The Golden Rule
Your subject line should look like something a colleague would send.
Would a colleague send: "Quick question about your marketing strategy"
Would a colleague send: "MASSIVE SAVINGS - ACT NOW!!!"
What Works
| Style | Example | Why It Works |
|---|---|---|
| Question | "Quick question about [X]" | Invites response |
| Direct | "[Company] + [Your Company]" | Clear purpose |
| Reference | "Following up on [topic]" | Implies relationship |
| Simple | "Hi [Name]" | Personal, human |
What to Avoid
| Style | Example | Why It Fails |
|---|---|---|
| ALL CAPS | "HUGE OPPORTUNITY" | Instant spam signal |
| Excessive punctuation | "Amazing!!!!" | Looks promotional |
| False urgency | "URGENT - Read Now" | Spammy pattern |
| Clickbait | "You won't believe this" | Trust destroyer |
| Overly salesy | "50% OFF TODAY ONLY" | Classic spam |
Character Length
| Length | Effect |
|---|---|
| Under 30 | Displays fully on mobile |
| 30-50 | Good balance |
| 50-70 | May get cut off |
| 70+ | Definitely truncated |
The First Name Debate
Using first name:
- Personalized: Yes
- Can backfire: If it's wrong
- Spam filters: Know this trick
Our take: First names work but aren't magic. A good subject line without a name beats a bad one with a name.
Preview Text Matters
The preview text (first line of your email) appears next to the subject in most email clients:
``
Subject: Quick question about your marketing
Preview: Hi [Name], I noticed that [Company]...
``
Make sure your first line continues the subject naturally.
Subject Line Formulas
Pattern 1: Question
"Question about [their specific challenge]"
Pattern 2: Mutual Connection
"[Connection name] suggested I reach out"
Pattern 3: Observation
"Noticed [something about their company]"
Pattern 4: Direct
"[Your Company] for [Their Company]"
Testing Subject Lines
With our multi-domain setup, you can test:
- Send Version A from domains 1-5
- Send Version B from domains 6-10
- Compare reply rates
- Scale the winner
Key Takeaway
Simple, clear subject lines that look human. Avoid anything you wouldn't send to a colleague. Test with real data, not assumptions.