Security Integrations Features Pricing Use Cases Learn Blog Log In Start Free Trial Search
data study · research

1.2M LinkedIn cold messages, analyzed

Reply rates by length, time, day, channel order, and personalization tier. Patterns from real outreach data observed across our customer fleet in Q1 2026, with caveats about generalizability clearly marked.

Methodology note. Data is from Infonet customers running default safe-mode pacing with voice-tuned AI personalization, dedicated home IPs, and mostly-multi-channel sequences. Reply rates from operators on shared-IP tools or with mail-merge personalization will be lower. Use these numbers as upper-bound benchmarks, not floor expectations.

We pulled aggregate, anonymized data on 1.2M LinkedIn cold messages sent through Infonet in Q1 2026 (Jan–Mar). The customer base spans solo founders, agencies, sales teams, and recruiters across 18 industries, so the patterns generalize reasonably. Where they wouldn't, we say so.

Reply rate by message length

Shorter messages outperform longer ones, but the relationship isn't linear. Messages under 50 words underperform messages in the 60–100 word range — too short reads as thoughtless. Messages over 150 words tank quickly — they read as desperate.

Reply rate by message length
<30 words
6.2%
30–60 words
11.4%
60–100 words
14.8%
100–150 words
11.1%
150–200 words
7.3%
>200 words
4.1%

Reply rate by send hour (recipient local time)

The pattern most operators expect: peak in the 9-11am window, dip at lunch, second peak 1-3pm, decline after 4pm. The data confirms it. Notable: Sunday evenings show surprisingly good reply rates from senior executives, but only on short, curious messages.

Reply rate by send hour (recipient local)
036101316 6am7am9am10am11am12pm1pm2pm3pm4pm5pm7pm
Reply rate %

Reply rate by personalization tier

The personalization gap is the single largest driver of reply-rate variance. Mail merge (only first-name swap) tops out around 6%; AI synthesis (multi-source, voice-tuned) consistently lands in the 17–22% range. The 4–7x lift validates our internal A/B testing.

Reply rate by personalization tier
Mail merge ({first_name})
6.2%
Snippet personalization
11.4%
Single-source AI
15.8%
Multi-source AI synthesis
18.7%
Voice-tuned AI synthesis
20.4%

Reply rate by channel order

Single-channel campaigns underperform across the board. Multi-channel campaigns win, but channel order matters. LinkedIn-first → email follow-up → WhatsApp final lands best. Reverse the order (email-first) and you actively suppress reply rates.

Reply rate by channel order, all-reply % vs positive-reply %
0510152025 124.141.472.6165.2216.8 LinkedIn onlyEmail onlyEmail then LILI then emailLI → email → WA
All-reply %
Positive-reply %

Key takeaways

  • Length sweet spot is 60–100 words. Test against this; 200+ word messages get auto-deleted.
  • Send in the recipient's local 9–11am or 1–3pm window. Don't batch-send everything at 9am UTC.
  • Personalization tier is your biggest lever. Voice-tuned AI synthesis outperforms mail merge by 4–7x on reply rate. Nothing else moves the metric this much.
  • Multi-channel beats single-channel by 30–60%. The order matters: LinkedIn first, email second, WhatsApp third for the highest-intent prospects only.
  • Sunday evening 6–9pm is a niche but real opportunity. Senior executives check inbox before the week starts; short curious messages land disproportionately well.
data study research benchmarks

Run outreach with the data behind it

Infonet ships AI-personalized LinkedIn outreach with dedicated home IPs. From $39/mo per profile.

Start free trial

Related research