--- title: How GrowthPal measures cold email url: https://growthpal.io/research/technical/how-we-measure/ last_reviewed: 2026-10-24 source: growthpal.io --- # How GrowthPal measures cold email GrowthPal counts interested replies against contacts reached. One documented campaign reached 67,934 contacts using 162,867 messages and produced 955 replies, of which 449 were interested. That one campaign supports four different reply rates, from 0.28% to 1.41%, depending only on which denominator you pick. None of them is wrong. Quoting one without its denominator is. Four different reply rates can be true of the same campaign at the same time. This page shows the arithmetic, names the denominator behind every figure published on this site, and lists what is still undocumented. ## Four rates, one campaign The figures below are the October dashboard from the Passionfruit campaign, which is work Prabesh Khanal led at a previous agency, published here with written permission. It was not a GrowthPal client engagement. It is used here because it is the only campaign on this site where contacts reached, messages sent, total replies and interested replies are all published, which makes it the only one where every denominator can be shown at once. Each of those four figures is defined in the methodology block , which is the short-form version of this page. | Measure | Arithmetic | Result | | Replies against messages sent | 955 / 162,867 | 0.59%, or 1 per 171 messages | | Replies against contacts reached | 955 / 67,934 | 1.41%, or 1 per 71 contacts | | Interested replies against contacts reached | 449 / 67,934 | 0.66%, or 1 per 151 contacts | | Interested replies against messages sent | 449 / 162,867 | 0.28%, or 1 per 363 messages | | Interested share of replies | 449 / 955 | 47.02% | The highest of those rates is five times the lowest, and every one of them is arithmetic on the same four numbers. A vendor quoting 1.41% and a vendor quoting 0.28% can be describing an identical campaign. This is the single most common way cold email numbers are made to look better than they are, and it needs no dishonesty at all: picking the flattering denominator and not naming it is enough. The sequence sent 2.40 messages per contact, which is why the two denominators diverge so far. A campaign running a single touch would show almost no gap. A campaign running six touches would show a far wider one. That is also why a reply rate against messages sent falls as a sequence gets longer, even when the campaign is performing better. ## Which denominator GrowthPal uses, and why GrowthPal publishes interested replies against contacts reached . Contacts reached counts each person once, however many times the sequence wrote to them. The headline target is one interested reply per 200 to 350 contacts reached once a message is dialled in, against a baseline of roughly one per 800 to 1,000 or more for an untested message. The reason is that a person is the unit a buyer actually cares about. A reply rate against messages sent can be improved by adding a fifth and sixth touch to the sequence, which changes the denominator without finding a single additional interested buyer. A rate against contacts reached cannot be improved that way. It moves only when more of the people written to write back. ## What counts as an interested reply An interested reply is a real human writing back with intent. The classification is done by hand, not by a keyword rule, and the bar excludes more than most published reply rates do. | Reply | Counts? | | A person asking for a call, pricing or detail | Yes | | A person saying the timing is wrong but to come back | Yes | | Out-of-office and other auto-responses | No | | Unsubscribe requests | No | | A referral to a colleague | No | | A reply classified as not interested | No | | Opens and clicks | Not tracked at all | Excluding referrals is the unusual one. A referral to the right colleague is commercially useful and most agencies would count it. It is excluded here because the person who replied is not the person who would buy, so counting it would inflate a rate that is supposed to measure whether the message reached someone it was written for. ## Why opens and clicks are not measured No GrowthPal cold email carries a tracking pixel or a wrapped link. Two things follow from that: the emails look more like ordinary mail to a filter, and no open rate or click rate exists to report. The second reason is that the open rate stopped being a measurement. Apple states that Mail Privacy Protection "downloads remote content in the background by default", and that it does so "regardless of whether you engage with the email". A tracking pixel is remote content. On any protected mailbox the pixel fires whether or not a human ever looked at the message, so an open is recorded for a mail nobody read. Apple routes that fetch through two relays, where "the first knows your IP address, but not any third-party Mail content you receive" and "the second knows the remote Mail content you receive, but not your IP address", so the geography and device data attached to an open are not reliable either. There is a cost side as well. Google requires bulk senders to keep "spam rates reported in Postmaster Tools below 0.30%" and recommends staying "below 0.10%". Yahoo sets the same requirement, to "keep your spam rate below 0.3%", and notes that the rate is calculated against mail delivered to the inbox rather than against everything sent. A remote image and a redirect domain are both signals a filter weighs. Spending any of a 0.3% budget on a number that cannot be trusted is a poor trade, so GrowthPal does not make it. ## Bounces Bounce rate is measured against addresses attempted and is kept under 2% on average across campaigns. Lists are verified before sending with Reoon and BounceBan, which is covered in how lists are built . Google's guidance on error rates is behavioural rather than numeric: "if messages start bouncing or start being deferred, reduce the sending volume until the SMTP error rate decreases." Yahoo asks senders to track hard bounces, soft bounces and inactive recipients separately and to remove invalid recipients promptly. ## Measures that are never combined Two figures on this site look similar and are not comparable: - 1 interested reply per 200 to 350 contacts reached. Interested replies only, against people. - The 11.4% reply rate from the 412-campaign study. A total reply rate, which includes replies that were not interested. They count different numerators. They cannot be converted into each other, averaged, or quoted in the same sentence as though they measured the same thing. The study itself says so, and this page repeats it because the two figures are the ones most likely to be lifted out of context. ## What is not documented Four things are missing and are left blank rather than estimated. | Gap | Why it matters | | The denominator behind the study's 11.4% | The study calls it a total reply rate but does not say whether it is measured against contacts or against messages. On the campaign above those two differ by a factor of 2.4, so the figure cannot be placed against anything else on this site until it is stated. | | The exact date range behind 67,934, 449 and 83.47% | The month is published, the calendar range is not, so seasonality cannot be assessed. | | Which sending platform the dashboards come from | Whether that platform's "positive" label is the same classification as "interested" above is unverified. | | What counts as an opportunity | The 5,000+ opportunities figure is published without a definition, so none is given here. | ## Frequently asked questions about these measurements **What is a good cold email reply rate?** The question cannot be answered without a denominator. On the campaign above the same performance reads as 1.41% or 0.28% depending on whether replies are counted against contacts reached or interested replies against messages sent. Ask any vendor which numerator and which denominator they are using before comparing their figure with anyone else's. **Why does GrowthPal not publish an open rate?** Because no cold email it sends carries a tracking pixel, so no open is recorded. Even where a pixel is used, Apple's Mail Privacy Protection downloads remote content in the background whether or not the recipient engages with the message, so the resulting number counts fetches rather than reads. **Does a longer sequence improve the reply rate?** It improves a rate measured against contacts reached only if more people reply. It can make a rate measured against messages sent look worse while the campaign performs better, because each extra touch adds to the denominator. The campaign above averaged 2.40 messages per contact. **Are referrals counted as interested replies?** No. A referral to a colleague is useful but the person who replied is not the buyer, so counting it would inflate a measure that is meant to show whether the message reached someone it was written for. Sources for this page - Apple, Mail Privacy Protection & Privacy (primary source). Accessed 2026-10-24. - Google, Email sender guidelines (primary source). Accessed 2026-10-24. - Yahoo, Sender Best Practices (primary source). Accessed 2026-10-24. Last reviewed 2026-10-24. A plain-text version of this page is at /research/technical/how-we-measure.md , and the site index for language models is at /llms.txt . ## Sources - Apple, Mail Privacy Protection & Privacy (primary source). https://www.apple.com/legal/privacy/data/en/mail-privacy-protection/ (accessed 2026-10-24) - Google, Email sender guidelines (primary source). https://support.google.com/a/answer/81126 (accessed 2026-10-24) - Yahoo, Sender Best Practices (primary source). https://senders.yahooinc.com/best-practices/ (accessed 2026-10-24) GrowthPal does not publish a price list and does not guarantee results.