How to structure and scale Meta campaigns

1 The bottom line

Structure Meta campaigns so creatives do the targeting, test many ads on small budgets, and scale only winners in steps. Meta delivers to the audience each creative attracts, so new concepts, not settings, drive scale.

2 Key takeaways

• Put your time into creatives, not Ads Manager settings. Meta now finds the audience through the ad itself.
  • Ethan Ethier, Built With Science24:40 says bids and targeting are largely out of your control, while better content moves the needle.
  • Marcus Burke38:26 sums up the audience as message plus media plus content type times funnel.
• Test at high volume on small budgets and expect most ads to lose. Winners are rare, and the losers are the price of finding them.
  • Alex, Boost App Social20:47 saw one video out of 85 take most of the spend and work.
  • Luke Martin-Fuller, Visible46:00 scales only one or two videos a week.
• Graduate winners through budget tiers instead of raising one budget. Each tier proves an ad before it gets real money.
  • Ryan Thorpe, Flo16:31 starts every video at $10 to $20 and ends with a few evergreens that take about 90% of spend.
• Once you can afford it, split ad sets by format and message. Otherwise Meta pours the budget into one ad.
  • Marcus Burke51:00 set budgets per ad set across five ad sets split by media, format and message.
• Raise spend slowly, and push hard only in the cheap seasonal window. Fast jumps raise CAC, but some weeks are much cheaper.
  • George, Wrestle AI30:27 grew from about 25k a month to about 41k MRR in 2 months with gradual increases.
  • Ryan Thorpe, Flo19:59 spent roughly 75% of the ad budget in January, when cost per download was about half.

a lot of the targeting on meta is specifically done through the creatives.

Ethan Ethier, Built With Science

2.1 Set up the account and send signals

Get the account, the channel and the event signals right before you spend.

• Send Meta early funnel events while you optimize for purchase. Source: Steve Young, App Masters34:45
  • Send registration, content view (used as a paywall view) and initiate checkout events, as Steve Young34:45 describes a past guest doing.
  • Optimize for the purchase event, as that guest did because he had no trial.
  • The guest's Meta account got noticeably better and smarter from the extra signals Steve Young34:45.
  • Steve and Joseph suspect Meta then retargets people who viewed the paywall, so they say to load all creatives and not try to outsmart the algorithm.
• Run Meta in-house so the ads match what the product does. Source: Ethan Ethier, Built With Science21:30
  • Pick someone passionate about the product who will not over-promise, over an agency.
  • Aim for a seamless handoff from the ad to the product experience.
  • The goal is to "promote the app in a way that isn't faking anything."
• Push one paid channel to its ceiling before you add another. Source: Josh, Blue Throne49:34
  • Keep scaling the current channel until ROAS curves turn unfavorable, then start the next one.
  • This builds real mastery in each channel Josh49:34.
  • Content type can pick the channel. Cedric Roberge, Pep AI50:40 found TikTok keeps rejecting peptide Spark ads, while Meta is more lenient and performs better.
  • Rival GLP apps in the top 100 of Health and Fitness run only Meta ads Cedric Roberge50:40.
• Expect account bans and keep fighting to recover. Source: Alex, Boost App Social8:52
  • Appeal persistently and explain internally why you should be allowed to advertise.
  • Once spend is high enough, a Meta rep and a warmed-up account help.
  • Accounts were restricted or banned six to eight times, including permanent bans, even with no policy broken. They always got them back. Early support was offshore and could not fix much.

2.2 Structure ad sets and campaigns

Start with one ad set, then split by format and message with a budget on each.

• Put all ads in one ad set while budget is tight. Source: Marcus Burke54:30
  • Throw everything in, see what comes out on top, and repeat next week.
  • Split by message and media only once ads are near break-even or you have funding to speed up learning.
  • Multiple ad sets would require doubling or quadrupling the budget.
• Then cluster ad sets by format and message, with budgets at ad set level. Source: Marcus Burke51:00
  • Cluster by media type, style (text walls, memes, UGC video) and message (overthinking, anxious attachment).
  • Set budget and cost-per-target goals on each ad set, or Meta pours budget into one best ad or message.
  • Marcus ran five ad sets split by media, format and message. Meme ad sets must deliver traffic at about half the price because their quality is lower.
  • For scale, he buys every distinct audience, not only the single best ad.
• Plan new ads as message plus media plus content type, times funnel. Source: Marcus Burke38:26
  • Message sets the psychographics and audience. Media (static or video) sets placements and age mix. Content type shapes who engages.
  • To scale, add new pain-point messages, new formats and media types, or new funnels. Check where ads were delivered to learn from it.
  • Video skews to a younger Instagram Reels audience, and statics to Facebook feed with older users. Street interview styles skew younger than raw text walls.
• Run web and app campaigns separately, and test each concept on both. Source: Leon Sasson, Rise Science10:19
  • Use web ads for problem-aware reach and app ads for solution-aware reach Marcus Burke17:10. An app funnel starts with a download, so the ad must be close to the solution.
  • Testing each concept twice gives more chances of a winner, because the audiences differ.
  • Separate campaigns also smooth swings. Web may be down in a month while app is up Leon Sasson9:00.
  • A creative that flopped on app often starts working on web, and the reverse. Rise did not do this at the start but now tests most creative on both.
  • Static ads deliver poorly in app promotion campaigns, so start with video and test statics Marcus Burke17:10. A Nibble geopolitics ad on the web side has no selling at all.

2.3 Test creatives at volume, in tiers

Most ads lose, so make many, test them cheaply and search for new concepts.

• Plan the whole operation around producing enough ads to find a winner. Source: Alex, Boost App Social20:47
  • Make enough ads that a winner appears often enough to be profitable.
  • Treat the losers as the price of finding the winner.
  • One video out of 85 took most of the spend and worked. Total spend was about 1.5 to 1.8 million dollars, with something like 800k wasted, and the winners paid back hundreds of thousands in profit.
• Explore new concepts instead of endless variations of one ad. Source: Patrick and Pablo, Social48:10
  • Test widely for new angles, concepts and formats, then go deep where something works.
  • Variations of one ad stay in the same region and perform less over time. Creative is an infinite search space, and there are not enough impressions to test every ad.
• Test every video cheaply and graduate winners through three levels. Source: Ryan Thorpe, Flo16:31
  • Level one: each video is its own ad set with $10 to $20 of spend. Kill weak ones after $5 to $10.
  • Level two: winners go in with four or five videos per ad set and $100 to $500 budgets.
  • Level three: the one or two that rise to the top become evergreens. Keep the lower levels running while they still work.
  • Level three evergreens take about 90% of ad spend, run by Ryan, a media buyer and a head of content.
  • Luke Martin-Fuller, Visible46:00 tests member-made videos on very small budgets, and only one or two a week ever get scaled spend.
• Accept that Meta has to like the creative, even a great one. Source: Alex, Boost App Social24:17
  • Do your best on quality, and expect the platform to decide the rest.
  • A creative that looked great took 20K of spend and still did not work. A winner needs both a Meta match and an audience match.

2.4 Steer who Meta delivers to

Settings matter less than creative, but a few levers still shape delivery.

• Spend less time in Ads Manager and feed Meta better signals. Source: Ethan Ethier, Built With Science24:40
  • Feed the highest-intent users back to Meta.
  • Use a custom dashboard to validate which trials truly come from Meta.
• Use value rules to push spend toward older, higher-value users. Source: Marcus Burke44:30
  • Keep targeting broad and tell Meta that someone 45 plus is, say, 30 to 40% more valuable.
  • Also make ads that attract older people. Low spend there often comes from not making content for them.
  • This works around the fact that trial-to-paid happens about 7 days later, too late for the algorithm to react.
• Split outlier ads into their own ad set so Meta spreads delivery. Source: Marcus Burke47:50
  • Chart each ad's placement mix against cost per target.
  • Pull ads with good cost per target but heavy Facebook feed share into a separate ad set with its own budget, broad targeting and all placements.
  • Compare them with the other ads to learn why they deliver differently.
  • Outlier ads with favorable cost per target spent 40 to 50% of budget on Facebook feed. Mixed with others, they get starved because Meta concentrates spend, such as 80% of budget on one ad.

2.5 Scale spend without breaking performance

Raise spend in steps, keep hunting for the breakthrough and time big pushes.

• Increase spend gradually. Source: George, Wrestle AI30:27
  • Scale in small steps, and make more creatives so spend can keep growing.
  • Scaling too fast raised CAC and then evened out, so the gain was lost. Gradual increases took them from about 25k a month to about 41k MRR in 2 months.
  • George says the unsolved problem is producing about 10x as many creatives.
• Treat stable CPM and CPA as a sign you stopped pushing. Source: Yuliya Lennox, Solid Starts4:00
  • Keep hunting for the creative that halves CPA or triples daily purchases.
  • It is tempting to relax when a $30 user returns $50. The breakthrough usually comes from one creative or concept, and early results should not be good.
• Spend most of the year's budget in early January. Source: Ryan Thorpe, Flo19:59
  • For health, fitness and productivity apps, spend about 75% of the ad budget in the first month, then use the rest of the year to test which creatives to rerun.
  • Cost per download was about half the rest-of-year price, maybe less.
• Time spend to the weekday and the days around New Year. Source: Ryan Thorpe, Flo25:15
  • Dec 25 is dead. Dec 26 to 28 is for testing at $5k to $10k a day. Conversion quality jumps again from Dec 28 to 30.
  • Push hardest on the Sunday before people go back to work. Go up to 10x spend from Saturday night if early metrics are good.
  • Done hit number one free app in the US App Store on Jan 2 and 3 two years in a row after heavy spend in this window.
  • If New Year's Day falls on a Thursday or Friday, expect no big jump. Sunday to Tuesday converts best for self-improvement apps, and the first few days of each month convert better for subscriptions.
• Avoid edits to live, high-budget campaigns when tired. Source: Alex, Boost App Social21:17
  • Alex duplicated an ad at 11:30 p.m. by mistake and wasted 8k. He treats such losses as part of the cost of testing at scale.

2.6 Where they disagree

Advice splits on how much to control, how to make new ads and how fast to scale.

• Control versus trust. Ethan Ethier24:40 says tinkering with bids and targeting is largely out of your control.
  • Marcus Burke44:30 still uses value rules, and splits outlier ads into their own ad sets.
  • Steve Young34:45 sits closer to Ethan: don't try to outsmart the algorithm.
• One ad set or many. Marcus Burke54:30 says use one while budget is tight.
  • He splits by format and message only when ads are near break-even or you have funding Marcus Burke51:00.
  • Ryan Thorpe16:31 starts every video in its own ad set at $10 to $20, which is a different way to keep tests cheap.
• Variations or new concepts. George, Wrestle AI30:27 makes variations of each winner and has other creators copy the main creator's style.
  • Patrick and Pablo, Social48:10 say variations of one ad stay in the same region and perform less over time.
  • George admits that producing about 10x as many creatives is still unsolved.
• Slow or fast. George, Wrestle AI30:27 raised spend gradually after a fast jump lost the gain.
  • Ryan Thorpe25:15 goes up to 10x on the Sunday before people go back to work, but only if early metrics are good.
  • That push fits seasonal health, fitness and productivity apps Ryan Thorpe19:59. It may not fit other products.