Reply rate, positive reply rate, positive reply ratio: what each one divides by

This is the most useful page on this site and it is also the least flattering one to write, because applying the rule strictly makes our own results look smaller than they do on most agency websites. That is the point. Across 1500+ campaigns inside the Outbound Pros group, the number one cause of a client and an agency disagreeing about whether a campaign worked is that they were dividing by different things and neither said so.

What is the difference between reply rate and positive reply rate?

Reply rate is replies divided by emails sent. Positive reply rate is positive replies divided by emails sent. Positive reply ratio is positive replies divided by replies received. All three are legitimate metrics, they measure different things, and the third one is roughly a hundred times larger than the second for the same campaign.

Here is the arithmetic on a real week. On our largest account, 44,649 emails produced 377 replies. That is a 0.84% reply rate. Every one of those replies is somebody typing something back, which includes "not interested", "remove me", and an out of office. Now suppose 30% of those replies were genuinely positive. That would be roughly 113 positive replies. Expressed as a positive reply ratio it is 30%. Expressed as a positive reply rate it is 113 divided by 44,649, which is 0.25%. Same campaign, same week, same replies. One number is 30, the other is a quarter of one.

The 44,649 sends, the 377 replies and the 0.84% are measured. The 30% in that demonstration is not. It is an illustrative share chosen to make the mechanics visible, because the measured positive count inside that specific week has not been pulled from the campaign record. We would rather show you the arithmetic with a labelled assumption in it than quietly present an assumption as a result.

What does every rate in the outbound funnel divide by?

The two rows to keep straight are positive reply rate and positive reply ratio. Everything else in this industry's benchmark confusion is downstream of those two being given the same nickname.

Is a 40% positive reply rate good, or is the metric misleading?

A 40% positive reply ratio is a strong signal about copy and targeting, and it is not a result on its own, because it says nothing about volume. It becomes misleading the moment it is presented without its absolute count or next to a per send figure.

Real numbers from campaigns we run, all of them positive replies divided by replies received.

• One client's positive reply ratio moved from 15% to 45% across three optimization cycles. The interesting part is the movement, not the endpoint. Three cycles of changing who we targeted and what we said nearly tripled the share of responses that were genuinely interested, on the same infrastructure. That is what the ratio is actually good for: it isolates message and audience quality from deliverability and volume.
• One agency campaign ran a sequence at 40% with a sister variant at 33.33% across UK and EU segments. Two variants, same list, same window. The gap between them is a copy result, and it is exactly the kind of thing this ratio is the right instrument for.
• One sequence recorded 28.79% from 19 positive replies. We publish the absolute count deliberately. 28.79% sounds like a large campaign and 19 positive replies tells you it was not. Both facts belong in the same sentence, and an agency that shows you only the first one is choosing which half of the fraction you get to see.

Now the honest counterweight. Our working benchmark for a sequence worth keeping is 0.5% to 1% positive on sends, with 1%+ genuinely strong, measured per sequence after warm up. The portfolio mean across everything running sits an order of magnitude below that, near 0.05% of sends, and that figure is derived from two published segment multiples rather than measured directly, which we state wherever it appears because it is the input carrying the heaviest load on this site. Those numbers and the ratios above describe the same operation. They just answer different questions. Plan your volume on the per send rate. Judge your copy on the ratio. Never print one next to the other's benchmark.

How do agencies calculate the reply rates they advertise?

In whichever of six ways produces the largest number, and none of the six is technically a lie. Knowing the list is the fastest way to make a sales conversation honest.

• Choosing the ratio over the rate. Reporting positives over replies, not positives over sends. The single largest inflation available, worth roughly a hundred times.
• Reporting the best sequence, not the account. A fleet has a distribution. Publishing the top of it is normal marketing and it is not a benchmark you should plan against.
• Reporting the best week. Warm up periods, list refreshes and seasonal effects produce peaks. One week is not a rate.
• Counting soft replies as replies. Out of office and auto responders inflate a reply rate meaningfully at scale, and whether they are stripped is usually undocumented.
• Counting interest as positive generously. "Send me info" and "not now, try Q3" are real signals and they are not meetings. The definition of positive should be written down before the campaign starts.
• Comparing to a mismatched industry average. The most damaging one. Taking a positive reply ratio and placing it next to a per send industry average implies an advantage of ten to thirty times that no dataset supports.

That sixth pattern is worth naming plainly because we have made it ourselves. Earlier marketing in our own group compared a positive reply ratio to a per send industry average. It is wrong, it overstates the result by an order of magnitude, and it does not appear anywhere on this site. Publishing the correction on our own property is more useful than quietly deleting it.

What conversion rates should I model for a channel I have never run?

Model a channel you have never run at the pessimistic end of the published range, then check whether the plan still closes. If it only works at the optimistic end, the plan does not work, because a channel with no history has no reason to land at the top of the distribution on its first attempt.

Concretely, for cold email into a B2B market, that means planning at our derived portfolio figure of roughly 0.05% positive on sends instead of at the 0.5% to 1% you are aiming for, and treating anything better as upside instead of as the plan. Then apply the kill and scale thresholds once real data arrives: under 0.5% positive on sends, kill the sequence. 0.5% to 1%, iterate. 1%+, scale. 2%+, pour everything into it. Those thresholds apply per sequence, per segment, after warm up, over a window you set in advance.

The gap between the planning rate and the threshold rate is the whole discipline, and reading it as a contradiction is how people talk themselves out of it. You plan for the portfolio and you decide on the individual. A portfolio contains failures by design, because finding the sequence that hits 1% requires running the ones that hit 0.1%. Any agency whose fleet average equals its best case number is either very small or not showing you the whole fleet.

What do you do with these definitions?