How to run and scale Facebook ads: the rules we use inside real ad accounts
How to scale Facebook and Instagram ads, which campaigns to run, when to pause an ad and what to ignore in Ads Manager. Written by AdBreakers, an ecommerce marketing agency whose founders have managed 100M+ dollars in ad spend for 100+ brands and run the accounts themselves.
Updated November 2026 · By AdBreakers · See our Meta ads agency page
How do you scale Facebook ads?
There are two ways to scale Facebook ads: add more products and angles, or put more budget and more creatives behind the one that is already proven. We do both, in that order of safety. The bar is always the same: ROAS against the break-even ROAS of the store. If the return is above break-even we raise the budget in steps and watch it daily. If it is below, we do not scale, we fix the creative or the page first.
What is the best way to scale a Facebook ads budget without losing ROAS?
Raise the budget gradually, about 20% at a time, and only raise again inside 48 hours if the last raise held its results. Bigger jumps of 50% or even double are for campaigns that are clearly far above break-even. When a budget goes up too fast, Meta has to go past the people who convert best and spend the extra money on colder people, so the cost per sale rises. If that happens we bring the budget back to the level that worked.
Are Facebook ads worth it for an ecommerce brand?
Facebook ads are worth it when the return stays above the break-even ROAS of the store, and that number comes from the margin. A store that keeps 40% of each sale after product and shipping costs breaks even at a 2.5x ROAS: at that point it is not losing money on ads, but it is not thriving either. We judge every account on that one number, at real spend. Cheap clicks, low CPMs and high click rates do not make ads worth it if the sales are not there.
Are Instagram ads worth it, or should I only run Facebook ads?
Instagram ads are worth it for the same reason Facebook ads are: both are bought in one auction through Meta Ads Manager, and Meta moves the money to wherever the buyers are. We launch every campaign on all placements by default, feeds, Stories and Reels on both apps, and let the purchase data decide the split. Choosing Instagram only or feed only usually just removes buyers the system would have found.
What is a Facebook ads agency and what should it do?
A Facebook ads agency runs a brand's paid advertising on Facebook and Instagram: it builds the campaigns, makes or directs the ad creative, decides budgets and reports the return. In our view a good one does three things most do not: it judges everything on ROAS against the store's break-even, it ships enough genuinely different creatives to compete in the auction, and it checks the store's own sales numbers before giving a verdict. At AdBreakers the founders do this work themselves.
What drives the cost of a Facebook ads agency?
Three things drive what a Facebook ads agency charges: how much ad spend it manages, how much creative it produces, and who actually works in the account. An agency that only adjusts budgets is cheaper than one that also makes the statics, videos and copy, but creative is the main lever on Meta, so the cheaper option often costs more in lost sales. AdBreakers charges a transparent monthly fee, sometimes with a performance component, with no long contracts.
What happens when you duplicate a Facebook ad?
A duplicated Facebook ad is a new ad: it starts with no delivery history, and the original keeps everything it has learned. That is why we use duplication for changes. When a winning ad needs a correction, we do not edit it, because an edit resets the ad. We duplicate it, make the change in the copy, and leave the original running next to it. If the new one wins, good. If not, the seller never stopped selling.
Should you edit or delete a Facebook ad that is working?
No. An ad that is working is not edited, closed or deleted. An edit resets the ad, so a correction goes into a new ad beside it while the seller stays on. A low quality ranking is not a reason to close an ad that converts either. We also never delete campaigns or ad sets: deleting throws away the spend history and the social proof for good. When something has to stop, it is paused, and anything rebuilt is built next to the old one first.
Why are my Facebook ads not working?
When Facebook ads are not working, we usually find one of four causes. The campaign is optimised for something other than purchases. There are too few different creatives, often three or four, against competitors running dozens. The ads send people to a collection or home page instead of the product page. Or the ads were judged before they had real spend. Before any verdict we also compare Meta's numbers with the store's own orders for the same days, because Ads Manager alone can mislead.
Which campaign objective should an ecommerce store use on Meta?
Purchase, always. The optimisation event is the targeting: when you ask Meta for website purchases, it looks for people who buy online. Ask for add to carts, link clicks or awareness and it finds people who do exactly that and nothing more. In one account we audited, an add-to-cart campaign looked cheap per cart but needed about 100 carts for one purchase. The only exceptions we accept are a deliberate funnel test or a large brand running an awareness budget the way it would buy television.
Should I switch to add to cart or initiate checkout when purchases are slow?
No. Switching the optimisation event from purchase to add to cart or initiate checkout because sales are slow is common advice, and in our experience across many ad accounts it does not work. Meta gives you what you ask for, so you get people who add to cart and leave. The cost per cart looks good and the return collapses. If purchases are slow, the problem is the creative, the offer or the page, and that is where the work goes.
Is broad targeting better than interest targeting on Meta?
Yes, for almost every ecommerce account we manage. Meta works best with a broad audience, and narrow interest stacks stop the algorithm from doing its job. Today the creative does the targeting: you reach a demographic with what the ad shows and says, not with the ad set settings. Splitting a campaign into five audiences is the old way. Five ad sets with five different angles for one product is how accounts scale.
How should a Meta ad account be structured for ecommerce?
We recommend one campaign per product, with each ad set built around a different angle and several creatives inside it. That said, structure is the least important part of an account. One campaign with one ad set is not a problem if the money reaches the good creatives. Structure only matters when it keeps budget away from winners. Things we never treat as findings on their own: the number of ads per ad set, campaign budget against ad set budget, naming, or learning-phase warnings.
How do you test new ad creatives on Meta?
Put the new creatives together in one ad set, 5 to 10 of them, with broad targeting and purchase optimisation, and change one thing at a time. A test that changes the product and the audience at once teaches nothing. For a pure concept test we even keep the same primary text and headline on every ad, so only the creative differs. What we avoid is one creative per ad set on a small budget: the spend gets split so thin that nothing receives enough to be judged.
How much do you need to spend before judging a Facebook ad?
Enough for a sale to be possible. An ad with a few dollars of spend cannot be judged, whether it shows a purchase or not. Our rule of thumb is the order value divided by the target ROAS: that is the spend one break-even sale needs. For a 3,000 dollar product at a 3x target, that is about 1,000 dollars per sale, so 250 dollars with no sales is not a failure, it is no data. For videos with real spend, a click rate above 2% is a good sign.
When should you pause a Facebook ad?
Pause an ad when it takes real money and stays below break-even or below the account average. Our line is spend of twice the break-even cost per sale with no sales. An ad that gets no spend is left alone, because it costs nothing. We read fatigue as 14 days in a row under the bar, not 7. And track record counts: a long-time winner with a weak fortnight gets its budget trimmed and fresh creatives of the same angle, while a new test that fails is closed.
Do I need a separate retargeting campaign on Meta?
Usually not. Meta is an acquisition channel, and on a small account the retargeting pool is too small to deserve its own campaign: a broad purchase campaign reaches recent visitors anyway. We also do not swing to the other extreme. In one account, excluding every past buyer took existing customers from about 20% of spend to almost zero, and the healthier level is a band of around 10%. A customer from two years ago otherwise has to be bought again at full price.
Does a high ROAS on a small budget mean my Facebook ads are working?
Not necessarily. A very good ROAS on tiny spend is often a sign that Meta is only reaching people who already knew the brand: visitors, followers and past buyers. The test is what happens to the return when spend goes up. If the ROAS only looks good in the low-spend months, the ads are being credited for customers you already had. Real proof is a return that holds while the budget grows and new customers come in.
Do CPM, CTR and frequency matter in Facebook ads?
Much less than people think. CPM, click rate, cost per click and frequency are all already calculated into ROAS, so ROAS against break-even is the number we look at. A cheap CPM with no sales is worse than an expensive one with sales, because it means the wrong audience. We do not report frequency at all: an ad is retired when its return decays over weeks, not when a frequency number crosses a line. The one place click rate helps is diagnosing a weak video hook.
Should I lower my ad budget after one bad day?
No. One bad day is not a trend, and daily numbers always move up and down. We look at the week, not the day, and compare the last 7 days before touching anything. If a weak trend continues, we lower the spend a little on the products that are slipping instead of cutting the account. The same patience applies in reverse: budgets go up in steps, so the campaign keeps what it has learned.
Should I narrow my Facebook ads to certain ages or regions?
Almost never. Cutting regions down to the ones that feel richer, or narrowing ages to a guess about the customer, takes away the room Meta needs to find buyers, especially on a small budget. The regions you would cut cost little, and the system already spends where sales happen. The one age change that makes sense in our accounts is removing 65 and over when that group takes the spend without buying. Everything else stays broad.
Should I trust the ROAS that Meta Ads Manager reports?
Use it, but never alone. Meta can over-report and it can under-report. Before we recommend a cut, a pause or a scale, we divide the store's real revenue by the ad spend for the same days and compare that blended return with what Meta claims. We have seen Meta show a 1.4x ROAS while the store showed 2.6x on the same spend. Pausing on the Meta number alone would have switched off ads that were paying for themselves.
