ReplyLead proprietary campaign analysis // updated August 13, 2026
Two platforms published cold email benchmarks this year from datasets in the hundreds of millions. Their headline numbers disagree by 4.6x. So we opened our own campaign book - every campaign, both denominators, nothing removed - and let the distribution speak.
Scroll: the claims, then the evidence
Scene 01 // what gets published
3.43%
"Tier 3 (Average) 3.43% reply rate" - defined as the "percentage of all replies received (including follow-up responses) divided by total emails sent." Their top quartile threshold is 5.5%, their top 10% is 10.7%.
Dataset described as "billions of cold email interactions across thousands of active workspaces". No median published. No campaign eligibility criteria disclosed. Source
0.74%
"The median sender earns a reply from about 0.74% of contacts - roughly one reply per 135 people contacted. The top 10% convert 2.63% or better." Built from "more than 850 million emails"; "only senders with at least 500 emails sent in the period are included."
A median, per contact, with a stated floor - the most reproducible methodology of the set. Source
5-10%
"Good cold email reply rates range from 5-10% for B2B teams, with top performers hitting 15%+ on focused campaigns." The same guide cites an outreach study averaging 8.5% "across millions of emails."
Note the shift: 5% - the report's own top-quartile threshold - is presented here as the floor of "good". Source
Both flagship reports are current. Both datasets are enormous. They disagree by 4.6x.
Neither is necessarily wrong. One is an average per email sent, the other a median per contact. But if the honest range spans 0.74% to 3.43% before anyone stretches the truth, a single quoted "average" tells you almost nothing - and "good means 5-10%" needs evidence. Here is ours.
Scene 02 // our complete book
Each dot is one real campaign with at least 500 contacted leads - all 80 of them, from 115 campaigns with sends across 7 client programmes. Hover or tab to any dot. Nothing is hidden and nothing was removed for looking bad.
| Campaign | Reply rate | Contacted | Emails sent | Steps |
|---|---|---|---|---|
| Campaign 001 | 0.36% | 1,395 | 1,941 | 2 |
| Campaign 002 | 0.57% | 702 | 1,318 | 2 |
| Campaign 003 | 0.60% | 5,490 | 6,899 | 2 |
| Campaign 004 | 0.62% | 5,629 | 8,502 | 2 |
| Campaign 005 | 0.64% | 7,644 | 14,980 | 2 |
| Campaign 006 | 0.76% | 5,231 | 10,311 | 2 |
| Campaign 007 | 0.78% | 1,927 | 3,818 | 2 |
| Campaign 008 | 0.79% | 2,266 | 4,452 | 2 |
| Campaign 009 | 0.80% | 2,500 | 4,927 | 2 |
| Campaign 010 | 0.82% | 735 | 1,452 | 2 |
| Campaign 011 | 0.82% | 2,326 | 4,519 | 2 |
| Campaign 012 | 0.87% | 6,066 | 11,606 | 2 |
| Campaign 013 | 0.97% | 3,305 | 6,504 | 2 |
| Campaign 014 | 0.98% | 7,618 | 15,097 | 2 |
| Campaign 015 | 0.99% | 3,640 | 4,201 | 2 |
| Campaign 016 | 1.11% | 1,349 | 1,349 | 2 |
| Campaign 017 | 1.15% | 1,392 | 2,760 | 2 |
| Campaign 018 | 1.22% | 1,806 | 3,574 | 2 |
| Campaign 019 | 1.23% | 653 | 1,291 | 2 |
| Campaign 020 | 1.37% | 1,536 | 1,680 | 2 |
| Campaign 021 | 1.40% | 4,504 | 4,509 | 2 |
| Campaign 022 | 1.52% | 528 | 1,042 | 2 |
| Campaign 023 | 1.60% | 2,506 | 4,924 | 2 |
| Campaign 024 | 1.70% | 528 | 1,041 | 2 |
| Campaign 025 | 1.74% | 1,553 | 1,553 | 2 |
| Campaign 026 | 1.82% | 1,707 | 3,330 | 2 |
| Campaign 027 | 1.84% | 2,613 | 5,000 | 2 |
| Campaign 028 | 1.85% | 3,306 | 6,489 | 2 |
| Campaign 029 | 1.88% | 957 | 2,805 | 3 |
| Campaign 030 | 1.97% | 1,626 | 3,212 | 2 |
| Campaign 031 | 2.00% | 1,800 | 1,800 | 2 |
| Campaign 032 | 2.01% | 1,936 | 4,118 | 3 |
| Campaign 033 | 2.02% | 741 | 1,452 | 2 |
| Campaign 034 | 2.04% | 736 | 1,442 | 2 |
| Campaign 035 | 2.06% | 4,133 | 8,164 | 2 |
| Campaign 036 | 2.06% | 972 | 2,723 | 3 |
| Campaign 037 | 2.09% | 1,674 | 2,117 | 2 |
| Campaign 038 | 2.10% | 2,624 | 5,193 | 2 |
| Campaign 039 | 2.10% | 1,903 | 3,748 | 2 |
| Campaign 040 | 2.12% | 2,637 | 5,133 | 2 |
| Campaign 041 | 2.22% | 3,061 | 5,298 | 2 |
| Campaign 042 | 2.27% | 528 | 1,035 | 2 |
| Campaign 043 | 2.33% | 8,083 | 12,386 | 2 |
| Campaign 044 | 2.33% | 1,886 | 1,886 | 2 |
| Campaign 045 | 2.35% | 2,729 | 5,353 | 2 |
| Campaign 046 | 2.36% | 3,132 | 5,847 | 2 |
| Campaign 047 | 2.43% | 2,962 | 5,853 | 2 |
| Campaign 048 | 2.48% | 5,041 | 5,368 | 2 |
| Campaign 049 | 2.52% | 3,294 | 6,315 | 2 |
| Campaign 050 | 2.56% | 4,261 | 8,386 | 2 |
| Campaign 051 | 2.58% | 2,523 | 4,970 | 2 |
| Campaign 052 | 2.62% | 2,744 | 5,401 | 2 |
| Campaign 053 | 2.64% | 3,488 | 6,554 | 2 |
| Campaign 054 | 2.73% | 4,877 | 9,592 | 2 |
| Campaign 055 | 2.73% | 3,370 | 6,559 | 2 |
| Campaign 056 | 2.77% | 723 | 1,418 | 2 |
| Campaign 057 | 2.80% | 1,681 | 3,290 | 2 |
| Campaign 058 | 2.84% | 1,479 | 2,656 | 2 |
| Campaign 059 | 2.84% | 3,379 | 6,632 | 2 |
| Campaign 060 | 2.97% | 1,246 | 2,456 | 2 |
| Campaign 061 | 3.09% | 3,984 | 7,849 | 2 |
| Campaign 062 | 3.22% | 3,790 | 7,459 | 2 |
| Campaign 063 | 3.22% | 6,269 | 12,296 | 2 |
| Campaign 064 | 3.22% | 2,575 | 5,044 | 2 |
| Campaign 065 | 3.23% | 1,518 | 2,986 | 2 |
| Campaign 066 | 3.30% | 4,850 | 9,576 | 2 |
| Campaign 067 | 3.36% | 804 | 1,574 | 2 |
| Campaign 068 | 3.50% | 2,372 | 4,650 | 2 |
| Campaign 069 | 3.55% | 4,203 | 4,875 | 2 |
| Campaign 070 | 3.57% | 9,194 | 15,780 | 2 |
| Campaign 071 | 3.76% | 2,126 | 4,164 | 2 |
| Campaign 072 | 3.79% | 4,013 | 7,705 | 2 |
| Campaign 073 | 3.94% | 4,567 | 8,524 | 2 |
| Campaign 074 | 4.55% | 616 | 616 | 2 |
| Campaign 075 | 4.71% | 2,018 | 3,905 | 2 |
| Campaign 076 | 4.86% | 4,981 | 9,777 | 2 |
| Campaign 077 | 4.93% | 1,054 | 2,084 | 2 |
| Campaign 078 | 6.47% | 2,783 | 2,839 | 2 |
| Campaign 079 | 7.19% | 6,298 | 9,644 | 2 |
| Campaign 080 | 8.05% | 5,924 | 9,075 | 2 |
Campaigns are anonymized and sorted by reply rate. The 35 campaigns with sends but fewer than 500 contacted leads stay in every total above and below; they are excluded only from per-campaign rate statistics, where a rate computed on a few dozen contacts is noise. Full rules in the methodology.
Scene 03 // the signature problem
Watch what happens to the mean - the number most benchmarks quote - when the best campaigns leave the room. Then watch the median.
This is the whole trick. Our top 4 campaigns hold 19.2% of all replies. Any dataset with a long right tail - which is every cold email dataset - produces a mean above its median. Publish the mean and the number is honest, headline-friendly, and systematically above what a typical campaign experiences.
Scene 04 // the definition trap
These three numbers are computed from the same 429,898 emails. Every one of them is a defensible answer to "what is your reply rate?". Tap each to see what it counts.
Under Instantly's definition our entire book runs 1.46% - well under their 3.43% average. Under Smartlead's per-contact family, our median campaign is 2.9x their 0.74% median sender. Same emails, same replies. The definition did all of that.
So when two benchmarks disagree by 4.6x, the first question is not "who is lying?" - it is "what is the denominator, and is that a mean or a median?" Ask those two questions and most of the gap disappears. The part that remains is population: whose campaigns got counted.
Scene 05 // every claim on one line
Every current headline claim and our measured figures, on one axis. These numbers use different definitions - the small print under each marker says which. That is the point: they are routinely quoted side by side as if they were the same measurement.
We could not reproduce the headline "average" or the "good = 5-10%" band from our own complete campaign data, and the published methodologies do not provide enough information to explain the difference. Smartlead's median, the one figure published with a floor and a definition, is the only one our data can even be fairly compared against - and our median lands 2.9x above it, which their all-comers population and our managed-campaign population plausibly explains.
Scene 06 // try it yourself
A working model of the arithmetic. Give a set of ordinary campaigns one or two great outliers and watch the mean drift away from the median - no dishonesty required.
The arithmetic: (10 x 2.2 + 1 x 12) / 11 = 3.09. The median of the same eleven campaigns is 2.20 - the outlier never touches it.
Scene 07 // follow-ups and bounces
What we deliberately cannot tell you: which reply came from which step. Our platform counters attribute replies to campaigns, not to sequence positions, so we publish no per-step reply claim. Instantly reports that 58% of replies arrive on the first email of a sequence - that is their measurement on their data, not something we can verify from ours.
Scene 08 // where do you stand
Compare your reply rate per contacted lead against the full distribution of ours. Percentile bands, not a single pass/fail number.
Enter a rate to see where it lands on the distribution of 80 real campaigns.
Common questions, answered from the data
Across 80 ReplyLead campaigns with at least 500 contacted leads (429,898 emails in total, analysis updated August 13, 2026), the median campaign reply rate was 2.17% per contacted lead, and the middle half of campaigns landed between 1.39% and 3.00%. Treat anything in that band as normal.
It depends on the definition, and the differences are large. On this same dataset: the unweighted mean campaign rate is 2.40%, the pooled rate is 2.59% per contacted lead and 1.46% per email sent. Instantly's current report publishes a 3.43% average per email sent; Smartlead's publishes a 0.74% median per contact. Any "average" quoted without a definition is unusable.
ReplyLead's median campaign replied at 2.17% per contacted lead (80 campaigns, 500+ contacted leads each). Smartlead's median sender earns replies from about 0.74% of contacts across 850M+ emails. The two medians differ because the populations differ: managed agency campaigns versus every account on a self-serve platform.
It is normal. A 2.0% reply rate per contacted lead sits at roughly the 38th percentile of our campaigns - just below the median of 2.17%. It is neither a problem nor a bragging right.
It is exceptional, not merely good. Only 3 of our 80 campaigns reached 5%, and our 90th percentile is 3.80%. When a vendor calls 5-10% "good", they are describing roughly the top 1% of our measured book.
We did not observe one. Our best campaign of 115 reached 8.05%. Rates of 10%+ do appear in published tier tables (Instantly's top 10% threshold is 10.7% by their per-email definition), but treat any 10%+ claim as an outlier, a warm list, or a different definition - not as a plannable target.
Four reasons, in order of size: the denominator (per email sent vs per contacted lead differs by your follow-up count - 43.6% of our sends were follow-ups); mean vs median (our mean sits above our median, and removing just 4 top campaigns moves the mean 9.3% but the median only 3.4%); who gets counted (eligibility floors, test accounts, dead campaigns); and what counts as a reply (with or without out-of-office and automated responses - often undisclosed).
Median, for judging a typical campaign. The mean of our campaigns is 2.40% against a median of 2.17%, and the gap is entirely the long right tail. If you must use a mean - for capacity planning where volume matters - use the pooled rate and say so.
Under 2%. Our pooled bounce rate was 1.31% and the median campaign bounced 1.33%; 64 of 80 eligible campaigns stayed under 2%. Instantly's guidance ("keep bounce rates below 2%") matches what our data shows disciplined lists deliver.
Most of our campaigns run 2 primary steps (88 of 115); a fifth run 3. Smartlead reports a median of 3 steps with diminishing returns per added step. Our data cannot attribute replies to steps, so we make no causal claim about the ideal count - only that 43.6% of everything we send is a follow-up.
On our book: 13 of 80 campaigns (16.2%) beat 3.43% - the number Instantly publishes as the average - and 3 campaign in 80 reached 5%. High reply rates are real, and rare enough that no one should budget on them.
The part that makes the rest credible
ReplyLead proprietary campaign data, pulled and analyzed on August 13, 2026 (UTC). Counters are cumulative platform totals at that timestamp for every campaign in the account: 131 campaigns across 7 client programmes, of which 115 have at least one send. Campaigns in the dataset were created between April 1 and August 12, 2026. Totals: 429,898 emails sent, 242,414 unique leads contacted, 6,268 unique replies, 5,652 bounces, 103 unsubscribes.
16 zero-send campaigns (drafts and not-yet-launched builds) are excluded from send-based metrics and stay in the inventory count. 35 campaigns with sends but fewer than 500 contacted leads are included in every pooled total and excluded only from per-campaign rate statistics (percentiles, mean, median), because a rate computed on a few dozen contacts is noise, not signal. Nothing else was excluded: no campaign was removed for performing badly, and internal-purpose campaigns (for example re-activation of old lists) remain in the dataset.
A reply is a distinct lead who replied at least once (unique replies) - out-of-office and negative replies included, threads not double-counted. A send is one delivered-or-attempted email; messages accepted but filtered downstream count as sent. A first touch is a lead's first email, so first touches equal leads contacted. A follow-up is any send after the first touch: follow-ups = emails sent minus leads contacted. A sequence step is a distinct primary email in the campaign sequence; A/B copy variants of the same step are not steps and were verified individually against the sequence structure of all 115 campaigns with sends.
Median and percentiles: linear interpolation on per-campaign rates over the 80 eligible campaigns. Pooled rates: summed numerators over summed denominators. Mean campaign rate: unweighted average of per-campaign rates. Every statistic on this page was computed twice by independent implementations that had to agree before publication.
This is one agency's managed book. Campaigns here get verified lists, warmed domains and reply handling; a platform-wide population (like Smartlead's 850M-email dataset) includes every unmanaged account, which plausibly explains why our median sits 2.9x above theirs. Reply sentiment is not split - a positive-reply rate would be lower than every reply rate shown. Counters are cumulative, not a fixed time window; active campaigns will keep moving. We found no meaningful correlation between send volume and reply rate (Spearman 0.16) or sequence length and reply rate (-0.07) on the eligible set, so we publish none. Correlation would not have been causation anyway.
Every figure on this page can be recomputed from the per-campaign table in the dataset section: rates, percentiles, means, pooled rates, the outlier scenarios, follow-up share and bounce statistics. If you cite this analysis, state the sample: 115 campaigns, 429,898 emails, median 2.17% per contacted lead, ReplyLead, August 2026.
Third-party pages were captured on August 13, 2026 and quoted verbatim. If a source page changes after that date, its live text governs; tell us and we will update the comparison.