Most subject line advice treats open rate as the finish line. It isn't. It's a checkpoint, and one that's gotten noisier since Apple Mail Privacy Protection started pre-loading tracking pixels on a large share of B2B opens, inflating the number regardless of whether a human ever looked at the message. If you're optimizing subject lines purely for opens, you can win the wrong game.

Gong's analysis of 85 million cold emails is the most useful dataset here precisely because it tracks both sides of the funnel. Subject lines that read as obviously salesy saw open rates drop as much as 17.9% compared to lines phrased like an internal note, and all-lowercase subject lines beat Title Case and Sentence case consistently across the sample. That's a cheap, immediate fix: stop capitalizing your subject lines like a press release.

Where the data splits: opens vs. replies

The harder lesson from Gong's data is that some tactics that lift opens actively hurt what you're actually selling for. Numbers in a subject line are the clearest case. Woodpecker's benchmark data, drawn from more than 20 million cold emails, found that including numbers in a subject line lifted open rates by up to 113%. Gong's 85-million-email study looked at the same tactic and explicitly advises against it, because in enterprise sales a number in the subject line reads as a sales pitch and drags down reply and meeting conversion even as it drives curiosity clicks.

That's not a contradiction to resolve, it's a signal to act on. If your goal is inbox visibility for a top-of-funnel nurture, a number might earn you the open. If your goal is a reply from a VP of Sales, skip it.

The emoji question has no single right answer

Reply.io's study of 2 million emails found a subject-line emoji lifted open rate to 37.8% versus 30.9% without one, a 22% relative gain. Lemlist's analysis of millions of cold sends found the opposite: emojis cut open rates by 6% and reply rates by 13%. Both numbers are real. The difference is almost certainly audience. Reply.io's sample skews toward broader SMB and agency outreach where a visual marker stands out in a cluttered inbox. Lemlist's sample leans more enterprise, where an emoji in a subject line from an unknown sender reads as consumer marketing, not a peer note, and gets treated accordingly by both spam filters and skeptical executives.

The practical takeaway: if you're selling into enterprise, finance, legal, or any buyer group that treats their inbox like a compliance queue, leave the emoji out. If you're selling into SMB, e-commerce, or more casual verticals, it's worth an A/B test rather than a blanket rule.

Length: two sweet spots, two different jobs

Subject line length isn't a single optimal number, it's two optimal numbers depending on what you're optimizing for. Gong's data shows ultra-short subject lines in the 1-to-4-word range consistently win on raw open rate, likely because they pass what amounts to a subconscious colleague test: in a mobile inbox preview, four words looks like an internal message, not a vendor pitch.

A separate cluster of data points to a different sweet spot for context and conversion. Overloop's cold email benchmark writeup and a 2024 study of 5.5 million B2B emails run by Belkins in partnership with Reply.io both land on 36 to 50 characters, roughly 6 to 10 words, as the range that best balances brevity with enough context to qualify the recipient before they open. That extra handful of words gives you room to name a company initiative, a mutual contact, or a specific trigger, which reduces the odds the recipient opens the email, doesn't recognize why they're being contacted, and archives it without reading further.

Worth flagging directly: a widely repeated claim attributes a "36 to 50 character sweet spot" finding to a Backlinko study of 12 million emails. That Backlinko/Pitchbox study is real, but it measured link-building outreach, not B2B sales cold email, and its actual headline finding was that longer subject lines outperformed shorter ones in that context. It doesn't support the sweet-spot claim as commonly cited, so we're leaving it out here and sticking with Overloop and Belkins, which measured the right kind of email.

The gap between open rate and reply rate is the real story

Gong's data includes one figure that should reframe how you read every open-rate stat in this space: empty subject lines boosted open rates by roughly 30% in their dataset, while cutting reply rates by 12%. An empty subject line is maximally curious and minimally useful. It proves that open rate, on its own, can reward exactly the wrong behavior.

If you're running cold sequences and only tracking opens, you're one clever but hollow subject line away from a vanity metric getting worse instead of better. Reply rate and meeting-booked rate are the numbers that tell you whether the subject line did its job, which is getting a real read from the right person, not just a pixel fire from an Apple privacy setting.

A short checklist

  • Default to lowercase, 1-to-4 words if your priority is opens; 6-to-10 words (36-50 characters) if your priority is qualified replies.
  • Skip numbers and emojis for enterprise, finance, legal, and other formal buyer segments. Test them cautiously for SMB and DTC.
  • Retire bare "quick question" openers. Longitudinal data from Woodpecker shows these generic phrases now register as automated outreach precisely because AI tools overused them, and they need a specific company detail attached to still work.
  • Judge subject line performance by downstream replies and booked meetings, not opens alone.

None of this is complicated once you separate the two goals, but it's easy to get backwards when you're staring at an open-rate dashboard under deadline pressure. Running your subject lines and full sequence through a second pair of eyes, or a grading tool built for exactly this kind of contradiction, catches the cases where a tactic that looks good on one metric is quietly working against you on the one that matters.