How to Scale Meta Ads: The Structure Behind $25M in Ad Spend
Want my team to build this for you? Work with Advergent. Rather build it yourself? Join the coaching waitlist.
Scaling Meta ads past $1,000 or $2,000 a day breaks for one of two reasons: you run out of fresh creative before the algorithm ever gets a real read on your offer, or you touch the budget before the ad set has told you anything. Fix both and scaling stops being a mystery. Build one testing campaign with one angle per ad set on broad targeting, let each ad set spend three times your cost per qualified call before you judge it, and bump budgets 10 to 30% every one to three days once a set earns it.
That's the short version. In the video above I walk through the rapid scaling system me and my team run on the biggest accounts at Advergent, built off $25 million in ad spend I've personally managed and well over $50 million in ad revenue generated from it. This post is the reference version: the campaign structure, the creative volume, the kill and keep math, and what to do the moment you find a winner.
One thing up front. This is not for a new or inexperienced media buyer who isn't fluent running $100 to $200 a day, and it's not for a business that can't produce 20 to 30 net new ad creatives every single week, which is the bare minimum we hold for clients at this level. If that's your situation, start with the small-budget version instead, because the volume rules below assume you can already feed the account.
What actually counts as a new ad creative?
A new ad is a net new, individualized creative built around a distinct angle, not a re-edit of old footage with a new cut or thumbnail. If you script something, record it, then re-edit the same footage later and count it as a fresh ad, the account doesn't see it that way, and neither should you.
What actually makes an ad new is going back to the numbers behind it: what the script needs to improve, what the callout to the persona needs to say differently, and whether the angle needs to get more specific into a sub-angle for a narrower slice of your audience. Aesthetics matter, but only if they're genuinely good, and they're a fine way to test cheaply before you commit to video. Past that point, especially selling B2B or selling info products, you need video ads, because you're targeting cold strangers who have never heard of you or your offer, and a well-edited, well-articulated video is the only thing doing the job an organic post or a warm email would normally do.
The comparison that makes this obvious: one business re-edits the same 10 ads it recorded six months ago and relaunches them against the same personas every week. Another pushes 25 to 30 fresh angles hitting new personas, opening up more of its total addressable market with every launch. The second one wins, every time, because it's the only one actually finding new demand.
And no AI slop. You can use AI inside your production workflow, but building your entire ad output out of AI-generated creative isn't something that can carry a scaling account right now.
Angle vs. format: which one do you test first?
There are only two levers in an ad creative: the angle, which is the persona, their situation and the problem your offer solves for them, and the format, which is the style it's shot in (direct to camera, selfie, carousel, B-roll, skit). Test angles first, because no format takes a business from $0 to $100K a month on its own. A winning angle does that. Format is how you squeeze more revenue out of an angle once you've already found it.
Once an angle wins, you don't move on to a new persona right away. You spin that same angle into other formats, static, talking head, carousel, split screen, B-roll, and launch all of them against the same audience pool that already responded. That's what I mean by scaling horizontally: instead of just raising the budget on the one ad that's working, you multiply the formats carrying that winning angle and let more of the account run on it. Raising the budget on a single ad set is the vertical move, and you'll use both, but horizontal comes first.
How do you structure a Meta ads account to scale?
Run one ABO testing campaign with one angle per ad set, and leave every ad set on broad targeting with no interest tags. The creative does the targeting: the algorithm reads the script and the frames of your video, learns who responds, and finds that audience faster than an interest stack would.
Inside that structure, angles either hit different problems for the same ideal customer, or they hit different ideal customers entirely, but every one of them ties back to the same product or service. Split your daily budget evenly across the angles you're testing. If you're spending $600 a day total with a $300 scaling campaign already running, that leaves $300 for testing. Split three angles evenly and each ad set gets $100 a day.
How many ads go in each ad set depends on what you can afford to feed it:
| Daily budget behind the angle | Ads to launch |
|---|---|
| Under roughly $200/day | 5 to 10 ads, go deeper on quality |
| Up to $1,000/day total | 20 to 30 new ads a week is the sweet spot |
| $1,000+/day behind one angle | 50 to 100 ads |
More volume means faster learning. A competitor launching 50 ads in a week gets the same learning you'd get from 10 ads a week, just five weeks sooner. If your budget can't support that volume yet, go deeper on quality with fewer ads rather than stretching thin.
When do you kill, keep or scale an ad set?
Let each ad set spend three times your average cost per qualified call before you judge it at all, then measure the result against your target with a 20% grace range. This single rule is 90% of the optimization game, because it stops you from killing an ad set on noise before it's had a fair chance to find its audience.
Say your target cost per qualified call is $100. Let the ad set spend $300 (3x) before you look at it. Land between $100 and $120 and you keep it, or scale it. Land above $125, or it hasn't booked a single qualified call by then, kill it. Note this is cost per qualified call, not cost per call. An ad set full of unqualified bookings isn't a keeper just because the cost per call looks cheap.
For anything you're keeping, bump the budget 10 to 30% every one to three days, not every day. Meta's algorithm needs a couple of days to re-stabilize after a budget change, and if you bump it daily, it gets confused, starts showing your ad to the wrong people, and your return on ad spend drops. How aggressive you go inside that 10 to 30% range depends on how much risk you're willing to take with a set that's already proven itself.
If the gap between cost per call and cost per qualified call is still fuzzy for your account, this breakdown of why cheap leads cost more goes into why buying booked calls, not raw leads, is the number that actually matters.
How do you know your numbers can actually scale?
Run your real funnel numbers, ad spend, cost per click, application rate, booking rate, show rate and close rate with average cash collected, through an ROI projection before you commit more budget. Your industry and ticket size set what a realistic cost per booked call looks like for you. Your ambition doesn't.
Here's the kind of projection this produces. Starting at $10,000 a month in ad spend, scaling 15% every month as winning ads compound, with a $200 target cost per booked call, a 70% show rate, an 85% gross margin and a $7,500 average deal at a 30% close rate:
| Scenario | Cost per booked call | Deal size | 12-month net profit |
|---|---|---|---|
| Base case | $200 | $7,500 | $1.7M (1,451 calls, ~$290K spend, 570% ROI) |
| Lower target | $150 | $7,500 | $2.3M |
| Higher-ticket account | $350 | $15,000 | $1.9M |
The middle row shows what happens if you're happy leaving more margin on the table versus pushing your target cost per call down to $150 instead of $200. The bottom row is a real pattern across some of the accounts we manage: a higher cost per booked call isn't a problem if the deal size absorbs it. What matters is knowing which one you're actually running, and planning both a best case and a worst case rather than setting your team's expectations off the optimistic number alone.
What do you do once you find a winning angle?
Double down immediately. Shift your creative output so 80% of new ads that week are on the winning angle and only 20% are fresh angle tests, and push every format you haven't tried yet, carousel, talking head, split screen, B-roll, skit, against that same angle before you move on. This is the point where 20 to 30 ad launches a week stops feeling like a grind, because the winner is doing most of the work.
Once you find it, you don't ration a winning angle, you run it until it stops paying you back. Kill the losers from your original test, pull the next-best angle from that list to fill the remaining 20%, and keep repeating the cycle: test angles, find a winner, squeeze it across formats, then test new angles again with whatever budget is left over.
The one caveat before any of this works
Everything above, the 3x spend rule, the kill and keep math, the campaign structure, assumes you already understand how Meta's algorithm is currently behaving, because that's what you're building the structure around. If your account's performance shifted after a recent platform update and you're not sure why, diagnose that first before you layer a scaling structure on top of an account that's already confused.
Where to start if you're scaling for the first time
- Confirm you can produce 20 to 30 net new creatives a week before you touch any of this. If you can't yet, that's the actual bottleneck, not your campaign structure.
- Build one testing campaign, one angle per ad set, broad targeting, no interest stack.
- Set a target cost per qualified call and let every ad set spend 3x that before you judge it, with a 20% grace range around your target.
- Bump budgets 10 to 30% every one to three days on anything you're keeping, never daily.
- Run your real funnel numbers through an ROI projection before you commit new spend, so your team's expectations are set by your market, not by hope.
- The moment you find a winner, flip to 80% of new ads on that angle, squeeze every format, and keep the remaining 20% testing fresh angles.
If you'd rather me and my team build this creative engine and run it inside your account instead of doing it yourself, apply here and I'll give you my honest read on the call about whether we can help.
FAQ
How many new ad creatives do you need to scale Meta ads?
At the level where scaling works, 20 to 30 net new creatives every week is the minimum I hold clients to. A re-edit of old footage doesn't count as new.
When should you kill or keep an ad set?
Let it spend three times your average cost per qualified call, then judge it against your target with a 20% grace range.
What do you do when you find a winning angle?
Double down: make 80% of that week's new ads on the winning angle and 20% fresh tests, and run every format you haven't tried against it.