Short answer: A winning Meta ad climbs a series of levels, and each one asks it to prove one new thing. In creative testing it must sell to strangers. At the first verdict it must earn more money. An extension or a runoff gives a fair second look to ads that were starved, promotion moves a proven seller into Scaling, and Scaling asks whether it can keep selling at real budget next to the ads already winning there. Ads that sell, but too expensively, leave by a side door into the Graveyard.
Most accounts call an ad a winner after a few sales. That shortcut is why so many accounts promote ads that flop and kill ads that would have worked. Seeing the journey as levels changes the question from "is this ad good?" to "what has this ad proven so far, and what must it prove next?"
What are the levels of a winning ad?
The levels of a winning ad are the checkpoints every new creative passes on its way from first launch to steady, profitable spend: creative testing, the first verdict, extension, runoff, promotion, Scaling, and the Graveyard as the side exit.
Not every ad touches every level. A strong ad might go straight from testing to promotion. A starved ad might need an extension or a runoff first. Most ads stop at the very first level, and that is how it should be. The structure only works because it is designed to fail most ads quickly and cheaply.
Here is the map:
- Creative testing. Where it happens: Creative Testing campaign, cold audiences. What the ad has to prove: It can get strangers to buy
- First verdict. Where it happens: Creative Testing, at a spend checkpoint. What the ad has to prove: It deserves more money than a first look
- Extension. Where it happens: Creative Testing, spread Ad Sets only. What the ad has to prove: It was starved, not bad
- Runoff. Where it happens: A new Ad Set built for runners up. What the ad has to prove: It can win budget on its own
- Promotion. Where it happens: The move into Scaling. What the ad has to prove: The sales repeat, at goal, on more than one source
- Scaling. Where it happens: Scaling Campaign. What the ad has to prove: It can sell at real budget next to incumbents
- Graveyard. Where it happens: Cost cap campaign. What the ad has to prove: It sells, and can be kept at a price you accept
This all lives inside our CTSG system, where Creative Testing, Scaling and the Graveyard each have one job.
Level one: what does an ad have to prove in creative testing?
In creative testing, an ad has to prove it can make strangers buy. Not past customers, not people who visited the site yesterday. Strangers.
That is why the Creative Testing campaign excludes past customers and recent site visitors, with the same exclusions in every Ad Set. If an ad is shown to warm people, it can look like a winner just because it reached people who were already about to buy. The test would tell you about the audience, not the ad.
The problem: a new ad sitting next to a proven ad never gets a fair look, because Meta feeds what already works.
The solution: new ads compete only with other new ads, in small groups inside an Ad Set, on a small budget. Each ad gets its chance to show a signal. We also watch daily frequency (how many times, on average, each person saw the ad that day). An ad that sells while its daily frequency stays close to one is selling to new people. That is the strongest kind of early signal. See why creative testing must run to cold audiences.
Level two: what is the first verdict, and what does it prove?
The first verdict is the moment an Ad Set has spent enough to judge the ads inside it, and it proves whether an ad deserves more money than a first look.
We set that checkpoint from the account's own cost per sale and buying cycle. Before it, we only cut single ads that burned their share with zero sales. At it, we look at every ad on its own numbers. The checkpoint is a time to look, never an automatic pause.
At this level the ad sorts into one of a few outcomes:
- Zero sales after a fair spend: paused. It had its chance.
- Sales, but cost and return both below your floor: moved to the Graveyard.
- One number passes and the other fails: held and checked again soon, because cheap products flatter cost per sale and pricey products flatter return.
- More than one sale, at goal, mostly to new people: on its way to promotion.
We cover the timing in the first verdict. The point here is what it proves. Passing the first verdict does not make an ad a winner. It makes it worth more of your money.
Level three: when does an ad get an extension?
An extension is extra, capped spend for an Ad Set whose money spread thin across its ads, and it lets an ad prove it was starved rather than bad.
Sometimes Meta splits the budget so evenly that no single ad got a real test. Pausing that Ad Set on its total would kill ads that never had a chance. So instead, we look for early signs of life, like a cost per add to cart in line with your best ads, and give the Ad Set a small, capped extension. Each step of the extension has a hard stop, and sales decide at the end.
The extension is not a reward. It is a fairness check. The ad still has to turn its early signs into sales at goal, or the Ad Set comes off. The full method is in spread vs concentrated spend and the extension ladder.
Level four: what does a runoff prove?
A runoff proves that an ad which sold well, but lost the budget fight to a sibling, can win money on its own.
Inside a testing Ad Set, one ad often takes most of the spend. The second and third place ads may have sold at goal and at low frequency, yet barely spent. They are not losers. They were outshouted.
A runoff pools several of these runners up into a fresh Ad Set with their own budget and judges them by the same rules as any test. It is used sparingly, only for ads that clearly earned it. One sale below goal, or no sales, does not earn a runoff. See runoffs: picking between two close ads.
Level five: what does an ad have to prove to be promoted?
To be promoted, an ad has to prove its sales repeat, at goal on cost, return and order value together, confirmed by a second source, over recent days.
Each part closes a door that luck could sneak through. One sale can be a fluke. One number alone can flatter an ad. Meta alone can over count or under count. And old sales say nothing about whether the ad sells now.
We also print the ad's daily frequency next to every promotion, as a flag, not a gate. Ads that sold to new people tend to get budget in Scaling. Ads that sold mostly to people who had already seen other ads tend to stall. The full checklist is in promoting a winner to Scaling.
Level six: what does an ad have to prove in Scaling?
In Scaling, an ad has to prove it can keep selling at real budget, next to established winners, on a slower clock.
This is the hardest level, and the one most people forget is a level at all. An ad that crushed a small test has not yet proven it can sell at many times that spend, to a wider mix of people, while older ads fight it for budget.
Scaling judges new arrivals over about two weeks, with early warning marks set at multiples of the Scaling Campaign's own cost per sale. A newcomer that lands a couple of sales at goal early stops being "new" and gets judged like any established ad.
Two things can go wrong here, and they are different problems:
- Starvation. The ad barely spends. It was never tested in Scaling. That is not a fail.
- The trap. The ad gets budget and does not convert. It was tested and did not hold up.
When several promoted ads starve at the same time, a runoff Ad Set inside Scaling with a small, guaranteed spend floor gives them a fair fight without taking much from the incumbents. We explain how to catch and rescue these in inside the Scaling Campaign.
What happens after an ad becomes an established winner?
An established winner's job changes from proving itself to staying alive and growing. It is now judged on cost per sale against the campaign, on fatigue signs, and on whether it still reaches new people.
This is where two later moves come in. The first is watching for winners that are quietly dying: cost per add to cart creeping up, frequency drifting higher, weekly sales slipping. That is the work of the weekly Scaling Report. The second is growing the winner on purpose, by making new versions of it for new people and new angles. We call that persona scaling.
Where does the Graveyard fit in the journey?
The Graveyard is the side exit for ads that proved they can sell, just not at a cost you can live with.
An ad can land there from the first verdict, from a runoff, or later in its life. Instead of turning it off, we move it to a cost cap campaign, where Meta only spends when it can find sales at or under the price you set. Only ads with sales go there. An ad with zero purchases has nothing for a cap to rescue. More in the Graveyard: why we keep losers, and in the Graveyard Report, which looks at what the ads that ended up there can teach you about your next round of creative.
Why does thinking in levels matter?
Because each level answers a different question, and mixing them up is how good ads die and bad ads get funded.
Promoting on one sale skips the "does it repeat?" level. Killing a starved ad skips its fair test. Calling a promoted ad a winner on day one skips the Scaling level, and plenty of testing winners never hold up at real budget.
Levels also give you a cleaner way to measure your creative program: count how many ads passed each one. That is the idea behind creative hit rate: the share of new ads that make it from testing all the way through Scaling. And when too few ads clear the top levels for too long, you have a winner gap to close.
Jake's rule of thumb: a winner is not something an ad is. It is something an ad keeps proving, one level at a time.
Key takeaways
- A winning Meta ad climbs levels: creative testing, first verdict, extension, runoff, promotion and Scaling, with the Graveyard as a side exit.
- Each level asks the ad to prove one new thing, from "it sells to strangers" to "it sells at real budget next to incumbents."
- Extensions and runoffs are fairness checks for starved ads, not rewards. Sales still decide.
- Promotion needs repeat sales at goal on cost, return and order value, confirmed by a second source.
- Scaling is its own level, judged over about two weeks. Starvation is not a fail; the trap is.
- Counting how many ads pass each level is the honest way to measure your creative program.
FAQ
How long does it take an ad to go from creative testing to Scaling? It depends on your spend and buying cycle. A strong ad on an impulse product can earn promotion within days of launch. Ads that need an extension or a runoff take longer, and every promoted ad then gets about two more weeks in Scaling before it counts as established.
Do all winning ads need a runoff? No. Most winners go straight from testing to promotion. Runoffs are for the exception: ads that sold at goal but lost the budget to a sibling in the same Ad Set.
Is an ad in the Graveyard a failure? Not exactly. It proved it can sell, just not at your target cost. Under a cost cap it can still bring in steady sales at a price you accept, which is better than throwing a working creative away.
What if my promoted ad gets no spend in Scaling? Then it has not been tested at that level yet. Check its share of spend, look at the frequency it sold at in testing, and consider a runoff Ad Set inside Scaling with a small spend floor so it gets a fair fight.
Can an ad drop back down a level? Yes. An established winner can fatigue and lose its place, and an ad that fails in Scaling can still be worth a spot in the Graveyard if it sells at some price. The levels describe what an ad has proven so far, not a permanent rank.
If you want to see where your own ads stall on this ladder, we offer a free Meta account audit. It includes a full account audit plus net new reach and cumulative reach reports, so you can see which ads are climbing and which are stuck. You can request one at our contact page.
Ready for your free Meta audit?
We will review your account and run net new reach and cumulative reach reports, free.
Check what you learned
Tap an answer to see if you got it right.
1. What is the first level a winning ad has to pass?
2. What are extensions and runoffs?
3. What is the honest way to measure your creative program?
Ready for the full test? Take the Scaling exam and get certified. Want this run on your account? Get my free audit.
Module 2: Scaling · article 1 of 11
Next: How to Tell if a Meta Ad Set Got a Real Test (Spread vs Concentrated Spend)
About the author
Jake Mehani, founder of Creative Analytics
Jake has spent 15 years running Meta and paid social ads for consumer brands. Meta published his RBX Active Advantage+ case study, he taught the DMAnc Meta and TikTok ads masterclass, and he spoke at Motion's 2024 Creative Strategy Summit. His senior team runs media buying, creative and measurement like an in house team.