The short version
- The 3-2-2 method runs 3 creatives × 2 primary texts × 2 headlines = 12 combinations in one ad set, usually through dynamic creative.
- It's fast to set up and tests visuals and copy together, but Meta won't split spend evenly, so read the results asset by asset.
- A winner is an asset that gets real spend and sells at or under your target cost per purchase, not the one with the best click-through rate.
- Meta has been folding dynamic creative into newer options such as flexible ads, so the exact setup screens vary by account.
- For cleaner reads or more ideas per round, test more concepts or give each creative its own ad.
What the 3-2-2 method is
The 3-2-2 method is a creative-testing structure that's popular with D2C media buyers. Into one ad set you put:
- 3 creatives — images or videos, ideally three genuinely different concepts,
- 2 primary texts — the copy above the image or video, and
- 2 headlines — the short line beside the button.
Usually this runs through dynamic creative, which mixes the pieces into 3 × 2 × 2 = 12 combinations and shows each person the ones Meta predicts will work best for them.
| Text 1 + headline 1 | Text 1 + headline 2 | Text 2 + headline 1 | Text 2 + headline 2 | |
|---|---|---|---|---|
| Creative A | A · 1 · 1 | A · 1 · 2 | A · 2 · 1 | A · 2 · 2 |
| Creative B | B · 1 · 1 | B · 1 · 2 | B · 2 · 1 | B · 2 · 2 |
| Creative C | C · 1 · 1 | C · 1 · 2 | C · 2 · 1 | C · 2 · 2 |
Its appeal is simplicity: one ad set, seven assets, one budget, and at the end a view of which visual and which lines did best. It spread through D2C media-buying circles because it's quick to build and simple to repeat every week.
How to set it up, step by step
- Choose three different concepts. Three versions of one photo in different colours won't teach you much. Pick three different angles or formats — say, a problem–solution photo, a customer-review graphic and a short product video.
- Write two primary texts that take different approaches. One might lead with the problem, the other with proof. Put the point in the first line: roughly 125 characters show before "See more", which is Meta's recommended length.
- Write two headlines that say different things. A benefit against an offer, not two wordings of the same benefit. Meta's recommended length is roughly 27–40 characters.
- Create a sales campaign optimised for purchases, using your Pixel's Purchase event.
- Create one ad set with broad targeting or Advantage+ audience and automatic placements. If your account shows a dynamic creative option at this level, turn it on.
- Build the ad. Upload the three creatives, add both texts and both headlines, choose one call-to-action button, then add the landing page and your URL parameters.
- Name everything so you can find it later. Our naming conventions guide has a system that scales.
- Set a budget that can reach a verdict (below), publish, and leave it alone for a few days.
An example round
Here's what one round might look like for a fictional cold brew brand:
| Slot | Example | What it tests |
|---|---|---|
| Creative A | UGC-style photo of a can on a cluttered desk at 3pm | Problem–solution: the afternoon slump |
| Creative B | Bold graphic: $0.45 a cup against $2.60 at a café | Price anchor |
| Creative C | Short video of the concentrate poured over ice | Sensory, product-first |
| Text 1 | "Your 3pm slump doesn't need a $2.60 café run. Café-strength cold brew, ready at your desk in 30 seconds." | Problem-first |
| Text 2 | "Smooth, never bitter, and one pouch makes 20 cups. Stir it into cold milk or water and you're done." | Benefit-first |
| Headline 1 | Café-strength coffee in 30 seconds | Benefit |
| Headline 2 | 20% off your first cold brew pouch | Offer |
Notice that every slot differs in kind, not just in wording: three different angles, a problem-first and a benefit-first text, a benefit and an offer headline. That's what makes the results worth reading. If Creative B wins, you've learned that price is a strong argument for this product — a far more useful lesson than "the blue version beat the green one".
How long a 3-2-2 round runs
Give each creative roughly 2–3× your target cost per purchase in spend. That's enough for a clear loser to show itself (no sale after about 2× target) or for a likely winner to collect a few sales. The copy doesn't need a budget of its own. It rides along with the creatives.
A test this size won't take the ad set out of the learning phase — Meta says that takes about 50 optimisation events in 7 days — and it doesn't need to. You're comparing assets within the test, not waiting for the ad set to settle. The creative test calculator works out the numbers for your own CPA.
How to read the results
Wait until the test has spent roughly what you budgeted — a few days at least — before reading anything into it. Then break the ad's results down by creative asset. Ads Manager has offered a breakdown by dynamic creative element for this, though its name and place in the menus vary. You'll see spend, clicks and purchases for each image or video, each primary text and each headline.
Two things to keep in mind:
- Spend won't be even. Meta pushes budget towards the assets it predicts will do best, often within a day or two. An asset with little spend hasn't lost. It hasn't been tested.
- You're reading assets, not combinations. The breakdown shows how each piece did across all the combinations it appeared in, not a clean result for each of the twelve. Treat it as a strong hint, not a lab result.
What "winning" means
A winning asset has both real spend and purchases at or under your target cost per purchase — three or more is a useful bar. A high click-through rate on its own doesn't count: some ads are brilliant at earning clicks and poor at earning sales.
| What you see | What it probably means | What to do next |
|---|---|---|
| One creative takes most of the spend and sells at or under target | A likely winner | Give it its own ad or more budget, and make variations of it |
| One creative takes most of the spend but sells above target | People engage but don't buy | Check the landing page matches the ad. Test new concepts against it |
| A creative gets almost no spend | Untested, not beaten | Bring it back in a later round if you still believe in it |
| The two texts or headlines finish close together | Copy isn't the lever this round | Keep the better one. Test bolder copy differences another time |
| Nothing sells at target | None of these ideas works yet | Try new angles, not new colours |
Then plan the next round: carry forward the winning creative and copy, and fill the other slots with variations of the winner or with new concepts.
Common 3-2-2 mistakes
- Editing the ad mid-test. Swapping a creative or rewriting a text partway through mixes two tests into one, and Meta says significant edits can send an ad set back into the learning phase. Let the round finish, then start a new one.
- Calling it on day one. Early results swing, and delivery is still exploring. Wait until the test has spent roughly what you budgeted.
- A budget too small to judge. If each creative can't reach about 2× your target cost per purchase, the test ends without verdicts. Run fewer tests with more money each rather than many starved ones.
- Hiding an offer test inside a copy test. If one headline carries a 20% discount and the other doesn't, the "winning headline" may simply be the discount. That's fine if the offer is what you want to test — just know that's what you're testing.
- Carrying losers forward. Keeping a weak text for another round "in case it works with the new creative" spends budget on something that has already had its chance.
- Keeping the winner but not the reason. Write down why a creative won — the angle, the format, the opening — so the next round tests that reason, not just more of the same pixels.
Pros and cons
| Pros | Cons |
|---|---|
| Quick to build: one ad set, seven assets | Spend is uneven, so some assets are barely tested |
| Tests visuals and copy in the same round | You read assets, not the twelve combinations |
| One budget and one learning pool, which suits smaller accounts | Only three concepts per round — slow when variety matters |
| Meta can serve different combinations to different people | Dynamic creative's options and availability keep changing |
| Easy to repeat every week | Copy results are often too close to call |
When to go beyond 3-2-2
3-2-2 is a sensible default, not a law. Move past it when:
- You need more ideas per round. Under Meta's current delivery, variety matters more than it used to — see our guide to Andromeda and creative diversity. If your budget allows, test five to ten genuinely different concepts instead of three.
- You want Meta to choose the format for each placement. Flexible ads take up to 10 images and videos in one ad and let Meta pick per placement and per person. They're built for delivery rather than for a tidy read.
- You need a clean read on each ad. Give each creative its own ad, with its best copy, inside one ad set. Every ad gets its own results, you can pause losers one at a time, and a winning ad's post — likes and comments included — can be reused when you scale it.
- Copy keeps finishing level. If your texts and headlines never separate, stop testing copy in every round. Test it on its own when you have a genuinely different message.
How 3-2-2 maps to bulk testing
Bulk testing keeps the core idea — several options in one ad set, and let results decide — but turns the dials differently:
| 3-2-2 | Bulk test | |
|---|---|---|
| Concepts per round | 3 | As many as the budget can judge, often 5–12 |
| Copy | 2 texts × 2 headlines, mixed by Meta | Written for each ad's own concept |
| Structure | One ad set, one dynamic ad | One ad set, one ad per creative |
| What you read | An asset breakdown | Each ad's own spend, sales and cost per purchase |
| Verdicts | By eye | By fixed rules tied to your target cost per purchase |
The trade-off is setup: separate ads take longer to build by hand, and Meta will still favour some ads over others within the ad set. In return, each result belongs to one complete ad. Matching the copy to each concept also avoids the odd pairings dynamic creative can produce, like an offer headline under a founder-story image.
This is the structure BulkAds uses. The bulk ad launcher creates one campaign with one ad set and one ad per creative, all paused until you approve, and their results sync back into BulkAds for you to judge. Autopilot builds its own tests the same way and gives each of those ads a verdict on real sales: it stops an ad that spends 2× your target cost per purchase without a sale, and calls a winner after 48 hours with three or more sales at or under target.
Questions people ask
What is the 3-2-2 method?
A creative-testing structure popular with D2C media buyers: 3 creatives, 2 primary texts and 2 headlines in one ad set, usually through dynamic creative, which Meta mixes into 12 combinations.
Is the 3-2-2 method still worth using?
Yes, as a quick way to test visuals and copy together. Many advertisers now also test more distinct concepts per round or give each creative its own ad, because variety matters more under Meta's current delivery.
How much budget does a 3-2-2 test need?
Roughly 2–3× your target cost per purchase for each of the three creatives. At a $6.20 target that's about $37–$56 for the test.
Can I run a 3-2-2 test without dynamic creative?
Yes. Build the combinations as separate ads in one ad set, or use whichever ad-level text and media options your account offers. Separate ads take longer to set up but give a cleaner read.
How do I see which creative won?
Break the ad's results down by creative asset in Ads Manager (the menu label varies) to see spend and purchases for each image, text and headline. Judge on purchases and cost per purchase, and treat assets with little spend as untested.
Should the three creatives be different concepts?
Yes. Three colour versions of one ad tell you very little, three different angles or formats show you which idea your customers respond to.
Keep going
- Features
Bulk ad launcher
A whole batch live on Meta in one go
- Features
Autopilot
Stops losers, scales winners, inside your caps
- Free tools
Creative test calculator
How many new ads you can test, and how long until every one has a verdict
- Guides
Flexible ads vs dynamic creative: which to use for testing and scaling
Three ways to give Meta several creatives. How dynamic creative, flexible ads and one ad per creative differ, when to use each and how to read the results.
- Guides
How many ad creatives should you test? The number, and the maths.
A straight answer, then the simple sum that sets your real number, how many new ads to launch each week at your spend, and when to stop testing and scale.
- Glossary
Dynamic creative
Dynamic creative is a Meta ad set option where you upload several images or videos, headlines and primary texts, and Meta mixes them into combinations, showing more of the ones it predicts will perform best.
Written by the BulkAds team. Published , last checked . Meta changes its tools often — if something here no longer matches Ads Manager, tell us at contact@bulkadsai.com.
BulkAds AI is an independent product and is not affiliated with, endorsed or sponsored by Meta Platforms, Inc. Facebook and Instagram are trademarks of Meta Platforms, Inc.