Community content. Review instructions before giving them to an AI agent — treat modules like open-source code.
Ad Copy Rules
Rules for writing paid ad copy that converts across Facebook, Google, and other placements: one promise per ad, benefit before feature, message match to the landing page, compliance basics, and single-variable testing.
Mby @markdownersPublished August 21, 2026 · ~4 min read
0 downloads · Used by 0 stacks
An ad has one job: get the right person to take the next step, not to explain the whole product. Every rule below follows from picking one promise for one audience and cutting everything that doesn't serve it — an ad trying to say three things says nothing memorably.
One promise, one audience
- Decide the single benefit and the single audience segment before writing a word of copy. An ad addressing "everyone who might want this" converts worse than an ad clearly speaking to one specific buyer's specific situation, even though the audience is smaller.
- If there are multiple genuinely different value propositions for different segments, write separate ads for each rather than one ad that tries to cover all of them — a combined ad reads as generic to every segment it's trying to reach.
Benefit before feature
- Lead with what the reader gets, not what the product does. "Get invoices paid 12 days faster" beats "Automated invoicing software with reminders" — the feature can appear as supporting proof after the benefit has already landed.
- Translate every feature into its outcome before it goes in the copy: not "256-bit encryption" but "your data is protected to bank-grade standards" — the reader is evaluating what changes for them, not cataloguing specs.
Concrete specificity over superlatives
- Replace unverifiable superlatives ("best," "#1," "revolutionary") with a specific, checkable claim ("used by 40,000 teams," "cuts setup time to under 10 minutes") — specific numbers read as credible; superlatives read as the default noise every ad uses and get discounted automatically.
- If a claim can't be made specific or backed up, cut it rather than keep it vague — vague claims do less persuasive work than concrete ones and add compliance risk for no benefit.
Headline formulas that survive
- Problem call-out: name the reader's specific pain directly ("Still tracking expenses in a spreadsheet?") — works because it self-selects the right audience before they even read further.
- Outcome plus timeframe: state the result and how fast ("Cut onboarding time in half within a week") — concrete and testable claims consistently outperform vague ones.
- Social proof as headline: lead with a number or a recognizable name ("Join 12,000 teams already using X") — borrows credibility before the reader has evaluated anything else.
- Avoid headline patterns that are pure curiosity with no information ("You won't believe this one trick") — they perform poorly in ad contexts specifically because readers expect ads to make a real claim, not tease one.
Funnel stage and message match
- Cold audience (haven't heard of the product): lead with the problem or the core benefit, keep the ask low-commitment (learn more, watch, read) — asking for a purchase decision from someone who doesn't know the brand yet wastes the impression.
- Warm/retargeting audience (already engaged): reference the specific thing they interacted with, address likely objections directly, and use a higher-commitment CTA (buy, sign up, book a call).
- Match the ad's exact language, imagery, and offer to what the landing page shows immediately on arrival — a mismatch between what the ad promised and what the page delivers (different headline, different offer, different tone) spikes bounce rate even when the underlying product is a good fit.
Compliance basics
- Never manufacture urgency that isn't real ("Only 2 left!" when there's no actual scarcity, a countdown timer that resets) — beyond the ethical problem, major ad platforms increasingly detect and penalize these patterns.
- Never make a claim the business can't substantiate if challenged (medical, financial, or guaranteed-outcome claims are the highest-risk categories) — an unsubstantiated claim risks both platform disapproval and legal exposure, not just a weak ad.
- State pricing, trial terms, and any auto-renewal clearly rather than burying them — platforms and regulators both treat hidden material terms as a compliance issue, not just a copywriting choice.
Length by placement
- Search ads (e.g. Google): headlines and descriptions are hard-truncated by character limits — write the most important word or phrase first in each field since truncation cuts from the end, and don't assume the full text will display.
- Social feed ads (e.g. Facebook/Instagram): primary text gets truncated after a few lines with a "see more" — front-load the hook exactly as in organic captions, and keep the total short since feed ads compete with organic content for the same fast scroll.
- Display/banner ads: assume only a headline and a few words of subtext will actually be read — cut every ad down to the single strongest line rather than trying to fit a full pitch into a small unit.
Testing discipline
- Change exactly one variable per test (headline, image, CTA, audience) and hold everything else constant — changing multiple elements at once makes it impossible to attribute a performance change to a specific cause.
- Let a test run long enough to reach a meaningful sample before calling a winner — early results on small ad spend are noisy and reversing course too fast wastes the learning phase most ad platforms need to optimize delivery.
- Keep a record of what was tested and the result, even for losing variants — a documented losing headline prevents re-testing the same idea again later under a different name.
Badge
Link back to this module from your own README.
[](https://markdowners.com/m/markdowners/ad-copy-rules)Discussions about this module
No discussions about this module yet.
Start a discussion
Comments (0)
Sign in to comment. Sign in
No comments yet. Be the first to add one.