The subject line is the only part of a cold email most prospects ever read. If it doesn't earn the open, every other thing you wrote is dead. That puts an unfair amount of weight on a six-to-eight word string typed in thirty seconds at the top of the draft. This post is a working library of cold email subject lines that consistently earn the open in B2B outbound, grouped by what makes them work and paired with the failure mode each one corrects. It is written for SDRs, founders running founder-led sales, and anyone whose reply rate is being capped at the inbox preview pane.
Key takeaways
- Subject lines compete with the inbox, not with other cold emails. The bar is “does this look like work I should do?”
- Five categories reliably outperform: specificity, curiosity, mutual connection, signal-based, and question-form.
- Length matters less than people say. Three words can win. Eight can win. Salesy phrasing kills both.
- Title case (“Quick Question About Your Onboarding Flow”) reads like marketing. Sentence case (“quick question about your onboarding flow”) reads like a coworker.
- Test categories, not individual lines. A category that works for one ICP often fails for another.
What a subject line is actually competing against
Most cold-email advice frames the subject line as a contest between you and other cold senders. It isn't. Your subject line is competing with the prospect's internal Slack threads, their direct reports' updates, the calendar invite their boss just sent, and the legitimate vendor email about an invoice they need to approve. The prospect is not asking “is this the best cold email I've seen today.” They are asking “does this look like work I am responsible for.”
That reframing changes what good looks like. A subject line that screams “sales pitch” loses, because sales pitches are not work the prospect is responsible for. A subject line that looks like an internal note about a specific operational thing the prospect owns wins, because that is exactly the kind of message they are scanning for. The job of the subject line is to look like one of those, while being honest about the fact that you are a stranger.
Five categories of subject lines that work
The categories below cover the bulk of what consistently outperforms. None of them are tricks. Each one corrects a specific failure mode in default outbound writing.
Specificity-based subject lines
The single most reliable category. Specificity signals you did the work; vague phrasing signals a list import. The win condition is referencing something only a human paying attention would write.
- question on your post about renewal motion
- your Q3 hiring page (head of partnerships)
- idea for the {Acme} onboarding step 3
- noticed your team just doubled in {Boston}
- about your {podcast episode} with {guest}
- {first name} — the part about {specific topic}
- follow up on your {conference} talk
- your case study with {customer} (one question)
Each of these forces you to know something specific. That is the point. If you can't fill the brackets without making something up, you don't have the specificity yet, and the email will fail no matter what subject line you put on it.
Common mistake: fake-specific
“quick question about your growth” is not specific. Every B2B company on earth is working on growth. Specificity means a noun the prospect could not confuse with another company's situation. “your Series B from April” is specific. “your funding” is not.
Curiosity-based subject lines
Curiosity works when there is a real payoff in the body. It dies fast if the body is generic. Use this category sparingly and never combine it with a salesy opener, because the gap between curiosity and pitch makes the prospect feel tricked.
- three things I noticed
- your {landing page} — small thing
- weird question
- am I reading this right
- this might be wrong
- not a fit, probably
- {first name}, quick read
- worth thirty seconds?
“Not a fit, probably” is a counterintuitive winner because it removes the implied sales pressure. The prospect opens the email because the framing itself is unusual for outbound. If you use it, the body has to actually examine whether you're a fit, not pivot into a pitch — otherwise you've burned trust on the second line.
Mutual connection subject lines
The strongest opener if you actually have one. The weakest opener if you fake it. Use mutual-connection lines only when the connection is real and the person whose name you're using has either consented or is genuinely close enough to the prospect that the reference passes a sniff test.
- {mutual} suggested I email
- {mutual} mentioned you
- via {mutual}
- {mutual} thought we should talk
- following up from {event}
- met {mutual} at {event} — one question
- your team works with {mutual company}
- via {LinkedIn group / community}
Signal-based subject lines
A signal is something that just happened to the prospect or their company. New funding, a hiring spike, a launch, a reorg, a podcast appearance, a public RFP. Signal-based subjects work because they reference something the prospect is currently thinking about. The decay is fast — a signal-based subject sent four weeks after the signal feels stale. Conversation hooks exist for exactly this category.
- congrats on the {Series B}
- noticed you're hiring {role}
- your {product} launch — one question
- caught your post on {topic}
- the {Acme} integration announcement
- your new {office / market}
- about your {partnership with X}
- caught the {SOC 2} announcement
Worked example: turning a signal into a subject line
Signal: company posted a new SDR job last week. Bad version: “quick question about your hiring.” Better: “your SDR hire (one question).” Best: “your SDR hire — re: the AE hand-off.” The best version both names the signal and previews the body's point. The prospect can guess what the email is about before opening it, which paradoxically makes them more likely to open it.
Question-form subject lines
Questions outperform statements when the question is genuinely answerable in thirty seconds. They underperform when they read as rhetorical or as a setup for a pitch.
- does {Acme} still use {tool}?
- are you the right person for {topic}?
- who owns {function} now?
- still hiring for {role}?
- still on {competitor product}?
- did you ever launch {thing}?
- worth a fifteen-minute call?
- open to one suggestion?
What kills subject lines
Some patterns are so worn that a prospect's eye skips them automatically. They aren't hypothetically bad. They are statistically worse than almost anything else you could write.
- “Quick question” by itself. The prospect knows it isn't. Add a noun.
- “Re: ” on a first email. Reads as a trick. The prospect notices.
- All caps or excessive punctuation. Triggers spam filters and human pattern-matching simultaneously.
- Buzzwords: synergy, leverage, unlock, supercharge, transform. None of these are how a coworker would write.
- Title Case Like A Marketing Headline. Internal email is sentence case. Match the format your prospect's coworkers use.
- Generic personalization fields. “Hi {first name}” that didn't merge. Easier to omit the name than to risk a broken merge.
How to actually test subject lines
Most teams test subject lines wrong. They run two lines against each other on a list of 200 and call a winner based on a 4 percent versus 6 percent open rate. At that volume, the result is noise. A real subject-line test needs at minimum a few hundred opens per variant before the difference clears the noise floor, which usually means a few thousand sends per variant.
The better unit of testing is the category, not the line. Send a hundred specific subject lines and a hundred curiosity ones to similar segments, compare the distributions, and let the winning category dictate where you spend writing time next month. Within a category, line-level differences usually wash out.
How AI changes the subject line workflow
The bottleneck on writing good subject lines was always research, not phrasing. Anyone can write “congrats on the Series B (one question)” if they already know the company raised a Series B last week. The job that took ten minutes per prospect was reading the company's press releases, scrolling their LinkedIn, finding the post worth referencing.
That part is now mostly automatable. Prsona's cold email generatorreads a LinkedIn profile, surfaces the specific signal worth referencing, and drafts an email and subject line in your team's brand voice. The writer becomes an editor, which is a faster job. The point isn't that AI writes better subject lines — it's that AI compresses the research time so a rep can ship fifty specific subject lines in the time it used to take to write five.
One last thing on length
People obsess over subject line length. The data is messier than the rule-of-thumb posts suggest. Three-word subjects can outperform eight-word ones; eight can outperform three. What predicts success is whether the line reads like internal email or like marketing. A long subject line written like a Slack message often beats a short one written like a banner ad. Pick length to match the voice you want, not to hit a magic word count.
Want to see this in practice?
Try Prsona free — open a LinkedIn profile, click the extension, and the subject line you get is built from the specific signal on that profile. No copy-paste templating, no generic merge tags. Solo plan is free, 10 lifetime credits, no card required.