Creative Testing 11 min read· Nov 18, 2026

Facebook Ads Multiple Text Options: How to Test Copy

Multiple text options let Meta rotate up to five primary texts, headlines and descriptions in one ad, but reporting will not tell you which line won. Here is when to split copy into separate ads and how to test it at volume.

How multiple text options work

Facebook ads multiple text options let you add up to five primary texts, five headlines and five descriptions to one ad, and Meta's delivery system decides which combination each person sees. It is a delivery optimization, not a test: results are reported for the ad as a whole, so use it to lift performance and use separate ads when you need to know which message won.

Meta's creative best practices for text in ads describe the feature as a way for the ad system to optimize delivery using variations of your text, matched to individual preferences that its machine learning models identify. It applies to single image and single video ads. The ceiling of five per field matches the asset feed limits in Meta's Marketing API, which also cap one ad at 30 assets in total.

Three mechanics matter for anyone trying to learn from it:

  • Combinations multiply. Five primary texts, five headlines and five descriptions make 125 possible versions of one ad. Meta serves the versions it predicts will work for each person, so some get most of the delivery and some barely run.
  • Every pairing has to read well. Any headline can land under any primary text, so nothing in one field can contradict another.
  • Engagement pools. It is one ad, so likes, comments and shares collect in one place. That is good for social proof and useless for telling you which line people reacted to.

Keep each line short enough to survive truncation. Meta recommends 125 characters for primary text, 40 for headlines and 25 for descriptions, and the Meta ad copy checker shows where each version gets cut.

What the reporting shows, and what it hides

This is where copy tests built on text options go wrong. Meta reports performance at the ad level, and its help page on text generation is blunt about AI variations: "you won't see performance details for specific text variations." Meta's guidance on dynamic creative makes the same point from another angle: results are aggregated across variations, so Meta does not recommend it as a substitute for split testing.

What you can get

  • Results for the whole ad, which you can compare against your other ads.
  • Asset breakdowns, on some ads. Meta's API exposes body_asset, title_asset and description_asset breakdowns for dynamic creative and asset customization ads, and Ads Manager offers a matching dynamic creative breakdown. Where those rows appear, they show impressions, spend and results per line.

What you cannot get

  • A fair head to head. Each line reached the people Meta predicted would respond to it, in unequal amounts. For example, a line with 70% of the spend did not beat a line with 5%; it was simply chosen more often. Asset rows describe allocation, not a contest.
  • The winning combination. Asset rows are totals per line, so a headline's row blends every primary text it ran under.
  • Attribution outside Meta. Every version shares one ad ID, one ad name and one destination URL, so a UTM built on {{ad.name}} cannot tell GA4 or your backend which copy a visitor saw.
  • Anything about generated lines. Variations added by text generation are not reported individually.

Use asset rows for one job: spotting a line Meta has almost stopped serving. That is a fair reason to replace it. It is not evidence that the other lines are better.

One ad or separate ads: the decision rule

Choose the container from the question you are asking, not from habit.

QuestionText options in one adSeparate ads in one ad setA/B test
What it answersWhich mix works best for this ad overallWhich message Meta favors, directionallyWhich message wins with comparable audiences
DeliveryMeta mixes lines per personMeta shifts spend between adsAudience split so nobody sees both versions
Data per copy lineAsset rows at bestFull results per adFull results plus a declared winner
Attribution in GA4Not by copyYes, through the ad nameYes
Social proofPools on one adSplits across adsSplits across versions
Use it whenScaling a proven creative with lines that differ in wording, not messageScreening three to five distinct anglesThe answer will change a landing page, offer or brief

Two rules hold up in practice. First, if the variants are different messages (price versus proof versus the problem), split them into separate ads, because the answer matters beyond this one ad. If they are different phrasings of one message, let Meta rotate them. Second, watch ad volume. Meta's learning phase guidance warns that the more ads you create, the less the delivery system learns about each one, so fanning ten creatives out by five lines (50 ads) in one ad set starves most of them. Split copy on one or two proven creatives, not the whole library.

When the decision is expensive, use Meta's A/B testing, which shows each version to a separate segment of your audience so nobody sees both. Our guide to A/B testing Facebook ad creatives covers the setup.

Text generation and Text improvements: turn them off for tests

Two Meta features can change the copy you think you are testing.

Text generation

Formerly called text variations, text generation writes up to five extra versions of your primary text and headline from what you entered, optionally shaped by personas or a brand tone with blocked words. It needs at least five words of primary text, works for English, Portuguese and Spanish inputs, and is not yet available to some regulated categories. Generated lines are added as extra text options, shown when Meta predicts they will help, and not reported individually. Meta says they are optional and can be turned off at any time: in Ads Manager, leave the suggestions unselected on test ads and remove any that were added.

Text improvements

This Advantage+ creative enhancement (text_optimizations in the API) can show the text options you provided as primary text, headline or description when Meta expects better performance, so your headline line can appear as primary text. Meta's text best practices page also notes that some enhancements are on by default, and that opting in to test new optimizations in advertising settings can apply test enhancements even when Advantage+ creative is off at the ad level.

For a copy test, switch both off on the test ads, or your variable is no longer your copy. The step by step process for Ads Manager and the API is in how to turn off Advantage+ enhancements. If you launch at volume, disabling Advantage+ enhancements automatically sets every eligible feature off on every creative and reads back what Meta applied. After the test, leaving these features on in scaling ads is a legitimate choice. Make it on purpose.

A copy testing workflow that scales

1. Write angles, not rewrites

Draft three to five lines that each make a different argument: price, proof, the problem, how it works, the objection. Put the argument in the first line, since that is what shows before "See more". Rewrites of one idea belong together in a single ad as text options.

2. Hold the creative constant

Run the lines on one or two creatives that already perform, so copy is the only variable. New copy on an unproven creative tests two things at once.

3. Size the test before you build it

Example: four primary texts on one proven video with a $30 target CPA. Reading purchases at a rough floor of 25 per line costs about 4 × 25 × $30 = $3,000, or 20 days at $150 a day, and an ad set will not split that spend evenly anyway. So screen on a faster signal first. At $150 a day, an even split gives each ad $37.50 a day, enough to compare cost per landing page view and outbound CTR after three or four days. Cut the bottom two, then confirm the top two on purchases, ideally in an A/B test.

4. Name it so the test reads itself

Give each line a short ID (P1 to P5 for primary texts, H1 to H5 for headlines) and put it in the ad name. The next section covers the setup.

5. Launch paused, check, then activate together

Preview each ad to confirm it shows the intended line with text generation and Text improvements off, then turn the ads on at the same time so they start level.

6. Read at the floor, then recycle the winner

Set the floor before launch and do not call it early. The winning line becomes the first text option on your scaling ad, the runner up joins it as an alternate, and the losing angles retire. The winning argument also belongs in the hook of the next creative and on the landing page. Once a winner is a proven post, scale it by Post ID so its engagement travels with it, and keep the cadence going with a weekly creative testing framework.

Naming and UTMs that keep copy tests readable

A copy test you cannot read in your analytics is half a test. Three habits fix most of it.

  • Put the copy ID in the ad name. A pattern of date, product, angle and copy ID (for example 20261118_serum_proof_P3) makes every Ads Manager row readable on its own and survives exports. The ad naming generator builds the pattern, and Meta ad naming conventions covers the reasoning.
  • Pass the name to analytics at click time. Set utm_content={{ad.name}} and Meta fills in the ad name on every click, so GA4 sessions carry the copy ID. The UTM builder assembles the string, and the Facebook ads UTM guide covers the rest.
  • Name before you publish. Meta's dynamic URL parameter specifications say name parameters such as {{ad.name}} keep the name the ad had when it was first published, so an ad renamed later still reports its old name.

None of this works for text options inside one ad, because every version shares the ad's name. That is the strongest practical argument for separate ads whenever a copy decision needs site or backend data behind it.

How Volume Creatives handles copy variations

Volume Creatives follows Meta's model by default and makes splitting an explicit choice:

  • Default: one ad per creative, variations inside. On the Copy step you add primary texts, headlines and descriptions. Each creative becomes one ad, and the extra lines go to Meta as text options, up to five per field. In a new or empty ad set, ten creatives with three primary texts make ten ads, each rotating three lines. Lead ads and carousels use the first text only.
  • Ad sets that already hold ads get a choice. Meta will not mix text option ads with standard ads in one ad set, so the review screen flags it and offers two options: one ad per creative with your first text, or every variation launched as its own ad. The real ad count shows before anything is created.
  • Splitting pairs lines, it does not cross them. Separate ads pair lines by position (primary text 1 with headline 1, 2 with 2) and repeat the shorter list. Example: three primary texts and two headlines give three ads per creative (P1 with H1, P2 with H2, P3 with H1), not six. The ad combination calculator shows how fast counts grow.
  • Copy can apply to all ads, each ad set or each ad, and every ad gets your naming pattern and UTM template.
  • Copy guardrails. The review warns when primary text passes 125 characters or a headline passes 40, Advantage+ creative enhancements are opted out on every creative by default, and every ad launches paused.

The short version

Use multiple text options to let Meta find the best mix on an ad you are scaling, and accept that it will not teach you much. When you need to know which message wins, launch each line as its own ad on a proven creative, put a copy ID in the name, tag it with {{ad.name}}, and turn text generation and Text improvements off for the test. Settle the expensive decisions in an A/B test. If an AI agent builds your tests, our guide to Meta ads MCP servers covers which actions to keep behind a human.

If building test ads by hand is the bottleneck, the Volume Creatives bulk launcher applies your copy, naming and enhancement settings to every ad and publishes the batch paused. See pricing; the first 10 ad uploads are free.

FAQ

How many text options can you add to a Facebook ad?

Up to five each for primary text, headline and description, which matches the limits on Meta's asset feeds. Text generation can add up to five AI variations of primary text and headline on top of what you write. Meta recommends keeping primary text near 125 characters, headlines near 40 and descriptions near 25 to avoid truncation.

Can you see which text option performed best in Meta Ads Manager?

Not cleanly. Meta reports results for the ad as a whole and says you will not see performance details for individual generated variations. Asset breakdowns, where available, show how delivery was spread across your lines, but each line reached different people in different amounts. For a real answer, run each line as its own ad or in an A/B test.

Should I test ad copy with multiple text options or separate ads?

Use separate ads when the lines are different messages and you need to know which one wins, especially if the answer will change a landing page or brief. Use multiple text options when the lines are variations of one message and the goal is performance rather than learning. Many teams do both: screen angles as separate ads, then load the winners into one ad as text options.

How do I turn off AI text variations in Meta Ads Manager?

Text generation is optional, so leave the suggested variations unselected when you build the ad and remove any that were added. Then switch off the Text improvements enhancement in the Advantage+ creative settings. Also check whether your account is opted in to test new optimizations in advertising settings, since that can apply test enhancements regardless of your ad level choices.

Does changing text options reset the learning phase?

It can. Changing text on a live ad is an ad edit, and Meta says editing an ad, ad set or campaign during the learning phase resets learning. Plan your text options before launch, and test new copy as new ads instead of editing an ad that is still learning.

Volume Creatives team
Volume Creatives team· Last updated Nov 18, 2026

Media buyers who run client accounts on Meta every week

Written by the team behind Volume Creatives, the bulk Meta ad launcher. We build the tooling we use to run client ad accounts every day.

Launch your next test in one click.

Volume Creatives bulk launches hundreds of Meta ads with enhancements off, naming and tracking applied on every one.

Start free

First 10 uploads free. No card required.