Cold email reply rate benchmarks across 9 industries from 1.1 million sends. What your campaign should actually be hitting in 2026
benchmark_data_nick · 2026-07-14 · 1,620 views
Most reply rate benchmarks you find online are either outdated, made up by a vendor with an incentive to make you feel bad, or averaged across such a wide range of use cases that they mean nothing for your specific ICP. Here is real data from 28 client accounts, 18 months, 1.1 million sends.
Median reply rates by industry, full sequence (not just email one).
SaaS and tech: 3.8 percent. Professional services (law, accounting, consulting): 3.2 percent. Marketing and advertising agencies: 2.8 percent. Financial services: 2.4 percent. HR and talent: 2.2 percent. Healthcare: 2.1 percent. Manufacturing and logistics: 1.9 percent. Real estate: 1.6 percent. Government and public sector: 0.9 percent.
Top quartile reply rates. What strong campaigns hit.
SaaS and tech: 5.5 to 6.8 percent. Professional services: 4.9 to 5.7 percent. Financial services: 3.6 to 4.2 percent. Healthcare: 3.0 to 3.8 percent. Manufacturing: 2.8 to 3.5 percent.
If you run a tech ICP campaign and pull under 2 percent on a full five-email sequence, something is broken. Run GlockApps before rewriting a word. It is almost certainly an infrastructure problem, not a copy problem.
If you run healthcare or manufacturing at 2.5 percent, that is a strong campaign for those verticals. Benchmark against your actual ICP. Do not compare a healthcare campaign against a SaaS benchmark and conclude your copy is failing.
The variables that move these numbers most.
Infrastructure quality accounts for the largest variance I see between operators. The gap between properly warmed PuzzleInbox Google Workspace inboxes and a poorly configured sending setup is often 1.5 to 2 percentage points on reply rate before any other variable is touched.
List quality is second. ZeroBounce-cleaned, signal-filtered Apollo lists outperform raw unverified pulls across every vertical in this dataset by a consistent margin.
Copy is third. Most operators optimize copy first. That is the wrong order. Fix infrastructure, then list, then copy. The sequencing matters more than most people admit.