Subject lines are the most obsessed-over part of cold email and one of the least important. A great subject line on a bad list still gets nothing. A plain one on the right list gets replies.
That said, there are a few things large datasets agree on, and they are cheap to apply. Here is what I would actually do, and where I would stop caring.
Keep it short
Saleshandy looked at about 53 million cold emails and found the 4 to 5 word range performed best. Leadhaste, citing Instantly's A/B tests, puts the sweet spot at roughly 25 to 45 characters.
There is a practical reason underneath the stats. A lot of your readers open mail on a phone, and many mobile mail apps cut a subject off somewhere around 33 to 43 characters. If the part that matters is at the end, it never gets read.
So write the subject like a note a colleague would send, with the useful word first.
Make it look like a person wrote it
The subject lines that work tend to look internal. Lowercase is fine. No emoji. No brackets. No "Quick question!!!".
A few shapes that read as human:
| Shape | Example |
|---|---|
| Their company plus a topic | "acme + outbound" |
| A plain question | "who runs outbound at acme?" |
| A reference to a real trigger | "your new head of sales role" |
None of those say anything clever. That is the point. You are trying to not look like a campaign.
What to avoid
Saleshandy's guidance, which matches what I have seen, is to avoid:
- Colons and quotation marks. They read as marketing.
- Fake "Re:" or "Fwd:" prefixes. They are an old trick, filters know it, and the person feels tricked when they open it.
- All caps and heavy punctuation. Same reason.
- Anything that overpromises. A subject that sells a result you cannot back up gets the spam complaint.
The fake thread trick is the one I would call out hardest. It might lift opens for a week. It also teaches the reader that you lie in line one.
Questions can help, but test it
One vendor dataset reported that question-style subject lines beat statements on reply rate. It is a single source, and vendors have an interest in the stories they publish, so I would not treat it as settled.
What I would do is run it on your own list. Two subject lines, same body, same list, split evenly. If the question version wins by a clear margin over a few hundred sends each, keep it. If not, move on.
One honest warning about open rates
Many sequencers now recommend turning open tracking off, because the tracking pixel can hurt deliverability and opens have become unreliable anyway. If you have it off, you cannot judge a subject line by opens at all.
In that case judge it the only way that counts: replies per hundred sends. It is slower to measure, but it is the real number. I wrote about how many sends you need before a result means anything in how many cold emails it takes to get one positive reply.
What matters more than the subject line
If replies are low, the subject line is rarely the cause. Check these first, in this order:
- Are the emails reaching the inbox at all? See why cold email lands in spam.
- Is the list right? See how I classify an ICP.
- Is the offer clear in the first two lines?
- Then the subject line.
Spend ten minutes on the subject, test one alternative, and go back to the list. That is where the replies are.