- Report interested replies separately. "Unsubscribe" is a reply; it is not a result.
- Measure per campaign and take the median, not the average, if you run several.
- Switch off open tracking and stop judging campaigns by opens.
- Count at least a few hundred sends before drawing any conclusion. At 0.5%, 200 sends produce one reply.
That last point matters more than any benchmark. At a 0.45% rate, detecting a real improvement between two versions of an email needs thousands of sends per version; our outreach automation guide has the sample-size table.
What to do with a low reply rate
Below about 0.3% on every email sent, the problem is usually the list or deliverability, not the words. Check that your mail reaches inboxes first (the manual checks are here), then that the people on the list are the right people with working addresses. Smartlead's own guidance says the same: "If you sit below that, the cause is usually the list, deliverability".
Between 0.5% and 1% you are typical. Above 3% on every email sent you are doing better than most published medians, and the next gain is more likely to come from a better list than from more follow-ups.
Which benchmark to use
- Judging your own campaigns: Belkins 2025 (0.45% of sends) or Smartlead's median (0.74% of contacts). Both are strict and recent.
- Setting a stretch target: QuickMail's top quartile (5.98%) or Instantly's top performers (10%+).
- Comparing tools' claims: ask each vendor what it divides by and whether auto-replies are removed. Most pages do not say.
- Budgeting meetings: start from the strict rate. At 0.5% of sends, 1,000 emails produce about five replies.
Practitioners have long quoted the lower end. "0.5–1.0% reply rates are typical", one Hacker News commenter wrote in 2020; another in 2024: "Cold email response rates are already ~1% industry average." That matches the strictest vendor data today.
What this page does not know
So what reply rate should you expect?
About 0.5% to 1% of every email sent, if your list is decent and your mail reaches the inbox. The 3% to 5% figures are real but measured more generously or skewed by top senders, and the 30/30/50 rule describes an exceptional campaign. Pick one denominator, measure against a benchmark that uses the same one, and judge your campaigns by interested replies rather than opens.