Back to Insights

The average cold email reply rate is not yours

TL;DR

  • Instantly puts the average cold email reply rate at 3.43%. Smartlead puts the median sender at 0.74%.

  • The gap is 4.6x. It comes from averages versus medians, and from counting emails versus counting people.

  • A typical sender needs 135 contacts for one reply. The top 5% need 26.

  • Small lists reply far better. Lists under 50 contacts average 5.8%. Lists over 1,000 average 2.1%.

  • Plan on the median. If you beat it, that is upside, not the plan.

You decide to do cold outreach. You look up the reply rate first, because you want to know how many emails it takes to get a meeting.

The first result says 3.43%. You do the maths. Five hundred emails, seventeen replies, a few meetings. That looks like a plan.

Then you send five hundred emails and get four replies. Two are polite noes. One is an out of office. You have one real conversation.

You did nothing wrong. You planned on the wrong number.

Why do two benchmarks disagree by 4.6x?

Because they measure different things, and both call the result a reply rate.

Instantly reports an average. It takes all replies, including replies to follow-ups, and divides by all emails sent, across thousands of workspaces, from January to December 2025. The result is 3.43%.

Smartlead reports a median, per person contacted. It takes more than 850 million emails from January to June 2026, looks at each sender, and finds the middle one. That sender gets one reply per 135 people contacted. That is 0.74%.

Divide 3.43 by 0.74 and you get 4.6. That is the whole disagreement. Not better campaigns. Not a different year. A different way of counting.

Founders Engine Room laid the two side by side in November 2026 and the gap survives every check. Other 2026 datasets land in the same spread. Belkins reports 0.45% per email sent. Hunter reports 4.5% per recipient. The range across honest sources runs from under half a percent to over four percent.

What does the median sender actually get?

Smartlead publishes the full spread, and the spread is the useful part.

  • Median sender: 135: 0.74%

  • Top 25%: 70: 1.43%

  • Top 10%: 38: 2.63%

  • Top 5%: 26: 3.85%

Read that table again. The top 5% of senders need 26 contacts for a reply. The median needs 135. The bottom 10% need 625.

Half of all senders sit below one reply per 135 people. If you are a normal founder sending your first campaign, the median is your benchmark. The average belongs to someone else.

And a reply is not a meeting. Some replies are no. Some are not now. Some are questions that go nowhere. Your reply count is the top of the funnel, not the result.

Why does the average look so much better?

Because a few strong senders pull it up.

An average adds everyone together and divides. A handful of senders with tight lists, real personalisation and good timing can clear 10%. Instantly's own data says its top 10% clear 10.7%. Those senders drag the average up for everyone else.

A median does not care about the best senders. It lines everyone up and takes the middle. That is why it is lower, and why it is honest about what a typical campaign does.

There is a second difference. Instantly counts replies per email sent, and follow-up replies count too. Smartlead counts replies per person contacted. One person can receive seven emails and reply once. Counted per email, that looks like one reply from seven sends. Counted per person, it is one reply from one contact. Same campaign, different number.

Neither source is lying. Both are vendor data, from their own users, and both say so. The mistake is reading the average as a promise.

How many emails do you need for one meeting?

Work it out from the median first.

Five hundred contacts at 0.74% gives you about four replies. Five hundred contacts at 3.43% gives you about seventeen. Same list, same emails, two different plans. One of them leaves you short by thirteen replies.

Now do the same for a thousand contacts. At the median, that is seven replies. At the average, thirty-four. If your month needs five meetings, and a reply turns into a meeting maybe a third of the time, the median plan says you need far more than a thousand contacts. The average plan says you are already done.

This is where founders get hurt. They build the month on seventeen replies, send five hundred emails, get four, and conclude outreach does not work. Outreach works. The maths was wrong.

Your lead went cold in seven days for the same reason. Follow-ups matter, and most of the replies in the Instantly average come after the first email. Smartlead's own data puts the first email at 1.18% and the seventh at 0.33%. The follow-ups add replies, but each one adds fewer than the last.

What actually moves the reply rate?

Three things show up in every dataset. None of them is sending more emails.

Small lists beat big lists. Belkins studied 16.5 million emails and found lists under 50 contacts average a 5.8% reply rate, while lists over 1,000 average 2.1%. A short list of people you actually researched replies nearly three times better than a long list you bought.

Real personalisation beats a template with a first name in it. Woodpecker's analysis of 20 million emails puts advanced personalisation at 17 to 18% replies, against 7 to 9% for generic templates. The difference is not the merge tag. It is whether the email shows you looked at their business.

Short first emails beat long ones. On Smartlead's data, first emails under 50 words reply at 1.47%, about double the typical 0.74%. Say one specific thing, ask one question, stop.

None of this is a trick. It is the same rule in three forms. A short list, a real reason, a short email. That is what gets a reply from a person who deletes most of their inbox before breakfast.

One more small change is usually the whole fix. You do not need a new tool. You need a smaller list and a better first line.

How should you plan your pipeline then?

Plan on the median, and treat anything above it as upside.

Start with the number of meetings you need. Work back to replies, then to contacts, using 0.74% as the reply rate. If that number of contacts looks too big, the answer is not to swap in the average. The answer is to shrink the list and raise the quality until your own rate beats the median.

Then run the campaign in small batches and watch your own number. After fifty contacts you will know if you are near the median or far from it. After a hundred, you can plan the rest of the month on your own rate instead of anyone's benchmark.

Do not judge the campaign on day one. A reply is not a meeting, and a meeting is not a close. One client is half your business, so a pipeline built on one channel and one hopeful number is a risk, not a plan.

And check the source before you trust the number. Verify your AI agent like you mean it applies to benchmarks too. A vendor report describes that vendor's users. It does not describe you.

The honest version of the plan fits in one line. At the median, 135 contacts buy one reply. If you want ten replies, you need 1,350 contacts, or a list good enough to beat the median. Everything else is hope with a spreadsheet around it.

Frequently Asked Questions

What is a good cold email reply rate in 2026?

For a typical sender, 0.74% of contacts replying is the median, based on Smartlead's 850 million email study from January to June 2026. The top 25% reach 1.43% and the top 5% reach 3.85%. Averages run higher, around 3.43% on Instantly, because a few strong senders pull the average up. Judge yourself against the median first.

Why is the median reply rate so much lower than the average?

An average adds all senders together, so a few excellent campaigns pull it up. A median takes the middle sender and ignores the extremes. The two also count differently. Instantly counts replies per email sent, including follow-up replies. Smartlead counts replies per person contacted. Same campaigns, different denominators, different numbers.

How many cold emails do I need to send to get one reply?

At the median rate of 0.74%, you need about 135 contacts for one reply. At the top 5% rate of 3.85%, you need about 26. These are contacts, not emails, so follow-ups to the same person do not count twice. A short, well researched list will usually need fewer contacts than a long, bought one.

Do follow-ups increase the reply rate?

Yes, but each follow-up adds less than the one before it. Smartlead's data puts the first email at 1.18% and the seventh at 0.33%. Follow-ups are worth sending, and most campaigns stop too early, but they do not turn a weak first email into a strong campaign. Fix the first email and the list first.

Does personalisation actually improve reply rates?

Yes, by a lot, when it is real. Woodpecker's study of 20 million emails found advanced personalisation replying at 17 to 18%, against 7 to 9% for generic templates. A first name in a template is not personalisation. Quoting something specific from the prospect's own site or business is.