Most cold email advice is recycled opinion dressed up as insight. "Keep it short." "Personalize the first line." "Follow up five times." Nobody shows you the data behind the claim.
So we ran the numbers ourselves.
Over Q1 2026, our platform processed more than 50,000 cold emails sent across B2B campaigns in SaaS, agency, and professional services verticals. We tracked every variable we could control: subject line length, send time, sequence depth, personalization method, domain age, and more. Then we compared results across campaigns with statistically meaningful send volumes.
What we found challenged several pieces of conventional wisdom. Here is what the data actually shows — and what it means for how you should be running your outreach today.
---
Before the findings, a baseline: the median cold email campaign in our dataset achieved a 1.8% reply rate. The top quartile hit 4.1%. The bottom quartile stayed below 0.6%.
That 6x spread between worst and best is not explained by who you are emailing or what product you are selling. It is almost entirely explained by the decisions made before the email hits the inbox — the infrastructure, the targeting, the sequencing, and the copy architecture.
The implication: most outreach failures are self-inflicted. The fix is systematic, not creative.
---
Everyone argues about subject lines. Short wins. No, curiosity wins. No, specificity wins.
Our data shows a clearer picture: subject lines between 4 and 7 words produced the highest open rates, averaging 38% across campaigns. Below 4 words, open rates dropped to 29%. Above 9 words, they fell to 24%.
The peak performers shared two traits: they implied a specific benefit or problem rather than stating the ask, and they avoided spam triggers (free, guaranteed, limited time, etc.) entirely.
The worst performers in our set used either generic curiosity bait ("Quick question") or over-explained their pitch in the subject line itself ("Looking to help [Company] increase B2B pipeline by 40% this quarter with automated prospecting"). The first generates opens but kills replies once the email body doesn't match the intrigue. The second trains inboxes to filter you before prospects even see the content.
What works: Treat the subject line as a pattern interrupt, not a pitch. Reference something true and specific about their world. "Your pricing page caught my attention" outperforms "Increase your revenue" every time.
---
We split-tested 2,200 email variants where the only variable was the opening line. The results were unambiguous: the first sentence of the email body drives reply rates more than any other single element we measured — including the subject line, the CTA structure, and the number of follow-ups.
The highest-performing opening lines were: - Trigger-based observations (referencing a company announcement, a job posting, a published article, or a hiring pattern) - Specific research signals ("I noticed you're scaling your outbound team — you have three SDR roles open right now") - Contrarian industry observations the prospect would likely agree with
The worst openers: "I hope this email finds you well," "My name is [Name] and I work at [Company]," and anything that started with "I" rather than "you" or the prospect's company name.
This is where automated prospecting earns its keep. Our platform pulls live signals — hiring data, content activity, funding news, product launches — and uses them to generate first lines that read like you did the research yourself, because the research was actually done. The difference between a generic first line and a signal-based one is worth an average of 1.4 percentage points in reply rate in our dataset. At volume, that is not a rounding error.
---
The conventional cold email sequence is 5-7 touches. Our data suggests this is wrong for most B2B contexts.
Across campaigns in our set: - 42% of replies came from the first email - 38% of replies came from the second email (sent on day 4-6) - 15% of replies came from the third email (day 10-14) - Combined 5% came from touches 4 through 7
In other words, you are capturing 95% of your potential responses within three touches. Touches 4 through 7 exist mostly to annoy people who already decided not to respond.
The exception: high-value accounts where you have genuine multi-angle research. In that case, a fourth touch with a different format (a voice note, a short Loom video, or a referral to a piece of content directly relevant to their problem) can produce disproportionate results. But only if touch 4 adds real signal — not "just wanted to bump this to the top of your inbox."
For most teams, the right answer is a tighter 3-touch sequence with genuinely distinct value at each step. We broke down the exact structure in our 3-touch email sequence framework if you want the play-by-play.
---
This is the finding that surprised us most.
We tracked domain health metrics across every sending domain in the dataset. By week six of active sending, domains that started with pristine reputations showed measurable degradation in inbox placement if they were sending more than 80 emails per day without rotating to secondary domains.
The pattern: small drops in engagement rate → gradual increase in promotions-tab placement → periodic spam filtering → hard bounces start rising → domain reputation collapses.
The progression moves fast. By the time you notice replies are down, you are often two weeks into a degraded reputation that takes months to rebuild.
The data reinforces something we have been telling customers for years: your main domain is not a sending domain. You need a portfolio of warmed sending domains, rotating on a schedule, with individual volume caps per domain per day. Teams sending more than 200 emails per day from a single domain are burning their infrastructure without realizing it.
Pair this with proper authentication (SPF, DKIM, DMARC are table stakes in 2026 — if these are not configured, stop here and go fix them) and you have a deliverability foundation that actually holds up at scale. Our cold email deliverability guide covers the infrastructure setup in full detail.
---
This was the starkest finding in the entire dataset, and the one with the most direct strategic implication.
We compared campaigns by ICP match quality — essentially, how well the lead list matched the stated ideal customer profile for each account. We split campaigns into three tiers: loose ICP (broad industry targeting, minimal filters), moderate ICP (industry + company size + title), and tight ICP (industry + company size + title + behavioral signals + tech stack).
The results:
| ICP Precision | Avg. Reply Rate | Meeting Booked Rate | |---------------|-----------------|---------------------| | Loose | 0.9% | 0.2% | | Moderate | 2.1% | 0.6% | | Tight | 4.8% | 1.4% |
Tight ICP targeting produced 5x the reply rate and 7x the meeting rate of loose targeting — at the same volume.
The implication is counterintuitive for teams that have been told "more pipeline = more success." Sending 1,000 emails to a tight ICP list will produce more meetings than sending 5,000 emails to a loose one. Volume is not the lever. Precision is.
This is the case for automated prospecting done right: not blasting your way to meetings, but identifying the 200 companies that are genuinely likely to buy right now, enriching every signal available on them, scoring them against your ICP, and reaching out only when there is a real fit. Our ICP scoring framework explains how we operationalize this at scale.
---
If you pull these five findings together, the picture that emerges is not complicated:
1. Infrastructure first. Get your domains, authentication, and warmup right before you send a single email. Deliverability problems are invisible until they are catastrophic. 2. Precision over volume. Invest in ICP definition, lead enrichment, and signal-based targeting before worrying about send volume. 3. First line is everything. The quality of the first sentence matters more than subject line, CTA, or sequence depth. It deserves proportional effort. 4. Three touches, each distinct. Stop running seven-touch sequences where touches 4-7 are the same "just bumping this" template. 5. Measure reply rate, not open rate. Open rate is a vanity metric in a world with open tracking inflation. Reply rate and meeting rate are the only numbers that matter.
The companies running outreach at 4%+ reply rates are not doing something exotic. They have simply operationalized these five principles consistently, at scale, without relying on a team of SDRs to do it manually.
---
OnyxSend was built to encode these principles into automated cold outreach that runs without manual intervention. Our platform handles lead enrichment and ICP scoring, signal-based first-line generation, domain rotation and deliverability monitoring, sequence management with reply detection, and meeting booking — end to end.
The companies we work with typically replace one to three SDR headcount with a single OnyxSend subscription — and produce more qualified pipeline in the process. If you want to see the math, our SDR replacement guide walks through the full cost comparison.
Ready to see what 4%+ reply rates look like for your ICP? Start your free trial and we'll set up your first campaign in under an hour.