A Facebook campaign that produces orders but erodes cash is not working. Established eCommerce brands need to launch profitable Facebook campaigns against a commercial target, not a platform metric. That means knowing what an acquisition is worth before Meta spends the first pound, then building a testing process that separates genuine demand from expensive noise.
Clicks, reach and even revenue can look impressive while contribution margin disappears. The difference between a scalable account and a costly one is usually not a clever interest audience. It is disciplined economics, clean measurement and the willingness to cut what does not earn its place.
Start with the number that protects profit
Before campaign structure, creative or audience selection, establish your break-even cost per acquisition and your target cost per acquisition. These are not interchangeable.
Your break-even CPA is the most you can spend to acquire a customer before that first order stops contributing to profit. It must account for product cost, delivery, payment fees, returns, pick-and-pack costs, discounts and the margin you need to retain. A target CPA sits below that ceiling, leaving room for volatility and profit.
For brands tracking return on ad spend, calculate the equivalent break-even ROAS as well. If your average order value is £80 and your acceptable acquisition cost is £24, your break-even ROAS is 3.33. If you need a meaningful contribution after advertising, your working target may be 4.0 or higher.
The correct target depends on your business model. A high-repeat-purchase brand may sensibly accept a lower first-order ROAS if cohort data proves customers reorder at healthy rates. A low-repeat, seasonal or heavily returned product needs stricter first-order economics. Do not use lifetime value as permission to lose money unless retention data can support it.
Make tracking credible before you spend
Meta cannot optimise reliably around data it cannot see. The purchase event needs to fire consistently, with correct values and currency, and deduplication needs to be in place where browser and server events overlap. Test a real checkout journey rather than trusting a green status in Events Manager.
For eCommerce brands, the priority event should normally be Purchase, with Add to Cart and Initiate Checkout retained as diagnostic signals rather than success metrics. A campaign may generate plenty of carts and still fail at checkout because of delivery costs, payment friction or an offer that does not hold up under scrutiny.
Attribution is useful, but it is not a profit and loss statement. Meta may claim credit for sales that would have happened through email, branded search or direct traffic. Compare platform reporting with your store data, blended revenue, new-customer rate and contribution margin. If the account scales while blended performance worsens, the platform result is not as strong as it appears.
Use UTMs consistently and keep a simple reporting view that shows spend, attributed revenue, blended revenue, CPA, ROAS, average order value and margin. This gives your team one version of the truth when decisions need to be made quickly.
Build a campaign structure that can learn
Many accounts are overbuilt from day one. Multiple campaigns, tiny ad sets and narrow audiences spread early data too thinly. Meta needs enough conversion signal to understand who is likely to buy. Fragmentation makes that harder and gives advertisers more places to hide poor performance.
For most established brands, begin with a simple prospecting campaign optimised for purchases, supported by a separate retargeting campaign once traffic volume justifies it. Prospecting should reach broad eligible audiences, with only sensible market, age or exclusion controls. Retargeting should exclude recent purchasers where repeat buying is not immediately relevant and should be kept proportionate to the size of the warm audience.
There are exceptions. A brand selling distinct product categories with radically different price points, margins or buyers may need separate campaigns. A retailer with a large catalogue may need product-set segmentation. But every split should have a commercial reason and enough budget to produce meaningful learning.
Campaign budget optimisation can work well once you have proven ads and sufficient conversion volume. At the start, ad-set budgets may give more control over a deliberate test. The point is not to follow a fixed rule. It is to prevent Meta from spending most of the budget on a weak idea before you have enough evidence to judge it.
Creative does the heavy lifting
Targeting has become less granular. Creative now carries more of the work of qualifying the right buyer, explaining the product and giving Meta signals about who responds.
A profitable creative test is not a collection of cosmetic variations. Changing a background colour or moving a logo rarely answers a useful question. Test angles that reflect real purchase motivations: problem and solution, product demonstration, social proof, comparison, gifting, quality, convenience, price justification or a specific use case.
For a premium product, lead with the proof behind the price rather than a discount. For a replenishable product, demonstrate the routine or result that makes another purchase likely. For fashion, show fit, movement and scale rather than relying on polished stills alone. The creative must make the product easier to buy, not merely easier to scroll past.
Give each test enough time and spend to be judged against your target CPA. The exact threshold depends on average order value, conversion volume and volatility, but killing an ad after a handful of clicks is not optimisation. Equally, allowing a poor ad to consume several times your target CPA because it has a high click-through rate is not discipline.
Watch the relationship between metrics. Strong outbound click-through rate with weak landing-page conversion points to a message-to-page mismatch, pricing concern or site issue. Weak click-through rate and weak conversion usually means the angle is not landing. Good CPA but falling average order value can still damage your overall ROAS.
Fix the destination, not just the advert
Facebook ads amplify the commercial reality of the website. They cannot compensate indefinitely for slow mobile pages, unclear delivery information, weak product pages or a checkout that feels risky.
Before increasing spend, inspect the mobile purchase journey. Product pages should answer the objections a buyer has before they need to ask: what is included, why it is different, when it will arrive, how it fits or works, and what happens if it is not right. Reviews, user-generated content and clear returns information matter most when the product requires trust.
Offer strategy also needs restraint. A permanent discount may improve reported conversion rate while training customers to wait and compressing margin. Test bundles, threshold-based incentives, gifts or value-led messaging before defaulting to money off. The best offer is the one that improves conversion without creating a profitability problem next month.
Scale only after the result repeats
One profitable day is not a scale signal. Look for repeated performance through normal demand cycles, across several attribution windows and against blended business data. A campaign with a £20 CPA against a £25 target is promising; it is not automatically ready for a 50% budget increase.
Scale in controlled increments, often around 10-20% at a time, then allow the account to stabilise before making the next decision. Bigger changes can work, particularly where conversion volume is high and demand is proven, but they create more volatility. If you double spend and CPA rises beyond your target, you have bought revenue at a worse rate, not necessarily created profitable growth.
Keep a pipeline of new creative while you scale. Fatigue is not always visible as a dramatic collapse. Frequency may rise, click-through rate may soften, or conversion quality may decline as the same message reaches less responsive people. Fresh angles protect performance better than endlessly adjusting audiences.
Know when to intervene and when to wait
The hardest part of paid social management is resisting activity for its own sake. Do not reset a campaign every morning because yesterday was expensive. Facebook performance moves with day of week, pay cycles, creative delivery and conversion lag. Make decisions using a defined evaluation window that matches your volume.
Intervene when the evidence is clear: CPA is consistently above target, a creative has spent enough without producing viable purchases, landing-page conversion has deteriorated, or reporting exposes a tracking fault. Wait when conversion volume is low and the data is genuinely inconclusive. Patience is not passive when it is tied to a clear decision rule.
A profitable Facebook account is built through repeated commercial judgement, not platform superstition. If the numbers are unclear, tighten the measurement. If the creative is tired, test a new buying angle. If the margin does not support the acquisition cost, fix the offer or do not scale. That is how spend becomes an asset rather than another line of wasted budget.