Guides

Cold email subject lines and opening lines: policy rules

How to write plain, accurate cold email subject lines and first sentences, and which wording rules to record with an approved policy.

A subject line has one job: to say truthfully what the email is about. The first sentence has one job too: to show the reader why this email reached them and not someone else. Weak copy often fails at one of those two jobs. It promises something the body does not deliver, or it opens with a line that could be sent to anyone.

This guide covers how to write both lines plainly, which wording rules to record alongside an approved policy, and what an approver should check before anything goes out. It gives no magic phrases. It gives rules you can write down, check and audit.

Who this is for

This guide is for three groups:

  • Founders and sales leads who write or approve outbound copy and want a short list of rules that every message follows.
  • Marketers and operators who run cold email for a team and need something concrete to put next to a policy.
  • Approvers who sign off on what goes out and want a review list rather than a feeling.

If you use Outlevo, one fact shapes how you work. Nothing is sent outside a policy that a person approved [[1]](#sources). Record your wording rules alongside that policy, so the approver reviews them together with who can be contacted and how often. If you write email by hand, the same rules still work as a team checklist.

What the law says about accuracy

Start with the one rule that is not a style choice. The FTC's CAN-SPAM guidance says the subject line must accurately reflect the content of the message [[2]](#sources). A common assumption is that business email is exempt. The same guide says: "The law makes no exception for business-to-business email." [[3]](#sources)

So accuracy applies to B2B cold email. A subject line that suggests a reply to a thread that does not exist, or an invoice that is not an invoice, does not match the content of the message. Write it out of your copy for that reason alone, before you consider how it performs.

Whether you may email a particular person at all is a separate question, and this guide does not answer it. The UK's ICO says: "You can email or text any corporate body (a company, Scottish partnership, limited liability partnership or government body)." [[4]](#sources) GDPR Recital 47 says processing personal data for direct marketing "may be regarded as carried out for a legitimate interest." [[5]](#sources) Those sources cover who you may contact. The wording of the message still has to be accurate whichever rule applies. Have someone who knows the rules for your situation check the permission question.

Four principles for both lines

1. Say what the email is about

A subject line should name the topic in the words the recipient would use for it. If the email is about a change on their pricing page, say that. If it is about a role they are hiring for, say that. If you cannot name the topic in a few plain words, the email may not have a clear point yet.

2. Name the specific change or problem

Generic openers say nothing about the reader. Specific ones name one change or one problem that you can point to. The point is not to flatter the reader. It is to show that the message was sent for a reason.

"Specific" has to mean something you can back up. In Outlevo, each qualified account gets a Readiness Card with fit, intent and data quality scores, and a card needs at least three observations, each with a source and a date [[6]](#sources). An opener that names a specific change can be tied to one of those observations. If you cannot point to the dated observation behind a sentence, do not write the sentence.

3. Match the line to the recipient's role

A change that matters to a head of finance may mean nothing to a head of operations at the same company. Write the opener for the role you are contacting. Write down which opener angle applies to which role, instead of leaving it to each writer. For more on the trade-off between writing by hand and writing from rules, see cold email personalization.

4. Do not present a guess as a fact

Outlevo labels guesses as hypotheses and never presents them as findings [[6]](#sources). Facts read from a website are kept only when their quote is found on the page [[6]](#sources). Borrow both rules for the first sentence:

  • A site fact goes in only if you can quote it from the page.
  • An inference is phrased as an inference: a question, or an "if".
  • A number comes from the sender's own facts. Outlevo applies that rule to ads: every number in an ad comes from the customer's own facts [[7]](#sources). It works for email openers as well.

Before and after examples

These examples are illustrations. Square brackets mark the parts you would fill from your own records. Each "after" assumes you hold a dated observation behind the claim.

Subject line: vague to specific

  • Before: Quick question
  • After: Your new returns page

The "before" says nothing, so it cannot be inaccurate, but it also gives no reason to open. The "after" names the topic. It is only acceptable if a dated observation of that page exists.

Subject line: pretending to be a thread

  • Before: Re: our call
  • After: Support lead role on your careers page

The "before" implies a prior conversation that did not happen, which fails the accuracy rule above. The "after" says what the email is about and where you saw it.

Opening line: filler to a named change

  • Before: I hope this email finds you well. I came across [Company] and was impressed by your growth.
  • After: Your careers page lists an open role for a support lead. [One sentence on what you do for teams in that position.]

The "before" could go to any company. The "after" can be traced to something on the page, and the claim is small enough to check.

Opening line: inference stated as fact

  • Before: You're clearly struggling with slow onboarding.
  • After: If onboarding is a bottleneck for your customer success team, this may be relevant. If it isn't, ignore this.

You have not observed the struggle. The "after" admits that and lets the reader decide.

Opening line: one line for every role

  • Before (sent to both a finance lead and an operations lead): We help companies like yours cut costs.
  • After, finance: [Company] changed its pricing page in [month]. If invoicing now has more plans to reconcile, here is how we handle it.
  • After, operations: [Company] is hiring for [role]. If you are building out that team, here is what we set up first.

Both "after" versions rest on one dated observation each, and each names a different change for a different job.

Steps

Use these steps to write the lines, then to record the rules with your policy.

  1. List the observations you hold for the account. Use only items with a source and a date. If you use Outlevo, these are the observations on the Readiness Card [[6]](#sources).
  2. Pick one observation that matters to the recipient's role. If none does, the account is not ready for a message to that role.
  3. Write the subject line first, from the topic. Name the change or problem in plain words. Then check it against the body: does the body deliver what the subject says?
  4. Write the first sentence as one claim. Make it one observation, with its quote or date available. Do not stack three compliments or guesses in front of the point.
  5. Mark anything you inferred. Rewrite it as a question or an "if". Never state it as a finding.
  6. Cut anything you cannot source. This includes numbers, praise and claims about the reader's feelings.
  7. Set the length limits. This guide gives no number for subject lines or first sentences. Choose a maximum for each, based on your own sends, and record the number where the approver will see it. Some channels carry a hard limit already. The LinkedIn invitation note is at most 300 characters [[8]](#sources), so note channel-specific limits too.
  8. Record the wording rules next to the approved policy. The next section lists them.
  9. Test one change at a time. Change the subject line or the opener, not both. For testing at low weekly volume, read how to A/B test cold email at low weekly volume before drawing conclusions.

Wording rules to record with the policy

An Outlevo policy sets who can be contacted, on which channels, how often and with what budget [[1]](#sources). Keep your wording rules in a short section next to it, so the person who approves the policy approves the wording rules in the same review. Every external action is checked against the approved policy before it runs and recorded in a ledger [[1]](#sources), so an approver can audit what was actually sent. Suggested rules:

  • Accuracy. The subject line must describe what the body contains. No fake replies, forwards, invoices or meetings.
  • Sourced first line. The first sentence names one observation that has a source and a date. A fact read from a site needs a quote from the page.
  • Hypotheses as hypotheses. Inferences are phrased as questions or conditions, never as findings.
  • Numbers. Any number in the opener comes from the sender's own facts or a dated observation.
  • Role fit. Each role has its own opener angle. A message for one role is not reused for another.
  • Length. A maximum for the subject line and for the first sentence, with the number chosen by your team.
  • Channel limits. Hard limits per channel, such as the LinkedIn note limit above [[8]](#sources).
  • No false familiarity. No claims of a past conversation, mutual connection or event unless it is recorded.
  • Opt-out wording. A plain way to decline, kept consistent across messages.

Outlevo writes and schedules the emails, and Smartlead sends them from the customer's mailboxes [[9]](#sources). Settle the wording rules before the campaign is approved, not after messages reach the mailbox.

If a rule turns out to be wrong, say a subject rule that makes messages unclear, you can stop sending without taking everything down. Kill switches pause a campaign, a channel or the whole workspace at once [[1]](#sources). A sensible order is to pause, change the rule, have a person approve the change, then resume.

What to avoid

  • Fake threads and urgency. "Re:", "Fwd:", "Final notice" and similar prefixes imply something that is not true.
  • Generic praise. "Impressed by your growth" is not an observation.
  • Unquoted site claims. If the quote is not on the page, do not state it as a fact from the page.
  • Stated guesses. "You're struggling with..." is a hypothesis dressed as a finding.
  • Invented numbers. A statistic in the first line that you cannot trace to your own facts is a liability.
  • One line for every role. It reads as sent to a list, because it was.
  • Claims about wording effects. Do not write rules on the belief that a certain subject format always lifts opens or replies. No such effect is established here, so test it on your own sends.

Reply rates in context

Reply rates for cold email are low, and published figures vary. One guide from SmartReach says the average reply rate slid from 8.5% in 2019 to about 5% in 2025 and roughly 3.4% in 2026, without naming a single dataset [[10]](#sources). Treat that as directional only. Smartlead's State of Cold Email 2026 covers more than 850 million emails sent from January to June 2026 and puts the median reply rate at 0.74% per contact, the top quarter of senders at 1.43% and the top tenth at 2.63% [[11]](#sources). The two sets of figures measure different things: one is an average from an unnamed dataset, the other is a median per contact. They are not a contradiction. Neither tells you what a given subject line will do.

What to check

Give the approver this list before a campaign goes live. Check a sample of the actual messages, not only the template.

  1. Does the subject line accurately reflect the content of the message? Read the subject and the body side by side. [[2]](#sources)
  2. Does the subject avoid fake replies, forwards or prior contact?
  3. Does the first sentence name one observation with a source and a date?
  4. If the first sentence states a site fact, is the quote on the page?
  5. Is every inference written as a question or a condition?
  6. Does every number come from your own facts or a dated observation?
  7. Does the opener fit the recipient's role, and is it different from the openers for other roles?
  8. Is the subject line within the length limit you set, and the first sentence within its limit?
  9. Does each channel's own limit hold, for example the LinkedIn note?
  10. Is it recorded who approved these rules and when?
  11. After sending, does the ledger show what was sent, so you can compare it with what was approved?

One more pattern is worth borrowing. Another vendor, 11x, says every message can be reviewed, edited or overridden before it goes out, or it runs on autopilot [[12]](#sources). Whichever tool you use, decide in advance which messages need a human read and which run from the rules, and write that down with your wording rules.

If sending infrastructure is part of what you are reviewing, these pages cover it: email deliverability tools, email verification tools and secondary vs primary domain for cold email. Good copy does not fix a bad address list or a damaged sending domain, so review those separately.

Next step

Write down the rules above as a one-page wording section kept next to your policy, with your own length limits, and have a named person approve it. Then pull a small sample of messages and run the review list against them.

If you want a starting point for who to write to, the free ICP assessment reads up to 5 public pages of a website [[13]](#sources). It returns facts labelled as verified observations or AI inferences, plus up to 5 experimental segment hypotheses [[13]](#sources). When you draft openers from the results, treat the verified observations as candidates for a first sentence and keep the AI inferences as questions. Outlevo asks which countries you sell into and never proposes markets [[14]](#sources), so that decision stays with you. For how approval, the ledger and kill switches fit together, see how Outlevo works.

Sources

  1. How Outlevo works
  2. FTC, CAN-SPAM Act: a compliance guide for business
  3. FTC: CAN-SPAM Act compliance guide for business
  4. ICO: Electronic mail marketing (PECR)
  5. GDPR Recital 47: legitimate interests of the controller
  6. Readiness Cards
  7. Reddit Ads integration
  8. LinkedIn integration
  9. Smartlead integration
  10. SmartReach, cold email statistics guide (2026)
  11. Smartlead, The State of Cold Email 2026: reply rate per contact (top quartile; January to June 2026, more than 850 million emails)
  12. 11x AI personalization
  13. Free ICP assessment
  14. ICP from evidence

Related

See what Outlevo finds on your own website

Free, no sign-up, about a minute.