How to test ad creative on Meta: a simple hook testing method

On this page
  1. What to test first
  2. The method: same body, new first 3 seconds
  3. The metrics: hook rate and hold rate
  4. How to read the results
  5. Keep a simple testing sheet
  6. Stay inside ad policy
  7. A one-week starter plan
  8. Questions people ask

Short answer: test one variable at a time, and start with the hook, because the first 3 seconds decide whether anyone sees the rest. Keep one ad body, write 3 to 5 different openings, run them side by side with the same audience and budget setup, and decide in advance how much each may spend. Use hook rate and hold rate to understand why an ad works, but pick winners on cost per purchase (or ROAS), because a hook can stop the wrong people.

What to test first

Creative tests answer different questions. From biggest effect to smallest, most media buyers test roughly in this order:

  1. Concept: the core idea. A demo, a problem and solution, a founder story, a comparison, a review-style video.
  2. Angle: the reason to buy you lead with. Saves time, looks better, lasts longer, costs less per use.
  3. Hook: the first 3 seconds. Opening line, on-screen text, first visual.
  4. Format: short video, longer video, static image, carousel.
  5. Details: captions, music, call to action, headline.

New concepts and angles find the big wins. Hooks are the cheapest way to get more from a concept that already works, because you don't need to film a new ad. You need one good body and several openings.

Testing button colours and headline tweaks rarely moves results enough to read. Make variants meaningfully different.

The method: same body, new first 3 seconds

  1. Pick one body. Everything after the hook: the demo, the benefits, the offer, the call to action. Keep it identical across versions.
  2. Write 3 to 5 hooks, ideally from different angles. A problem callout, a curiosity hook and a demo hook teach you more than three versions of one idea.
  3. Change only the first 3 seconds. Opening line, on-screen text and first visual. If you also change the music or the offer, you won't know what made the difference.
  4. Launch them together in one ad set, so they share the audience, budget and dates. If you use Advantage+ creative enhancements, consider turning off the ones that change the part you're testing (for example, automatic text or visual changes), so the platform doesn't blur your variants.
  5. Set a spend limit before you start. Decide how much each version may spend before you judge it. Many buyers tie this to their target cost per purchase, for example "each hook gets up to 2x target CPA". Pick a rule that fits your budget and write it down. Deciding in advance stops you killing a hook after one bad afternoon.
  6. Don't edit mid-test. Changing budget, audience or creative during the test adds a new variable and muddies the result.

If you need a cleaner answer to a bigger question (two concepts, two audiences, two landing pages), Meta's A/B test tool under Experiments in Ads Manager splits the audience so each person only sees one version. It's slower and needs more budget, so most buyers save it for high-stakes questions and use simple side-by-side tests for hooks.

The metrics: hook rate and hold rate

Neither is a default column in Ads Manager. You can add them as custom metrics, or work them out in a spreadsheet.

Metric Formula What it tells you
Hook rate 3-second video plays ÷ impressions How many people stopped scrolling
Hold rate ThruPlays (or 15-second views) ÷ 3-second video plays Whether the hook led into the body, or grabbed attention and lost it
CTR (link) Link clicks ÷ impressions Whether the ad made people want to see more
Cost per purchase Spend ÷ purchases Whether it makes money. The deciding number

Some buyers use 15-second views for hold rate, others use ThruPlays. Pick one definition and stick to it, or your comparisons break.

About benchmarks: you'll see "good hook rate" numbers online. They vary a lot by niche, price point, placement and source. Compare hooks against each other and against your own account's history. Your own median is a better bar than someone else's average.

How to read the results

Pattern What it usually means What to do
High hook rate, low hold rate The opening works, the body loses people Keep the hook, tighten the next 5 to 10 seconds so they keep paying off the hook
Low hook rate, good hold rate The people who stay like it, but too few stop Keep the body, test new openings
High hook and hold, low CTR People watch but don't act Look at the offer, the call to action, or how clearly you show the product
Good CTR, poor cost per purchase The ad works, something after the click doesn't Check the landing page, price, shipping costs and checkout. Also check your tracking is counting purchases right
Good on everything A winner Reuse the hook on other bodies, try it with a new creator, or film the same line in a new setting

The winner is the version that leads to purchases at a cost you can live with. A clickbait hook can win on hook rate and still lose money.

Keep a simple testing sheet

One row per ad. These columns are enough:

After a few weeks this sheet becomes your most useful creative asset. You'll see which angles your buyers respond to, which is more valuable than any single winning ad.

Stay inside ad policy

Hooks that call out the viewer can break Meta's rules. Avoid hooks that assert or imply personal attributes (health, body, finances, identity), like "Struggling with your weight?". Talk about the product or the situation instead. Keep before-and-after visuals to objects and spaces, not bodies. Only use claims, numbers, reviews and deadlines you can prove. Policies change and vary by country, so check the current rules for your category before you launch.

A one-week starter plan

Questions people ask

What is a good hook rate on Meta?

There's no reliable universal number. Published benchmarks disagree and vary by niche, placement and price. Compare hooks against each other and against your own account's history.

How long should I run a creative test?

Until each version has spent the amount you set in advance, and long enough to get past the first day's swings. On small budgets that can mean a week or more. Judge on cost per purchase, not on early CPA.

How many creatives should I test at once?

3 to 5 versions of one variable is a practical range. More versions split the budget so thinly that none get enough spend to read.

Should I test hooks or new concepts?

Both, for different reasons. New concepts and angles find big wins. Hook tests get more out of a concept that already works, cheaply, because only the first 3 seconds change.

Does this work on TikTok too?

The method does: one body, several openings, same setup, judge on cost per purchase. TikTok's reporting names and video view metrics differ, so map the formulas to its columns.

Platform menus and defaults change often. This guide reflects them as of 18 Nov 2026. Nothing here promises results: they depend on your store, offer, budget and execution.

Keep reading

More free guides