Published:
Cold email reply rates: what the data says
Most cold emails get no answer. We read the largest public datasets and set them against the numbers of a live pilot: Clara, the AI salesperson Arlo grew out of, writing to marine companies on the German coast since May.
The baseline: one in thirty
- Average3.43%
- Top 25%5.5%
- Clara, our pilot9.3%
- Top 10%10.7%
The average cold email gets a reply 3.43% of the time. A quarter of senders reach 5.5% or more, the best tenth 10.7% or more. Hunter’s data for 2024 lands close, at 4.1%: about 96 of every 100 cold emails go unanswered.13
A pilot, in numbers
Clara is Arlo’s predecessor, built for one of our pilot customers, a company that sells a software platform to manufacturers and fleets. Since May Clara has written to marinas, shipyards, rentals and boat builders: five emails per contact over about five weeks, in the contact’s own language.6
- FindA lead agent searches one region at a time, finds companies and their decision makers, and files them in the CRM.
- WriteClara reads what each company does and writes to the contact in their own language.
- Follow upFive emails over about five weeks, Tuesday to Thursday only. She stops the moment someone replies or signs up.
- Hand overA reply agent reads every answer, handles opt-outs and passes warm leads to the sales team within a day.6
- companies reached
- 442
- contacts
- 464
- emails sent
- ~1,000
- replies
- 43
- reply rate
- 9.3%
- warm leads to the sales team6
- 8
That is 2.7 times the average, from the first campaign. One reply in five was warm enough to hand to a person: one warm lead for every 58 contacts. Clara runs on the pilot’s own tools, Zapier and HubSpot; Arlo grew out of what she taught us. These are Clara’s numbers, not yet Arlo’s.16
Where you write changes everything
- Lower Saxony and Bremen (66)16.7%
- Schleswig-Holstein (221)8.6%
- Mecklenburg-Western Pomerania (160)6.3%
Nearly three times the replies between two neighbouring coasts, with nothing else changed. A benchmark tells you where you stand. Your own numbers per region and target group tell you where to write next.6
Fewer recipients, more replies
- Up to 50 recipients5.8%
- Over 1,000 recipients2.1%
- Not relevant to them71%
- Impersonal43%
- Don’t trust the sender36%
The reason people ignore cold email is rarely the subject line or the hour. It is that the email has nothing to do with them. Small lists of companies that fit get almost three times the replies of mass sends.3
Short, and about them
- 0–19 words3.7%
- 20–39 words4.5%
- 40–59 words3.4%
- 60–79 words4%
- 80–99 words4.1%
- 100–119 words3.7%
- 120–139 words3.6%
- 140–159 words3.5%
Emails of 20 to 39 words do best in Hunter’s data, but the spread is small, and Hunter itself warns that cutting words alone lifts nothing. Instantly finds its best campaigns under 80 words. Length follows from content: say one thing that matters to the reader, and stop.12
Clara reads what each company does and writes 20 to 39 words in the contact’s language: one observation, one line on us, one soft question.
The follow-ups bring half the replies
- With follow-ups4.9%
- First email only3%
Between 41% (Belkins, 7.5 million emails) and 58% (Instantly) of replies come from the first email. The rest come from follow-ups. At Belkins the third email alone books more than a third of all meetings from email. Instantly puts the sweet spot at four to seven emails, three to four days apart.14
The cost is real: a quarter of decision makers find follow-ups annoying, and unsubscribes rise after the third. Clara sends five, on Tuesday to Thursday only, and stops the moment someone replies or signs up.36
The day matters less than you think
- Monday0.45%
- Tuesday0.45%
- Wednesday0.48%
- Thursday0.48%
- Friday0.44%
- Saturday0.35%
Monday to Friday differ by four hundredths of a percentage point. Wednesday and Thursday lead by a hair, Saturday trails, and the morning beats the afternoon slightly (0.54% against 0.47%). The send time is a detail. Who you write to, and what you say, is not.5
What the numbers add up to
- Write to fewer companies that fit. Relevance is the main reason people answer, or don’t.
- Say one thing about them, one about you, and ask one question. Under 80 words.
- Plan four to seven emails, and stop the moment someone answers.
- Read your numbers per region and target group, not only the total: Clara’s best region replies at 16.7%.
- Keep bounces under 2%. Check addresses before you write.136
How Arlo works the same way
Arlo, our salesperson, is built around the same lessons. It finds companies region by region and per target group, each with the reasons it fits and the sources it read.
It writes first emails and follow-ups in a language of the prospect’s country that you name, and sends on days and at times in the recipient’s local time. The first email to someone new waits for your approval, unless you let Arlo send first emails on its own.
A reply pauses its other emails to that company. A no ends it, at once and for good, in every mailbox. A daily limit and a slow start hold the volume, and Arlo pauses on its own when bounces or complaints rise.
It learns from the replies, shows you what it learned before it uses it, and keeps it in your account. We will publish Arlo’s own numbers here once it has written to enough real people to count.
Sources
- 1Cold Email Benchmark Report 2026, Instantly
- 2What's the Best Word Count & Length for Cold Emails, Hunter
- 3The State of Cold Email 2025, Hunter
- 4Sales follow-up statistics in B2B, Belkins
- 5What is the best time to send B2B emails? Belkins' 2026 data study, Belkins
- 6Clara in Numbers, report of our pilot customer, HubSpot data as of 20 October 2026