Cold email reply rate benchmarks

What is a good cold email reply rate?

Across 6,074,430 emails, the average human reply rate is 1.6% per prospect contacted, or 0.8% per email sent. Including auto-replies, those become 3.2% and 1.6%. A campaign landing above 2% human replies per prospect is doing well. Below 1% usually means the list, not the copy.

That is the whole answer, and the rest of this page is why those four numbers differ so much from the ones usually quoted.

Sample: 6,074,430 emails to roughly 3.0 million unique prospects, across 28 Instantly workspaces and 1,568 campaigns, sent between 30 October 2023 and 12 October 2026. About 87% of that volume is from 2026.

These are averages across every campaign in the dataset, including ones that failed. A well-run campaign on a well-built list clears them by a wide margin, and the section on why a good campaign beats the average covers what that looks like.

Why are most published reply rates twice what they should be?

Because out-of-office replies get counted as replies. In this dataset there were roughly 49,000 auto-replies against 48,000 human replies, so auto-replies make up 51% of everything that lands in the reply folder. Any benchmark that does not separate them is reporting a number about twice its real value.

This is the single most common distortion in published cold email statistics, and it is not usually deliberate. Sending platforms count a reply as a reply. Nobody opens 97,000 messages to sort them.

The practical consequence: when a vendor quotes a 3% reply rate, ask whether that is human replies. If they do not know, it is probably about 1.5%. If a campaign dashboard shows replies climbing while the calendar stays empty, the auto-reply share is usually the explanation.

There is a second reason the gap matters. Auto-replies are not noise, they are information. An out-of-office confirms the address is live and monitored, which is a deliverability signal worth tracking. It is just not interest, and counting it as interest is how a campaign looks healthy for six weeks while producing nothing.

Cold email reply rates by industry

Reply rate varies more by industry than by almost anything else a sender controls. Cleaning and janitorial replies at 1.2% per email sent; SaaS and tech at 0.4%. That is a threefold spread driven by inbox volume and how used to cold outreach each market has become.

All figures are per email sent. The dataset averages about two emails per prospect, so per-prospect rates run roughly double these.

Three patterns are worth pulling out of that table.

The hardest markets are the ones that buy software. SaaS and tech sits last at 0.4%, and marketing agencies are second from last at 0.5%. Both are populated by people who run outbound themselves, receive a great deal of it, and recognise a sequence on sight. Selling to someone who writes cold email for a living is the hardest version of this job.

Professional services has the widest gap between human and auto. 0.6% human against 2.0% including auto, so more than two thirds of what arrives is an out-of-office. Accounting and legal firms run heavy auto-responders. A campaign into that market will look like it is working long before it is.

Volume is not quality. Home services is the largest segment in the dataset at 770,000 emails and sits mid-table on reply rate with the second-lowest positive rate. A market being large and reachable does not make it responsive.

Should reply rate be measured per email or per prospect?

Per prospect is the more honest number and the one most benchmarks use without saying so. A sequence sending three emails to one person who replies once is a 100% prospect reply rate and a 33% email reply rate. Both are true. Only one describes whether the market is interested.

This is why published benchmarks disagree so wildly. A page quoting 5% and a page quoting 1.5% can be describing identical campaigns measured two different ways, and neither usually says which.

Two questions settle any benchmark: is it per prospect or per send, and does it include auto-replies. Every figure on this page states both.

What share of replies are actually positive?

About 13%, or roughly one in seven. Of 48,000 human replies, 6,472 were marked positive, meaning interested or a meeting booked. Of those, 1,728 were a confirmed booked meeting. Per email sent that is a positive reply rate of 0.1%.

That ratio is the number most senders have never calculated, and it changes how a campaign should be judged. Six replies out of a thousand emails sounds like failure. If one of those six is interested and the deal is worth $30,000, the campaign is working.

Those are averages including every campaign that failed, so treat them as a floor rather than a forecast. The useful thing about them is the shape: the drop from replies to positive replies to booked meetings is steeper than most forecasts assume, and planning against reply rate alone consistently overestimates pipeline.

Why a well-run campaign beats every number on this page

These averages cover three years, 28 workspaces and 1,568 campaigns, including the first ones ever sent, clients who churned, and campaigns that never worked. That is what an honest aggregate looks like, and individual accounts run well above it.

From the same dashboards, in accounts Grow Surely runs:

The gap between those and the aggregate is the point of publishing both. An average that includes every failure is the only kind worth benchmarking against, and the distance between it and a working campaign is the actual measure of whether targeting and message are doing their job.

Three things move a campaign from the aggregate to the top of that table, and none of them is writing.

List precision. A reply rate below 1% per prospect is almost always a list problem rather than a copy problem. Copy changes move a reply rate by fractions. Pointing the same copy at the right 2,000 people instead of the wrong 20,000 moves it by multiples.

Deliverability. None of this matters if the emails are not arriving. Reply rate is the last metric in the chain and the first one blamed. Domain health, warmup and sending volume decide whether a campaign ever gets the chance to be judged on its message.

Offer. The widest variation in the dataset is between offers, not industries. The same market responds very differently to a request for a call and to something given away first, which is why comparing two campaigns across different offers tells you almost nothing.

How this was measured

Stated in full so the figures can be checked, argued with, and cited.

Source. Campaign analytics pulled directly from Instantly across every workspace Grow Surely operates: 26 client workspaces, Grow Surely's own, and one shared pilot workspace. Duplicates removed.

Period. 30 October 2023 to 12 October 2026. About 87% of volume is from 2026, 11% from 2025, 2% from 2023-24.

Sample. 6,074,430 emails to approximately 3.0 million unique prospects across 1,568 campaigns, an average of 2.0 emails per prospect.

Definitions. A human reply is a reply from a person, excluding auto-responders and out-of-office messages. A positive reply is one marked as interested or meeting booked. Reply rates are given both per unique prospect and per email sent, labelled on every figure.

Exclusions. Three industry rows were removed because each was driven mostly by a single client's offer, which would describe that offer rather than the industry: fitness and gyms, tours and venues, and car dealerships. One workspace showing roughly ten times the average reply rate from 25,000 sends was excluded as a labelling artefact. Industries under 50,000 sends were not published.

Coverage. The industry table covers 4.5 million emails sent by campaigns that still exist; 2.95 million of those sit in the ten published rows. A further 345,000 sent by campaigns whose names carry no industry, and 545,000 from campaigns since deleted, are in the totals but not the industry breakdown.

Why the headline figure is a floor. 6,074,430 excludes clients who have churned and whose workspaces are no longer connected, one workspace that returned an error, and any sending outside Instantly. The true figure is higher and is not published because it cannot be verified.

What this data does not show

Worth stating plainly, because a benchmark that only lists its strengths is marketing.

No open rates. Open tracking is switched off across these accounts, since tracking pixels hurt deliverability and Apple Mail Privacy Protection made the numbers meaningless in 2021. There is no open-rate benchmark here and anyone publishing one post-2021 should be asked how they measured it.

Positive replies are not labelled identically everywhere. Some workspaces use an AI classifier, some are marked by hand. The 13% positive share is a good estimate rather than an exact figure.

Industry is derived from campaign names. It is accurate for most campaigns and unavailable for deleted ones, which is why 545,000 sends sit outside the breakdown.

One agency's clients are not the whole market. This is a large dataset from a single operator with one methodology and a particular client mix, weighted toward North American B2B and service businesses. It is not a random sample of all cold email everywhere.

Averages hide offers. The widest variation here is between offers rather than industries or copy, and offer is the one variable this breakdown cannot isolate.

Citing this data

These figures are free to quote, republish or chart, with attribution and a link. No permission needed and no form to fill in.

Suggested wording: According to Grow Surely's analysis of 6,074,430 cold emails, the average human reply rate is 1.6% per prospect contacted, and auto-replies account for 51% of all replies received.

One request rather than a condition: if a figure is quoted, quote the denominator with it. The whole argument of this page is that a reply rate without "per prospect" or "per email sent" attached, and without saying whether auto-replies are counted, is a number that cannot be compared to anything. Repeating it bare recreates the problem the data exists to fix.

The dataset will be refreshed as volume grows. The figures above carry the date they were pulled so an older citation stays checkable rather than silently going stale.

How to use these benchmarks on your own campaigns

Four comparisons that produce an action rather than a feeling.

1. Recalculate your own rate the same way. Unique human replies divided by unique prospects contacted, auto-replies stripped out. Most senders have never done this and are comparing an inflated internal number against someone else's inflated published one.

2. Compare against your industry row, not the overall average. A 0.9% human reply rate is below average overall and well above average for SaaS. Benchmarking against the wrong row is how a working campaign gets cancelled.

3. If you are under 1% per prospect, look at the list first. That is the level at which targeting is almost always the constraint. Rewriting copy at that point is the most common wasted fortnight in outbound.

4. Judge on positive replies, not replies. One in seven human replies is positive on average. A campaign with a high reply rate and a low positive share has usually found the wrong people and annoyed them efficiently.