Email Marketing Copywriting

How to Write Email Subject Lines That Get Opened

Length limits by client, the four subject line jobs, and why the highest open rate is often the wrong thing to optimise for.

4 min read 2 of 10 in this topic Updated August 2026

On this page

The short version

  • Truncation is the constraint nobody designs around. Roughly the first thirty-five characters survive on a phone, which is where most of your list reads.
  • The highest open rate is not the target. A subject that overpromises buys opens from people who then do not click, and teaches the rest of the list to trust you less.
  • Test the angle, not the wording. Two phrasings of the same promise rarely differ enough to resolve on any list under six figures.

A subject line has one job and one constraint. The job is to make a specific promise the email then keeps. The constraint is that it is displayed inside a narrow column next to dozens of others, and most of it is cut off.

Nearly every subject line failure is one of those two things: a promise the body does not keep, or a promise that arrives after the cut.

Where the line gets cut

Display limits vary with window width, font size and client version, so treat these as the order of magnitude rather than exact figures — but the ranking is stable.

Approximate characters visible before truncation
Phone, portrait 35 chars; Phone, landscape 45 chars; Desktop webmail 60 chars; Desktop client, wide 75 charsPhone, portrait35 charsWhere the majority of opens happenPhone, landscape45 charsDesktop webmail60 charsDesktop client, wide75 charsThe width most subject lines are written in

Approximate and client-dependent. Send yourself a test and look at it on the device your list actually uses before trusting any published figure, including this one.

Four angles that keep working

There are more than four formulas in circulation, but most reduce to one of these. Pick by what the email actually contains rather than by which sounds most compelling — an angle the body does not support is the overpromise problem in a different coat.

The angles

Each suits a different kind of email, and misfires on the others
AngleExample shapeSuitsMisfires when
Specific outcome"Cut your bounce rate in half"How-to and tactical contentThe body is opinion, not method
Named question"Why did opens drop in June?"Diagnostic and analysis piecesYou do not answer it clearly
Concrete number"Nine things that trip spam filters"Lists and reference materialThe number is padded to reach it
Direct statement"The pricing page is live"Announcements and transactionalUsed on content that needs framing

Curiosity-only lines — the ones that withhold the subject entirely — belong on none of these rows. They lift opens and depress clicks, because the people they attract were promised nothing in particular.

Why the highest open rate is the wrong target

It is easy to raise open rate. Vague urgency, a fake reply prefix, a first name in the subject, a suggestion of something personal — all of these work, and all of them are borrowing against the next send.

The reader who opens on a promise you did not keep learns something about your from name, and applies it next time. The measurable version of this shows up as an open rate that holds while click rate declines, and eventually as an open rate that declines too.

Judge a subject line on clicks per delivered rather than opens per delivered. It is the same denominator, and it prices in whether the promise was kept.

Two lines for the same email

The email is a guide to fixing authentication records.

Higher opens

Promises nothing specific

  • "You need to see this"
  • "Quick question"
  • "Are you making this mistake?"
  • Attracts everyone, satisfies few

Higher clicks

Promises exactly what arrives

  • "Your SPF record is probably over the limit"
  • "DKIM in fifteen minutes"
  • "Three DNS records, written out"
  • Attracts fewer, satisfies most of them

Testing subject lines honestly

Most subject line tests cannot resolve. The effect you are looking for is a few percentage points, and detecting that reliably needs a sample most lists do not have. Running the test anyway produces a winner that is noise, and acting on noise is worse than not testing.

Two things make testing worthwhile at ordinary list sizes. Test angles rather than wordings, because the effect is larger. And accumulate: run the same angle comparison across ten sends and look at the aggregate, rather than declaring a winner from one.

Subject line check

  • The promise is contained in the first thirty-five characters
  • The body keeps that exact promise
  • It is not the same shape as the last four you sent
  • No manufactured urgency and no fake reply or forward prefix
  • It reads sensibly beside the preview text, not as a repeat of it
  • It has been looked at on a phone, not just in the composer

Frequently asked questions

Do emoji in subject lines help?

They help visibility in a crowded list and hurt when they are the only differentiation. They also render inconsistently and count against your character budget. Use sparingly, never as a substitute for a promise, and check how they display on the clients your list uses.

Should I put the recipient's first name in the subject?

It lifts opens slightly and is now common enough that many readers register it as automation. It also fails loudly when the data is missing or badly cased. If you use it, have a fallback and check what the fallback looks like.

Is all-lowercase or sentence case better?

Sentence case is safe. Lowercase reads as personal on lists that expect a personal sender and as careless on lists that do not, so it is a fit question rather than a best-practice one. Title Case reads as a headline, which suits announcements and not much else.