Quick answer
Woodpecker's cold email statistics page states that the average reply rate "declined from 5.1% in 2024 to 3.43% in 2026," and hyperlinks the 3.43% figure to Instantly's 2026 Cold Email Benchmark Report. I fetched that report directly. It contains the 3.43% average, but no 2024 figure, no year-over-year comparison, and no mention of a decline at all. The citation supports half the sentence it's attached to and is silent on the other half. Treat the "5.1% in 2024" half as unverified until someone points to where it actually comes from.
The claim: reply rates fell from 5.1% to 3.43%
I'm Hlib Storchak. I build outbound systems for B2B founders and sales teams, and I spend a fair amount of that time tracing where the stats clients quote back at me actually come from. Woodpecker's cold email statistics page, updated June 23, 2026, states plainly that "the average platform-wide reply rate has declined from 5.1% in 2024 to 3.43% in 2026." That reads like a clean, citable, two-point trend line. It's also the exact figure I'd already flagged once before, in a separate piece on this blog, as a number that turns up across dozens of 2026 "cold email benchmark" roundups with no dated primary source attached anywhere. This time it showed up with a link attached to it, which is normally how you'd expect a number like this to get resolved. So I followed the link.
What Woodpecker's page actually says
Woodpecker is a real cold email platform with its own sending data, and its statistics page mixes several kinds of numbers: some drawn from Woodpecker's own platform, some attributed to Snov.io, some to Belkins, and this one attributed, via a hyperlink on the "3.43%" portion of the sentence, to Instantly's 2026 Cold Email Benchmark Report. The sentence structure matters here. It doesn't say "our own data shows a decline." It presents the decline as an established fact and points the reader to Instantly for where that fact comes from. That's a normal, reasonable way to cite something, as long as the thing you're pointing to actually contains the claim you're making.
Why I checked this one specifically. A number with no citation at all is easy to flag as unverified. A number with a citation that looks correct is the more dangerous case, because it reads as settled and gets repeated with confidence. That's the exact pattern behind most of the stats I've had to walk back on this blog.
Following the link to Instantly
Instantly's 2026 Cold Email Benchmark Report describes its own dataset as billions of cold email interactions across thousands of active workspaces, covering January 1 through December 18, 2025, with email send and reply events, sequence performance, and timing and engagement patterns aggregated and anonymized. That's a real, large, disclosed sample. It's the same report I've verified directly for other pieces on this blog, and the 3.43% average reply rate is genuinely in there, along with a top-quartile figure of 5.5% and an elite tier the report describes as clearing 10%+. What I could not find anywhere in that report, reading it directly rather than trusting a summary of it, is a 2024 figure of any kind, a year-over-year reply rate comparison of any kind, or the word "decline" applied to reply rates over time.
What Instantly's 2026 report actually shows
To be precise about what is and isn't in the source Woodpecker points to:
| Figure | In Instantly's 2026 report? | Notes |
|---|---|---|
| 3.43% platform-average reply rate | Yes | Stated directly, for the Jan 1 to Dec 18, 2025 window |
| 5.5%+ top-quartile reply rate | Yes | Stated directly as its own tier |
| 10%+ elite-tier reply rate | Yes | Described as the top-performer band |
| A 2024 reply rate figure of any value | No | No 2024 number appears anywhere in the report |
| A stated year-over-year decline | No | The report is a single-period snapshot, not a multi-year trend study |
So the report backs the number the hyperlink lands on and says nothing at all about the number attached to the other end of the same sentence. That's not a rounding difference or a methodology quibble. It's a claim built from one verified figure and one figure the cited source simply doesn't contain.
A working link is not the same as a source
This is the part worth generalizing past this one page. A hyperlink on a statistic reads, visually, as "here's where this came from." Readers, myself included on a first pass, tend to treat a linked number as verified and an unlinked number as a claim to check. That habit breaks down the moment a link is attached to only part of a compound claim. Woodpecker's sentence pairs two numbers, 5.1% and 3.43%, into one trend, and the citation sits on only one of them. Anyone scanning quickly sees a link, assumes the whole sentence is sourced, and moves on. The mistake I see most often when a client hands me a "the data says" stat to build a campaign narrative around is exactly this: they checked that a link existed, not what the link actually supports.
Where "5.1% in 2024" seems to have come from
I don't have a clean answer for where the 5.1% figure originates, and I'd rather say that plainly than invent one. What I can say is that it isn't unique to Woodpecker's page. It shows up attached to the same "decline into 2026" narrative across other 2026 cold email roundup posts, usually with no citation at all, and in at least one case attached to Instantly by name with no working link back to anything. Woodpecker's page is the first version of this claim I've found that at least attempts a citation, which is progress in the sense that it's checkable, and a problem in the sense that the citation doesn't hold up once checked. The most likely explanation is the ordinary one: a number appeared in an early aggregator or roundup post, other posts repeated it because it fit a plausible-sounding "outbound is getting harder" narrative, and at some point one of those posts got a real citation bolted onto it without anyone re-reading the source it now points to.
Claim versus what checks out, side by side
Laid out plainly, the difference between the claim as written and what a direct read of the cited source supports:
| As stated on Woodpecker's page | What checks out | |
|---|---|---|
| 2024 reply rate | 5.1% | Unverified, no dated primary source found |
| 2026 reply rate | 3.43% | Verified directly against Instantly's 2026 report |
| Framing | A measured year-over-year decline | A single 2026 snapshot with no historical comparison in the cited source |
| What you can safely repeat | "Reply rates have declined since 2024" | "Instantly's 2026 report puts the average at 3.43%," full stop |
What Woodpecker's page actually gets right
None of this makes Woodpecker's statistics page useless, and it would be unfair to leave the impression that it is. The 27.7% to 44% open rate range it cites from Snov.io, its own internal framing of a 5 to 10% reply rate as a realistic "good" benchmark, and the personalization comparison it draws, roughly 7 to 9% for basic personalization versus 17 to 18% for advanced personalization, all read as reasonable and are attributed to identifiable sources rather than presented as unsourced facts. The problem is narrow and specific: one sentence, built from two numbers, where only one of the two is backed by the source attached to it. That's worth fixing on their end, and worth knowing about on yours before you build a client deck around "reply rates have fallen by half since 2024."
A five-step checklist before you repeat a benchmark
This is close to the process I actually run before I let a stat into a client-facing deck or a piece of content:
1. Read the cited source directly, not a summary of it. A page that quotes a report is not the same as the report.
2. Check what the link is actually attached to. If a sentence has two numbers and one link, assume only one number is sourced until you've confirmed otherwise.
3. Look for a date range and sample size in the source itself. Instantly's report discloses both. A source that won't tell you either is a source to treat as directional, not definitive.
4. Search for the exact figure in quotes. If a number only ever appears inside roundup posts citing each other, and never on a vendor's own dated page, that's a signal it's drifted rather than been measured.
5. Decide what you can actually repeat. Usually it's the verified half of the claim, not the whole narrative someone built around it.
Why this keeps happening
Cold email benchmark numbers get recycled constantly because a specific, well-formed statistic is genuinely useful content, and because checking the primary source is slower than restating a number someone else already published. I've walked back similarly-specific claims on this blog before: an uncredited multichannel-response percentage, an old outreach study freshened with a new "last modified" date, a subject-line lift percentage that didn't appear on the vendor page it was credited to. The pattern is always the same shape: a number that sounds precise enough to be true gets attached to a source that sounds credible enough not to double-check, and by the fourth or fifth repost, nobody left in the chain has actually read the thing being cited.
Key takeaways
- Woodpecker's cold email statistics page states reply rates "declined from 5.1% in 2024 to 3.43% in 2026" and links the 3.43% figure to Instantly's 2026 report.
- Instantly's 2026 report verifies the 3.43% average, the 5.5%+ top quartile, and the 10%+ elite tier, but contains no 2024 figure and no year-over-year comparison at all.
- A hyperlink attached to one number in a two-number claim does not verify the other number, and scanning readers routinely treat it as if it does.
- The "5.1% in 2024" figure has no traceable dated primary source found across Instantly, Woodpecker's own data, or any other named vendor.
- What's safe to repeat: "Instantly's 2026 benchmark puts the average cold email reply rate at 3.43%." Not safe to repeat as fact: any specific year-over-year decline built around it.
My take
I don't think Woodpecker set out to mislead anyone here. This is what citation drift looks like from the inside: a plausible number, a real link, and nobody re-reading the destination once the sentence felt finished. The reason I keep writing these teardowns isn't to score a point off one vendor's blog post, it's that this exact pattern is what I find myself untangling on a client account roughly once a quarter, someone's target reply rate or campaign narrative traces back to a stat that was never actually measured the way it's being cited. Read the source before you build a deck around it. It takes ten minutes and it's the difference between a defensible number and one a prospect's own research team unwinds in the first meeting.
FAQ
Did cold email reply rates really fall from 5.1% in 2024 to 3.43% in 2026?
The 3.43% figure is verified directly against Instantly's 2026 Cold Email Benchmark Report. The 5.1% figure and the year-over-year decline framing are not verified anywhere in that report, or in any other dated primary source found while checking this.
Where does the 3.43% cold email reply rate figure actually come from?
Instantly's 2026 Cold Email Benchmark Report, described as billions of analyzed cold email interactions across thousands of active workspaces, covering January 1 to December 18, 2025.
Is Woodpecker's cold email statistics page unreliable overall?
No. Most of the figures on it, including its own internal reply-rate benchmarks and its cited open-rate and personalization figures, are attributed to identifiable sources and read as reasonable. This is one specific sentence where the citation only supports half the claim.
How do I check whether a cold email benchmark stat is real before I repeat it?
Read the cited source directly rather than a summary of it, check exactly which number in the sentence the link is attached to, confirm the source discloses a date range and sample size, and search for the exact figure in quotes to see if it only ever appears inside roundup posts citing each other.
What's a defensible cold email reply rate benchmark to use in 2026?
Instantly's 2026 report's own three tiers: 3.43% platform average, 5.5%+ for the top quartile, and 10%+ for the elite tier, all measured inside the same dataset and time period rather than stitched together from different sources.