What Is a Good Reply Rate for B2B Cold Email? 2026 Benchmarks, Plus 137,962 Emails From One Client

Say your cold email campaign just finished its first month, and you want to know if the reply rate is any good. One tool's report says you're below average. Another tool's data says you're ahead of most senders.

Both can be right, because they count replies differently and compare you with different senders. This post shows how each source counts and the lines we use at Cymate. Then it walks through one client's funnel, from emails sent to booked meetings, with every ratio worked out.

I run Cymate, an outbound agency, so I have my own numbers and my own bias. The funnel comes from one client's campaign over 5 months. It's an example, not a benchmark, and it isn't all of Cymate's results.

What is a good reply rate for B2B cold email?

We aim for a reply rate of at least 2% of emails sent. Before we scale a campaign, we want 3%, with 20% of replies positive.

Typical published rates run from 0.74% to 4.5%. There's no single definition of reply rate, so each source counts its own way. Each one also studies its own users.

Some sources divide replies by emails sent, some by emails delivered, and some by people contacted. The number you divide by is called the denominator. It's the column to read first in the table below.

The outside figures come from Smartlead's 2026 benchmarks, Instantly's 2026 benchmark report and Hunter's State of Email Outreach 2026 report. Cymate is a Smartlead certified partner, and Smartlead is the tool we send with.

A vendor's "good" and its "average" can be different numbers, too. Instantly's report puts its platform average at 3.43% of emails sent, while Instantly's blog says "5-10% is solid for B2B."

So pick the benchmark that counts the way you do. If you divide by emails sent, compare yourself with Instantly's 3.43%. If you divide by people contacted, use Smartlead's bands.

Why do the benchmarks disagree so much?

Partly because they divide by different numbers. It's a boring detail that can double the result, like comparing one player's points per game with another's points per shot. They also count replies differently, use a median or an average, and study different users.

1. What you divide by. Instantly's report divides by emails sent. Its blog divides by emails delivered, with hard bounces taken out. Smartlead divides by people contacted.

2. Per email or per sequence. Hunter explains its 4.5% as "for every 1000 recipients that received a cold email in 2025, only 45 replied." That reads as replies per person across a whole sequence, so follow-up emails don't add to the number you divide by.

3. What counts as a reply. Smartlead's numbers count "every reply, not only positive ones." None of the three big reports says whether auto-replies, like out-of-office notices, are in its count. Apollo's definition of total reply rate includes them.

4. Median or average. Smartlead's 0.74% is the rate of its median sender, the one in the middle of the pack. Instantly and Hunter report averages, which, in Smartlead's words, "a few high performers can pull up."

5. Whose campaigns are in the data. Each tool reports on its own users, and they vary a lot. Smartlead's top 5% get 3.85%, while its typical sender gets 0.74%. Hunter's 4.5% covers marketing and link-building emails too, and its sales outreach sequences average 3%.

Here's a made-up example of how much the denominator alone can move things. Say Jake emails 1,000 people, two emails each, and gets 30 replies.

• Per email sent: 30 / 2,000 = 1.5%

• Per person contacted: 30 / 1,000 = 3%

Same campaign, same replies, and one number is twice the other. That's why you can't hold a per-email rate against Smartlead's per-contact bands.

Jake's example also shows the denominator isn't why Smartlead's figure is the lowest. Counting per person makes a rate bigger, so Smartlead's typical sender would look even lower per email. The likelier reasons are that 0.74% is a median while the others are averages, and that each tool has different users.

Whichever count you use, write it next to the number, and say whether auto-replies count.

• Reply rate per email sent = replies / emails sent

• Reply rate per email delivered = replies / emails delivered (sent minus bounces)

• Reply rate per person contacted = replies / people contacted

• Positive share = positive replies / all replies

I'd use emails sent. It's the number you control, and it's how Instantly's report counts.

What does a full cold email funnel look like with real numbers?

This is one client's campaign over 5 months, from emails sent to booked meetings. The counts are as our dashboard shows them, and they're also on our homepage. The reply count includes every reply, even automatic ones like out-of-office notices.

Because auto-replies are in this one client's 2,336 replies over 5 months, the rate for human replies alone was lower than 1.69%. The 576 positive replies are the cleaner signal.

Every other ratio comes from the same five counts.

In plain terms, it took about 240 emails for each positive reply and about 361 emails for each booked meeting. The campaign averaged about 76 booked meetings a month.

These are booked meetings, as our homepage labels them, not held or qualified ones. For how those differ, see what counts as a qualified meeting.

For a different send, see the raw results from 25,000 emails in 7 days.

How does that client compare with the benchmarks?

It depends on the yardstick. Counted per email sent, this client's reply rate is below Instantly's average. Counted per person, it could pass for a top-10% Smartlead sender or an average Hunter sales sequence.

I'd rather show you both than pick the one that flatters it. The per-person figure is a ceiling, because the 2,336 replies include auto-replies and may count some people twice.

The numbers after the reply are strong either way. For this one client over 5 months, there were 382 booked meetings for 576 positive replies. To me, that's the best number in the chain.

Still, this is one client. Your numbers will depend on your offer, your market and your list.

What is a good positive reply rate?

We want at least 20% of replies to be positive, and under 5% means something's wrong. I couldn't find a neutral benchmark, so that's our own rule of thumb.

Apollo defines a positive reply as "responses that express genuine interest," and teams draw that line in different places, so write yours down.

Instantly's and Hunter's reports don't give a positive-reply figure. Smartlead's numbers count "every reply, not only positive ones."

For the one client in this post, 576 of the 2,336 replies over 5 months were positive, or 24.7%. Since the 2,336 includes auto-replies, the share of human replies that were positive was higher than that.

Counted against emails sent, the positive rate was 0.42%, about 1 positive reply for every 240 emails.

How many emails does it take to book one meeting?

One Cymate client campaign took about 361 emails per booked meeting (137,962 emails and 382 booked meetings over 5 months). Our own outbound, the emails we send to find Cymate's clients, takes about 500 emails per booked meeting.

That 500 comes from our own sending, separate from the client above. We send about 40,000 to 50,000 emails a month from about 200 inboxes, for 80 to 100 booked meetings.

Instantly's rule of thumb is far lower. Its guide to meeting metrics says "average B2B cold email campaigns see roughly 1% of sends turning into booked meetings." That's about 100 emails per booked meeting, by our math.

The page gives no source for that 1%. Its own formula divides by prospects emailed rather than sends, and its table lists 0.5-2% as the 2025 average. The three big reports in the first table don't measure booked meetings at all.

So the client took about 3.6 times as many emails per meeting as that rule of thumb, and our own outbound about 5 times. Counted per prospect, the way Instantly's formula counts, this client needed about 200 prospects per booked meeting, or twice the rule of thumb.

I doubt a meeting for every 100 emails is typical. We do see it on some of our accounts, not all, when we're targeting the right audience with a great offer.

Treat any "meetings per 100 emails" claim as a rough guide, and ask for the raw counts behind it. For the cost per meeting, see in-house SDR vs outbound agency.

What reply rate means something is wrong?

A reply rate under 2% of emails sent is below our baseline. If fewer than 5% of replies are positive, something's wrong.

At Cymate, we read the numbers like this.

For an outside reference point, Smartlead's typical sender gets one reply per 135 contacts, which is its 0.74%. Smartlead's reply-rate guide puts the bottom 25% of senders at 0.37%, or one reply per 270 contacts. The bottom 10% are at 0.16% or lower, which is 625 or more contacts per reply.

When a campaign is below our baseline, I go down this list in order.

1. Offer: what you're selling, and why it's worth a meeting.

2. Positioning: how you describe it. The best version uses your buyer's own words for the problem.

3. Copy: the emails themselves.

4. Data: who's on your list.

5. Infrastructure: the domains, inboxes and sending setup.

When founders tell me cold email didn't work for them, they've usually done the bottom three: the emails, the list and the inboxes. The problem is higher up, and I explain why in why most cold email campaigns fail at step one.

If you're hiring help, the 10 checks in our guide to choosing an outbound agency include the numbers to ask any agency for.

How do you raise your reply rate?

Work down the same list from the top. Offer and positioning matter most, and infrastructure is the easiest part to replace.

Dealight is a good example. It sells an AI tool that books demos on LinkedIn.

Its first emails pitched it as an AI sales rep. They got replies but few demos, because buyers saw that space as crowded.

We changed the pitch to lead with the LinkedIn angle, narrowed the list to sales leaders at small service firms and tightened the emails. Reply rates rose by over 40%.

Its case study reports 24 demos booked in 3 weeks from 30,000+ emails. Reply rates averaged 4-6% across campaigns, though the page doesn't say per email or per contact.

The rest of our setup, one line each.

• Copy: 50-75 words. A hook that's specific to the buyer, the pain, how you solve it, and a soft ask.

• Sequence: two emails, the first and one follow-up about 3 days later.

• Sending: Google Workspace inboxes only, at 10 emails per inbox per day. Each new inbox gets at least 3 weeks of warm-up, sending small amounts so email providers learn to trust it.

Also set up SPF, DKIM and DMARC on every sending domain. These settings prove your email really comes from you, and Google's sender guidelines recommend all three.

Hunter's report has a few useful cuts. Sequences from a custom domain got 5.2%, against 2.5% from free email addresses. Sequences sent to 21-50 recipients got 6.2%, against 2.4% for 500 or more.

Two findings differ from our rule. Hunter saw 6.8% with three messages against 3.3% with one. Instantly's report found that 42% of replies came from follow-ups, and it puts the sweet spot at 4-7 emails.

We send two and judge that by our own results. Whatever length you pick, don't skip the follow-up.

For the three-part system we use to book sales calls at volume, see how to get 100 sales calls per month with cold email.

Does cold email still work in 2026?

Yes. One client's campaign booked 382 meetings over 5 months, and our own outbound books 80 to 100 meetings a month.

I grew Cymate from zero to seven figures with cold email as our main way to win clients. Smartlead's 2026 benchmarks page says results "depend heavily on the sender." In our experience, the offer and positioning matter most.

The bottom line

A good reply rate depends on how it's counted. We treat 2% of emails sent as the floor, want 20% or more of replies positive, and look for 3% before we scale. Say how you count, auto-replies included or not, and compare yourself only with benchmarks that count the same way.

If you'd like a second set of eyes on your numbers, bring them to a 25-minute intro call. You can Run a Trial with Cymate, or see what our outbound services include first.