Do Cold Emails Work? What a Test of 3,570 Emails Showed

Whether cold email works for agencies, shown with a controlled test of 3,570 emails, the arithmetic from reply to client and the rules your email has to pass.

Thijs Veer, Founder

9 min read

Contents

Do cold emails work? Yes, when the list gives each business a reason to reply. In a test of 3,570 cold emails that we published in October 2026, a random sample of other sites got 0.50% positive replies. Businesses with the weakest 20% of websites got 1.60% with a generic email and 2.35% when the email named their problem. Who you email moved the result more than what you wrote.

The gap between 1.60% and 2.35% was not statistically significant. The rest of this page shows that test, why other sources quote such different numbers, and the arithmetic from a reply rate to a signed client.

Do cold emails work for agencies?

Cold email works for agencies that sell websites and marketing to businesses with a visible problem. It works badly for anyone who emails a broad list with a generic pitch. The channel is the same in both cases. The list and the reason to write are different.

"Work" can mean four things.

  1. Delivered. The email reaches the inbox and avoids the spam folder.
  2. Replied. Anyone answers, including "remove me" and "not interested".
  3. Positive reply. A positive reply is an answer from someone who wants to talk.
  4. Client. Someone pays you.

Only the fourth one pays the bills. A campaign can look good at level 2 and still produce nothing at level 4. Our guide on what a cold email is covers the basics. This page is about results.

What did our test of 3,570 cold emails show?

The test showed that choosing who to email made a positive reply about 3.2 times as likely. We scored 20,000 Dutch company websites and took the 20% with the most problems as prospects. Then we sent 3,570 emails over 5 working days from 24 sending addresses (our write-up).

GroupWho got the emailWhat the email saidPositive repliesPer 1,000 emails
AA random sample from the other 80% of sitesA generic email0.50%5
BThe weakest 20% of sitesA generic email1.60%16
CThe weakest 20% of sitesAn email that names the site's problem2.35%23.5

The last column is our arithmetic. It multiplies each rate by 1,000.

Three findings stand out.

  • Selection did most of the work. Group B got the same generic email as group A and still replied positively more than 3 times as often.
  • Naming the problem helped less than it looks. The gap between B and C was not statistically significant, p = 0.238. It may be chance.
  • The cost was small. The whole experiment cost €242.25. Generating 1,190 problem descriptions cost about €5. Each positive reply in the selected groups cost €5.15.

The limits matter too. The study was on Dutch companies. It tested email only. It left out businesses with no website, and it says nothing about letters or LinkedIn. Treat it as one controlled test. It is no law of nature.

Why do cold email reply rates differ so much between sources?

Reply rates differ because sources count different things. Some count every reply. Some count only interested ones. A benchmark from a vendor's own platform reflects that platform's users, lists and markets.

SourceYearSampleWhat it countsRate
Backlinko and Pitchbox201912 million outreach emails, link requests includedAny reply8.5%
It'sJustContent campaign, from our October 2026 write-upBefore Oct 20261,382 Dutch companiesPositive replies0.36%
Our test, group APublished Oct 2026Random sample from the other 80% of Dutch sitesPositive replies0.50%
Our test, group CPublished Oct 2026Weakest 20% of Dutch sitesPositive replies2.35%

An 8.5% reply rate and a 0.50% positive reply rate can both be true. The first counts every answer to any outreach. The second counts only people who want to talk.

Before you compare your campaign to any benchmark, check three things. What counts as a reply, who was emailed, and who published the number.

How many cold emails does it take to win one client?

With our test's reply rates and a made-up close rate, one client takes about 383 emails from a chosen list and 1,800 from a random one. Here is the arithmetic, step by step. Swap in your own numbers where you have them.

  1. Pick your positive reply rate. We use 2.35% for a chosen list and 0.50% for a random one, from our test. The 2.35% was not significantly different from the 1.60% a generic email got on the same list.
  2. Pick your call rate. In the It'sJustContent campaign, 5 positive replies led to 3 sales calls. That is 60%, from one small campaign.
  3. Pick your close rate. Say 1 in 5 calls becomes a client. This figure is made up. Use your own.
  4. Work back from one client. One client needs 5 calls. 5 calls at 60% need 5 ÷ 0.6 = 8.3 positive replies. Round up to 9.
  5. Work out the emails. 9 ÷ 0.0235 = 383 emails for a chosen list. 9 ÷ 0.005 = 1,800 emails for a random list.
StepChosen list (2.35%)Random list (0.50%)
Clients wanted11
Calls needed at 1 in 555
Positive replies needed at 60%99
Emails needed3831,800

In this arithmetic, the same client costs 4.7 times as many emails from a random list. That is the strongest answer to "does cold emailing work". It does, and the list decides how hard you work for each client.

One warning from the same campaign. It'sJustContent got 5 positive replies and 3 sales calls, then 0 clients. Replies are a start. The call, the offer and the price still have to do their part.

Is cold emailing worth it for a small agency?

Cold emailing is worth it when your cost per positive reply is lower than what one client is worth to you. In our test, a positive reply from the selected groups cost €5.15.

The real cost is time. Reviewing the 1,382 sites from the earlier campaign by hand would have taken about 184 hours, at about 8 minutes a site. Nobody did it. That review is the step that turns a random list into a chosen one, so it is where your hours go.

Three questions tell you whether it is worth it for you.

  • Can you name a reason to write to each business? A slow site, a broken form or a page that fails on a phone is a reason. "We build websites" is no reason.
  • Can you send from a domain that passes authentication? If not, fix that first.
  • Can you follow up? In the Backlinko and Pitchbox study of 2019, campaigns of three or more messages did best. Our cold email follow-up templates give you the next three emails.

If the answer to all three is yes, cold email is cheap to test. Send 400 emails to a chosen list and count positive replies.

Why does cold emailing fail for most senders?

Most cold email fails for one of four reasons. The email never reaches the inbox, the list has no reason to reply, the message could go to anyone, or there is no follow-up.

It never arrives. Since February 1, 2024, Google requires every sender to personal Gmail accounts to set up SPF or DKIM and to keep spam rates in Postmaster Tools below 0.3%. Senders of more than 5,000 messages a day to Gmail accounts also need SPF, DKIM and DMARC (Google). Miss these and your email may be rejected or marked as spam. Our guide to cold email deliverability covers the setup.

The list is random. Group A in our test is the random list. It got 5 positive replies per 1,000 emails.

The message is generic. A first line that could go to any business tells the reader it went to every business. A first line about their own site shows you looked.

There is no follow-up. One email and silence gives up the replies a follow-up would bring. In the 2019 Backlinko and Pitchbox study, one follow-up brought 65.8% more replies.

The law matters as well. In the US, the CAN-SPAM Act covers business email too. It bans false header information and deceptive subject lines, and it requires a valid postal address and a working opt-out (FTC). Our post on whether cold email is legal has the full list. This is not legal advice.

Are cold emails effective on their own?

Cold emails can be effective on their own, and our test shows it. The test covered email only, so it cannot tell you how email does next to other channels. We have no measured reply rate for letters or LinkedIn steps, and we claim none here.

Our view, which no test has measured yet, is that an email lands better as a second touch after a letter. Leadodo's Outbound Machine runs six steps over 19 working days. A posted letter arrives on day 1. Emails follow on days 4, 7 and 11, and LinkedIn on days 15 and 19. The sequence stops for a business the moment it replies.

The letter gives the email something to refer to. In the US a letter costs about 86 cents. That covers one page in black and white, sent First Class, with printing, postage and the envelope. Your Outbound Machine sends it under your agency's name. Leadodo never contacts anyone itself.

Choosing the list first applies to every step. Twip AI scores every website in a search out of 100 across 10 domains, and sites that score below 40 form the shortlist. Twip's findings become the opening line of each letter and email (AI Analysis). The full method sits in our pillar on cold email for agencies.

Questions

Is cold email dead?

No. The test we published in October 2026 got 2.35% positive replies from a chosen list, against 0.50% from a random one. What has changed is the bar for getting delivered. Gmail's sender requirements have applied to everyone who sends to Gmail accounts since February 1, 2024.

What is a good response rate for cold emails?

It depends on what you count. A positive reply rate of 2% or more from a list of businesses with a clear problem is a strong result by our test. An "any reply" rate will be higher, so compare like with like.

Under federal law, yes, if you follow the CAN-SPAM Act (FTC). Its rules include honest headers and subject lines, a valid postal address and a working opt-out that you honor within 10 business days. This is not legal advice.

Why are my cold emails getting no replies?

Check delivery first. Send a test to a Gmail account and confirm SPF or DKIM pass. Then look at the list. If you chose businesses at random, you are in group A territory, at about 5 positive replies per 1,000 emails.

How long before cold email shows results?

You can read a first result after a few hundred emails and their follow-ups. At the 2.35% rate from our test, 400 emails would bring about 9 positive replies. That is arithmetic on one test, and your list will differ. Fewer than 100 emails tells you little, because chance decides too much.


Next step. See how your Outbound Machine posts the letter first and times each email after it. Outbound Machine.

Keep reading

All posts

Your next client is already in the database.

Get started, or book a demo first.