Cold Email Open Rate Benchmark Report 2026

Key metrics

Data scope: Since June 6, 2026 (when ECS pixel tracking came online). Split by enterprise email and personal email channels. Only records with complete tracking data are included. We track four main metrics: unique open rate, avg reads per person, repeat open rate, and forward signals — first publicly reported in a cold email benchmark.

Forward signals are expressed per 100 delivered: enterprise email 505÷1,956=25.8, personal email 35÷440=8.0, combined 540÷2,396=22.5. This metric is first disclosed by TokoAI in a cold email benchmark; Woodpecker/GMass/Mailshake/Instantly reports have no comparable value.

Industry benchmarks sourced from Woodpecker, GMass, Mailshake, Instantly 2025 cold email reports. Manufacturing unique open rate ~20-22%, repeat open rate ~10%. Data queried 2026-07-19, full scan of runtime.db outcomes table (since June 6, 741 unique openers, 1,774 total open events). Forward detection uses IP diversity, device diversity, and UA differences.

Variant performance ranking

Variants with over 100 sends, ranked by open rate:

Key observation: v24-b (40.9%) vs v10 (0.0%) — a 40.9pp gap with the same send channel and domain. The difference comes entirely from strategy choice. V10 and l4 used "negative framing" that likely triggered spam classification in manufacturing environments. V24's buyer-personalized templates significantly outperform generic question templates. On avg reads, pain-point questions (v21 2.6x, v22-c 2.7x) outperform templates (2.0x) — questions spark contemplation, making recipients re-read. p1 (50 sends only) had 78.0% open rate and 3.5x avg reads — small sample, not statistically significant.

Methodology

Data source & technical approach

All data sourced from the TokoAI production database runtime.db outcomes table — full scan, no sampling, no estimation. Tracking data written via each channel's API.

• Send channels: Enterprise email service (enterprise email) + personal email service (personal email), used to distinguish recipient types
• Tracking method: Native API open tracking + ECS pixel tracking; open events written to outcomes.open_count
• Data range: 2026-06-06 to 2026-07-16 (queried 2026-07-19). Starting from June 6 when ECS pixel tracking came online; earlier data excluded due to incomplete tracking
• Recipient profile: Manufacturing exporters (small-to-medium factories), covering North America, Europe, Southeast Asia
• Variant strategy: 43 active variants across 24 iterations, one variable changed per iteration

Definition notes

• Unique open rate: records with opened=1 / records with status='sent'. Deduplicated per recipient
• Avg reads per person: SUM(open_count) / opened=1 count. Measures how many times one interested person reads the email. This report baseline: 2.39x
• Forward detection: Multi-signal approach using IP diversity, device/OS/browser diversity, and X-Forwarded-For chain analysis. Each forward signal scores 1-3; cumulative score above threshold counts as a valid forward

Limitations & biases

• Sample is concentrated among manufacturing exporters; not directly generalizable to other industries
• Apple Mail Privacy Protection (MPP) may inflate unique open rates by 3-5pp
• Email security gateways may block tracking pixels, undercounting opens
• Send timing varies across variants (early ad-hoc sends, later batch sends), which may introduce a confounding effect
• Personal email channel open log retention is 7 days; historical avg reads per person may be understated

Data evolution

* Cumulative delivered includes early data with incomplete tracking. From June 6 onward tracking coverage is 100%.

Citation

@misc{tokoai2026coldemail,
  title={Cold Email Open Rate Benchmark Report 2026},
  author={{TokoAI}},
  year={2026},
  howpublished={\url{https://51toko.com/en/report/cold-email-benchmarks/}}
}

Changelog