Small budget runs great, then you scale and it all falls apart :same old story

YUNJI Agency

Newbie
Joined
Jul 10, 2026
Messages
18
Reaction score
5
Seen this play out so many times managing accounts. Small budget, numbers look solid, you say push it, budget goes up, CPL spikes, conversions dry up, you're confused asking if the ad's broken

Took a while to get it , that early "good" performance is often a mirage. Algorithm just landed on a small pocket of easy converters and kept grinding on them at low spend. Scale up and it has to look outside that pocket, and quality shows fast

Now we check two things first: did CPM spike, or did CPM barely move but conversion rate dropped on its own. Totally different problems, gotta separate them before touching anything

Biggest lesson: you love changing budget, creative, and targeting all at once. Then nobody can tell what broke it. One variable at a time now, even if slower

anyone else run into this before scaling their next campaign?
 
Why don't you do multiple small budgets runs instead of one big one that fails. Success leaves clues, stick to it.
 
Why don't you do multiple small budgets runs instead of one big one that fails. Success leaves clues, stick to it.
exactly my point i dont just jump into bigger budget campaign like that run some testing with smaller budget and check what works for your niche
 
exactly my point i dont just jump into bigger budget campaign like that run some testing with smaller budget and check what works for your niche
Nice, after that then you upscale. Neat delivery too
 
This exact question has been making rounds for a full year now.

People love to blame meta or think that something is broken but the way meta works now is based on angles and variations.

If you are selling a pillow for example,

And your setup and ads just point out it's a pillow, good price, good quality it's not going to go anywhere, as you said it might pick up a handful of buyers in a super small sized audience but it will be impossible to scale outside that.

It's just how meta works now.

What you want to do instead:

Write down at least 4 different angles and variations and if you are using avatars make sure to switch them up.

Examples:

1: My skin has never looked better after 3 nights sleeping with this pillow (Skin care, acne angle)

2: I would always wake up still feeling tired but due to the XYZ fabric from Tibet, I finally wake up energized.

3: My anxiety almost completely vanished after a week sleeping with this pillow versus my Ikea ones, turns out the silk used blabla has effect on XYZ

4: Struggling to fall a sleep? Is your mind still raising, this pillow is for you.


What this does is allowed your campaigns and adsets to reach multiple audiences.

Anxiety, Skin care, Bad sleepers and so on.

Top, Mid and Bottom funnel should all be in one campaign.


People that are still not convinced should also be shown ads with for example proof.

So on top of the variations, show ads with reviews, ratings, social proof and so on.

I am NOT saying you can't sell pillows just based on them looking good and are nicely priced, it's just that meta wants ads that can resonate with a much broader audience before you can scale, and those ads just don't cut it anymore.

Avatars AKA the if you are running reels you might want to NOT just use a white older lady but mix it up with ages, race, possible a doctor of sorts as a young person might resonate so much more with someone closer to their age, or a older person might not resonate with someone younger and so on.

And with AI it's now easier than ever to get these angles, variations and in general all these ads together.

Meta now basically runs your entire funnel on their platform while before your entire funnel used to mostly be your landing page.

Provide meta what they want instead of going against it.

In most cases the reality = YOUR ADS SUCK & YOU ARE BEING LAZY.

Might sound harsh but it's just facts.

Start with ABO setups, and test as much as possible but don't touch touch and touch get some solid data first, move your winners to a new campaign (CBO) and start that with a much higher budget and keep scaling that while still adding new ads in your ABO (Prospecting is what I like to call it) and rinse and repeat,

Might sound like a lot of work but once you get the hang of it, it's honestly not at all.

You also don't need thousands of ads a month but at least keep adding 20 or up a month or so and base them of the angle and variations that are doing the best.


BUT obviously there are also still so much other things you can do such as using RULES, decreasing your spend on ages, placements, genders that don't work etc

Duplication still works, it's weird but still does.

Make sure you create custom audiences that you can exclude as soon as possible too.

SCALE.
 
To understand why this happens you need first to answer to this question:
Why meta ads exist and what is their primary goal?

- Making money (same as our goal).

That means they must make you spend as much as possible (they don't care are you ROI positive or not, the goal is to make you spend).

How is this working?
1. you create a campaign
2. it brings your high intent people
3. you love the results and increase budget
4. meta widens the audience to whom they show your ads, that means it will now include low intent, medium intent too... (It is not a secret, but a documented fact, that meta, google, microsoft have a profile for each user and score them by their ads interaction and etc...)
5. your ROI drop

But this is not a big problem if you continue to be ROI positive.

Most popular strategy is to:

create many small campaigns with low budge, ROI positive ones keep increasing budget slowly and remove the negative ones, do this until your ROI drops to an acceptable one compared to the volume you get.
 
yeah that's basically what we do now, just clone the winning ad set and run a few side by side instead of dumping more budget into one. keeps the working pocket intact instead of forcing it to hunt for new audience overnight
Why don't you do multiple small budgets runs instead of one big one that fails. Success leaves clues, stick to it.

exactly my point i dont just jump into bigger budget campaign like that run some testing with smaller budget and check what works for your niche
yeah exactly, and once you find that smaller budget that's actually working, don't just crank it up either, clone it a few times instead so you're not messing with the thing that's already working

This exact question has been making rounds for a full year now.

People love to blame meta or think that something is broken but the way meta works now is based on angles and variations.

If you are selling a pillow for example,

And your setup and ads just point out it's a pillow, good price, good quality it's not going to go anywhere, as you said it might pick up a handful of buyers in a super small sized audience but it will be impossible to scale outside that.

It's just how meta works now.

What you want to do instead:

Write down at least 4 different angles and variations and if you are using avatars make sure to switch them up.

Examples:

1: My skin has never looked better after 3 nights sleeping with this pillow (Skin care, acne angle)

2: I would always wake up still feeling tired but due to the XYZ fabric from Tibet, I finally wake up energized.

3: My anxiety almost completely vanished after a week sleeping with this pillow versus my Ikea ones, turns out the silk used blabla has effect on XYZ

4: Struggling to fall a sleep? Is your mind still raising, this pillow is for you.


What this does is allowed your campaigns and adsets to reach multiple audiences.

Anxiety, Skin care, Bad sleepers and so on.

Top, Mid and Bottom funnel should all be in one campaign.


People that are still not convinced should also be shown ads with for example proof.

So on top of the variations, show ads with reviews, ratings, social proof and so on.

I am NOT saying you can't sell pillows just based on them looking good and are nicely priced, it's just that meta wants ads that can resonate with a much broader audience before you can scale, and those ads just don't cut it anymore.

Avatars AKA the if you are running reels you might want to NOT just use a white older lady but mix it up with ages, race, possible a doctor of sorts as a young person might resonate so much more with someone closer to their age, or a older person might not resonate with someone younger and so on.

And with AI it's now easier than ever to get these angles, variations and in general all these ads together.

Meta now basically runs your entire funnel on their platform while before your entire funnel used to mostly be your landing page.

Provide meta what they want instead of going against it.

In most cases the reality = YOUR ADS SUCK & YOU ARE BEING LAZY.

Might sound harsh but it's just facts.

Start with ABO setups, and test as much as possible but don't touch touch and touch get some solid data first, move your winners to a new campaign (CBO) and start that with a much higher budget and keep scaling that while still adding new ads in your ABO (Prospecting is what I like to call it) and rinse and repeat,

Might sound like a lot of work but once you get the hang of it, it's honestly not at all.

You also don't need thousands of ads a month but at least keep adding 20 or up a month or so and base them of the angle and variations that are doing the best.


BUT obviously there are also still so much other things you can do such as using RULES, decreasing your spend on ages, placements, genders that don't work etc

Duplication still works, it's weird but still does.

Make sure you create custom audiences that you can exclude as soon as possible too.

SCALE.
yeah the ABO to CBO thing is pretty much what we do too, but honestly the 20 ads a month only helps if you're actually tagging which angle each one's testing, otherwise it's just a pile of creative with no clue what's working. and lol duplication being "weird but still works" is so true, never figured out why that holds up either

To understand why this happens you need first to answer to this question:
Why meta ads exist and what is their primary goal?

- Making money (same as our goal).

That means they must make you spend as much as possible (they don't care are you ROI positive or not, the goal is to make you spend).

How is this working?
1. you create a campaign
2. it brings your high intent people
3. you love the results and increase budget
4. meta widens the audience to whom they show your ads, that means it will now include low intent, medium intent too... (It is not a secret, but a documented fact, that meta, google, microsoft have a profile for each user and score them by their ads interaction and etc...)
5. your ROI drop

But this is not a big problem if you continue to be ROI positive.

Most popular strategy is to:

create many small campaigns with low budge, ROI positive ones keep increasing budget slowly and remove the negative ones, do this until your ROI drops to an acceptable one compared to the volume you get.
yeah this is basically why we stopped chasing "acceptable ROI" as one number and just watch when the intent pool starts thinning out instead. once meta's clearly reaching past your actual buyers you're just paying for reach at that point, not conversions
 
Last edited by a moderator:
Yeah this happens a lot scaling usually exposes whether the campaign is actually stable or just benefiting from a small audience pocket.
 
This exact question has been making rounds for a full year now.

People love to blame meta or think that something is broken but the way meta works now is based on angles and variations.

If you are selling a pillow for example,

And your setup and ads just point out it's a pillow, good price, good quality it's not going to go anywhere, as you said it might pick up a handful of buyers in a super small sized audience but it will be impossible to scale outside that.

It's just how meta works now.

What you want to do instead:

Write down at least 4 different angles and variations and if you are using avatars make sure to switch them up.

Examples:

1: My skin has never looked better after 3 nights sleeping with this pillow (Skin care, acne angle)

2: I would always wake up still feeling tired but due to the XYZ fabric from Tibet, I finally wake up energized.

3: My anxiety almost completely vanished after a week sleeping with this pillow versus my Ikea ones, turns out the silk used blabla has effect on XYZ

4: Struggling to fall a sleep? Is your mind still raising, this pillow is for you.


What this does is allowed your campaigns and adsets to reach multiple audiences.

Anxiety, Skin care, Bad sleepers and so on.

Top, Mid and Bottom funnel should all be in one campaign.


People that are still not convinced should also be shown ads with for example proof.

So on top of the variations, show ads with reviews, ratings, social proof and so on.

I am NOT saying you can't sell pillows just based on them looking good and are nicely priced, it's just that meta wants ads that can resonate with a much broader audience before you can scale, and those ads just don't cut it anymore.

Avatars AKA the if you are running reels you might want to NOT just use a white older lady but mix it up with ages, race, possible a doctor of sorts as a young person might resonate so much more with someone closer to their age, or a older person might not resonate with someone younger and so on.

And with AI it's now easier than ever to get these angles, variations and in general all these ads together.

Meta now basically runs your entire funnel on their platform while before your entire funnel used to mostly be your landing page.

Provide meta what they want instead of going against it.

In most cases the reality = YOUR ADS SUCK & YOU ARE BEING LAZY.

Might sound harsh but it's just facts.

Start with ABO setups, and test as much as possible but don't touch touch and touch get some solid data first, move your winners to a new campaign (CBO) and start that with a much higher budget and keep scaling that while still adding new ads in your ABO (Prospecting is what I like to call it) and rinse and repeat,

Might sound like a lot of work but once you get the hang of it, it's honestly not at all.

You also don't need thousands of ads a month but at least keep adding 20 or up a month or so and base them of the angle and variations that are doing the best.


BUT obviously there are also still so much other things you can do such as using RULES, decreasing your spend on ages, placements, genders that don't work etc

Duplication still works, it's weird but still does.

Make sure you create custom audiences that you can exclude as soon as possible too.

SCALE.
Yes you've gotta test multiple variations for the same ad, and see how it goes before you even scale it. Also you need to be solving people's problems, otherwise people won't buy your product/service.
 
Yes, you should warm up your account first with small budgets to test it out. If the small budget works well, then you can gradually expand it by 20-30% until you reach your desired budget level. That's the campaign strategy I usually use.
 
I completely agree with this perspective! I always break down my initial budget into smaller amounts to test different advertising approaches, making sure they resonate with consumers, and then gradually increase spending when things start to become profitable.
 
Back
Top