Glossary
Ad Hook
Ad hook examples are the first line, frame or sentence of an ad that stops the scroll, sorted into six mechanisms. Check each against Meta's personal-attribute rules before testing it.
By Sachit Sharma, CEO & Founder · Updated 24 Oct 2026
Also called Hook or Ad opening
On this page
What is an ad hook?
An ad hook is the first thing a viewer meets in an ad, built to make them stop scrolling.
Key takeaways
- Pick a hook by mechanism, then check it against Meta's rules.
- Hooks that use "you" are allowed unless they assert or imply a personal attribute, such as health, disability or financial status.
- A competitor's live ad shows the ad is live. It does not show the hook works, so copy the mechanism, not the words.
- Keep the body of the ad identical and change only the opening, so a result can be pinned on the hook.
Quick facts
- Where it lives
- Spoken line, on-screen text, first frame, or primary text
- Mechanisms
- Problem, number, contrast, proof, curiosity, offer
- Rules that apply
- Meta's Advertising Standards, reviewed before an ad runs
- Often confused with
- Hook rate, angle and headline
The same hook line can be a clear winner in a test and an ad Meta rejects before it runs. Most hook examples are copied without checking either.
Two questions decide whether a hook is worth using. Will Meta let the line run, or reject it at review? And what would a result from it prove about the hook itself, not the rest of the ad?
The first can be answered before you spend anything. Each ad is reviewed before it runs, and the advertiser is responsible for following the rules. Meta's ad rules bar ads from implying they know something about the viewer, and some common hook lines do exactly that. The second depends on treating each hook as one variable.
Below you will find a plain definition, the six mechanisms most examples reduce to, a policy check of 11 example hook lines, and a one-row ledger to fill in for each hook before it enters a test queue.
What good ad hook examples look like
A hook lives in one of four layers: the spoken line, on-screen text, the first frame, or the primary text (the caption above the ad) shown before "See more." Most examples reduce to six mechanisms.
- Problem: "Still chasing leads that never reply?"
- Number: "Three checks before you raise a budget"
- Contrast: "More ad spend won't fix a weak first line"
- Proof: "[Real number] [real customers] use it" (only with a number you can back)
- Curiosity: "What changes when you cut the intro from every video"
- Offer: "[Offer]. Ends [day]."
Any of these can sit in any layer. A problem hook can be a spoken line, a text overlay or the first sentence of the copy, and the viewer meets it differently in each.
Which ad hook examples Meta's rules will reject
The ones that imply you know something private about the viewer. Meta's Advertising Standards say "ads must not contain content that asserts or implies personal attributes," including health, disability and vulnerable financial status, and bar ads from implying knowledge of the medical information of a user or user's family. Ads can use "you" and "your" language, provided no personal attribute is attached (Meta, Privacy Violations and Personal Attributes, as of October 2026).
Advertisers must follow these standards, and each ad is reviewed before it runs (Meta, Advertising Standards). Only Meta decides, but here is a reading of 11 example hook lines:
- Rewrite (2): "Runners with bad knees, watch this." "Bloating? Try Amla." Both address a health condition the viewer is assumed to have.
- Borderline (3): "Why do you wake up tired after eight hours?" "Keratosis pilaris affects 40% of adults." "WARNING: Your daughter will get hooked." These touch a health state, a named condition or a family member. The family reading is a cautious one, since Meta's clause covers medical information of a user's family.
- Pass (6): "Are you still creating ad briefs in Google Docs?" "Still chasing leads that never reply?" "Struggling with slow Wi-Fi?" "If you're a founder who hates selling." "Your ROAS (return on ad spend) drops every Friday." "More cushioning won't save your knees."
That is a judgement, not Meta's decision: lines that address a health state the viewer is assumed to have are rewritten or flagged, while lines about a product, a workflow or a stated role pass. "More cushioning won't save your knees" passes because it describes the product and does not assume a condition in the viewer. The fix is to describe the product, not the viewer. "Running shoes with [measured cushioning feature]. Watch the test." keeps the mechanism and drops the assumption. Have someone read any rewrite against the policy before it runs.
How to sort a hook before you test it
Fill in one ledger row per hook. Each field has a pass or fail, so the row is a check, not a note.
The four proof levels:
- Seen: the ad is live. Fail: being live shows nothing about results.
- Lasting: the ad has run for weeks. Fail as proof, though it is a fair reason to shortlist.
- Claimed: the brand publishes results. Fail until you test it in your own account.
- Yours: tested in your own account. Pass; it is the only level that supports a spending decision.
The Facebook Ad Examples page gives more detail on the four levels.
- Field
- Hook, verbatim
- Fill in
- Exact words, or the first frame described
- Fail if
- You cannot write it down
- Field
- Layer
- Fill in
- Spoken, on-screen text, first frame, or primary text
- Fail if
- It spans two layers
- Field
- Claim about the viewer
- Fill in
- Nothing, a role, a situation, or a personal attribute such as health, disability or financial status
- Fail if
- It is a personal attribute
- Field
- Policy read
- Fill in
- Pass, borderline or rewrite
- Fail if
- Borderline or rewrite and still unrevised
- Field
- Mechanism
- Fill in
- One of the six above
- Fail if
- More than one
- Field
- Proof level
- Fill in
- Seen, Lasting, Claimed or Yours
- Fail if
- The only evidence is that the ad is live
- Field
- One variable
- Fill in
- What changes against the control (your current opening, which the new hook is compared with): the opening only
- Fail if
- The body also changes
What people confuse with an ad hook
Hook rate, angle and headline are the three most often mixed up with a hook. Hook rate is a measurement: the share of impressions that become 3-second video plays. A hook is the thing being measured. The hook rate page gives the formula.
An angle is the reason to buy, and one angle can open with several hooks. A headline is a separate text field in the ad. The phrase "an ad with three hooks" means three different openings on the same body, so you can compare them.
A high hook rate shows attention, not sales. The hook signals vs. conversion data page covers which metric decides what. Hold rate measures how many viewers keep watching after the hook, and its formula varies by dashboard, so see the hold rate page. For how to run the comparison and when to stop, see the Meta ad hook testing framework and when to stop testing an ad hook pages.
How Deepsolv handles it
We track competitor hooks, offers and formats, then combine them with ad-performance data, customer signals and past ad learnings into a ranked weekly test plan and execution-ready creative briefs. We treat a competitor's hook as a pattern to rewrite, never a line to copy.
The ledger above is the manual version of that sort.
Example
Worked example: "Bloating? Try Amla." Layer: on-screen text. Claim: a health state. Policy read: rewrite. Mechanism: problem. Proof level: Seen, since all that is known is that someone ran it. One variable: not yet set. It fails three fields (claim about the viewer, policy read and proof level), so it goes back for a rewrite, not into the queue.
Frequently asked
Copy the mechanism, not the words. Keep the problem mechanism and drop the assumption about the viewer: "Runners with bad knees, watch this" becomes "Running shoes with [measured cushioning feature]. Watch the test." Put your own proof in the brackets.