Seven subject line rules agencies should enforce across client sequences

Subject lines are where client voice briefs fall apart first. One client wants title case because it looks professional. Another's founder writes everything lowercase. A new SDR on the account adds "Re:" to a first touch because a forum post said it lifts opens. Six weeks later nobody can say what the rules are, and the A/B tests are measuring opens that a privacy proxy fired.

Here are seven rules we would put in every client's sending policy. Some are about the law, some are about mailbox provider guidance, and some are about what the published data says. Where the data disagrees, we say so.

1. Never fake a thread

No "Re:" or "Fwd:" on a message that is not a reply or a forward. Google's email sender guidelines say exactly this: subject lines and other message elements should accurately represent the sender and the content, and you should not start subject lines with Re: or Fwd: unless the messages are actual replies or forwards.

It also tends to backfire with the people you most want to reach. A prospect who opens a fake reply and finds a cold pitch now knows the first thing you did was mislead them. Make this a hard QA reject, not a style note.

2. The subject has to match the body

The FTC's CAN-SPAM compliance guide lists "don't use deceptive subject lines" as one of the main requirements, and defines it simply: the subject line must accurately reflect the content of the message. The law covers business-to-business email too.

In practice, this rules out the clickbait patterns that creep into AI-written subject lines: "quick question" over a pitch with no question in it, "your account" when there is no account, a prospect's company name next to a claim that isn't about them. A quick test for the reviewer: if the prospect read only the subject and then the first line, would they feel misled?

3. Keep it to a few words

The length data is fairly consistent. Reply.io analyzed 712,000 first-step emails with unique subject lines and found 1 to 4 words got the highest open rates, and the average reply rate declined at 5 or more words. Gong's analysis of more than 85 million cold emails, as summarized by 30 Minutes to President's Club, also puts the ideal range at 1 to 4 words. ReachIQ's own dataset found 3 to 4 words did best on both opens and replies, with both falling off past seven words.

Set a ceiling in the client policy, something like six words, and let the write step work inside it. Short subjects also survive mobile truncation, which long ones do not.

4. No pitch in the subject

Gong's data, again via 30MPC, found "salesy techniques" reduced open rates by as much as 17.9%, and the advice that follows is to resist talking about your solution in the subject line. Use the prospect's world instead: an initiative, a problem, an outcome they care about.

For an agency this is also a client-isolation rule. Product names and offer language belong in the client's voice brief and the body copy. The subject line should read like something a colleague might send, not a campaign name.

5. Don't leave it blank

Blank subjects keep coming back as a hack. The data argues against them. Reply.io found emails with no subject line got almost 36% fewer opens than average. The Gong numbers cited by 30MPC went the other way on opens, a 30% lift, but replies fell 12%. Either way, the reply side looks worse. A blank subject is also harder to defend under rule 2: there is nothing for the body to match.

6. Set a casing rule per client, then test it on replies

This is where the published data disagrees with itself. Gong's data, per 30MPC, found all-lowercase subject lines had the highest open rates. ReachIQ found lowercase and sentence case close together, title case behind, and all caps far behind on both opens and replies. Reply.io found the opposite of Gong for its own sample: all lowercase produced 3.46% fewer opens and 11% fewer replies, while mostly capitalized words did best.

Two takeaways. First, never use all caps. Every dataset that tested it puts it last. Second, the rest is a per-client decision, and it should be tested on reply rate, not opens. Most of these studies lean heavily on open rate, which Apple Mail Privacy Protection and security scanners have made unreliable. A variant with fewer opens and more replies is the better variant. We wrote more on why open rate should leave client reports.

7. Every token needs a fallback, and the subject gets checked after rendering

Reply.io's study found subjects with two or three variables did better on opens than average, while subjects with a single variable had a reply rate 6.6% below average. It also noted that a relevant subject with no variables works fine for a narrow segment. Personalization in the subject is optional. A broken token is not.

A subject that ships as "{{Company}} hiring?" or "Hi ," is worse than a plain one. Require a fallback for every token used in a subject, and run QA on the rendered subject, not the template. If a company name comes from enrichment, check that it is the name the company actually uses, not the legal entity with "Inc." and a state suffix.

Putting it in the client policy

Write these into each client's sending policy as checks the QA step can run:

• Reject: fake Re:/Fwd:, all caps, blank subject, unrendered tokens
• Reject: subject that does not match the body
• Flag for review: more than six words, product or offer name in the subject
• Per client: casing rule and a short list of approved subject patterns
• Testing: variants compared on replies and booked conversations, never on opens alone

Display names follow similar logic. Google's guidelines say the display name should identify the sender and should not carry subject or message content, so do not let urgency words drift from the subject into the From field either.

Where GTM fits

GTM writes from prospect evidence in each client's voice and runs send, reply, and book in isolated client workspaces, with a compliance gate in the send path and human QA available before anything goes out. Subject line rules belong in the same per-client policy as voice and do-not-say lists, so one client's habits never leak into another's sequences.

If you run outbound for several clients and want those rules enforced per workspace, join the waitlist.

Sources

• Google Workspace Admin Help, "Email sender guidelines": https://support.google.com/a/answer/81126
• FTC, "CAN-SPAM Act: A Compliance Guide for Business": https://www.ftc.gov/business-guidance/resources/can-spam-act-compliance-guide-business
• Reply.io, "Reply Research: How to write better subject lines, according to data": https://reply.io/reply-research-subject-lines/
• 30 Minutes to President's Club, "4 data-backed subject lines to get your cold emails opened": https://www.30mpc.com/newsletter/4-data-backed-subject-lines-to-get-your-cold-emails-opened
• ReachIQ, "What millions of B2B emails reveal about subject lines": https://reachiq.ai/resources/blog/what-b2b-emails-tell-us-about-subject-lines/