Facebook ads structure for cash on delivery: a step-by-step guide

The exact Meta ads structure we run across 9 COD markets: one ad account per market, one scaling CBO, $20-30 product tests, and our weekly routine.

DFDiego FuentesCo-founder, COD Confirm
Published · 8 min read

Our Meta ads structure for cash on delivery fits in a few rules: one business manager per brand, one ad account per market, one scaling CBO per product and market, new ads added two to three times a week from an inactive upload campaign, and new products tested at $20-30 a day for two to three days in four markets. That is what runs nine markets for us without eight hours a day in Ads Manager.

This guide is for COD sellers who run, or want to run, Meta ads in several European markets at once. After reading it, you will be able to set up your accounts so a ban never stops the business, build the campaign structure we use, test a new product before buying stock, and scale what works.

Last month our brands did more than $320k. The main thing we learned to get there: the more complex your media buying structure, the harder it is to scale. The goal is not to optimise every detail. It is a structure simple enough to run many products in many markets, so your energy goes into internal processes, not into the ad account.

Step 1 — How should we structure ad accounts across several COD markets?

You will get banned on Meta. It happens even on products that are not sensitive, over and over, because Meta is unpredictable. So we build the structure so that one ban never stops everything:

LevelRule
Business managerOne per brand
Ad accountOne per market, per brand
ProductsSeveral products can share the same ad account

If three ad accounts get banned, the six other markets keep running and keep making profit. With every market in one ad account, a single ban cuts the whole business, and when you have high costs, that is dangerous. Relaunching every ad of every product takes a lot of time.

Step 2 — Keep the running business manager clean

The business manager (BM) that runs all the ad accounts must stay clean:

  • create no ad account, no page and no pixel inside the BM that runs the ads;
  • create the ad accounts in other business managers, then share them into the clean BM.

Why: if the running BM created an ad account that later gets disabled, the BM carries a red flag with Meta and can be banned too.

Step 3 — Which campaign structure works for COD?

We used to test ads in a testing CBO, then duplicate the winners into a scaling CBO. The more markets we added, the less time we had for anything else. Every time we over-complicated the structure, we paid for it. Here is what we run today:

ElementOur rule
Scaling campaignOne CBO per product, per market. Only one
Testing campaignNone. New ads go straight into the scaling CBO
Inside the CBODiversified ads: several formats and concepts, videos and statics (with Andromeda, diversity matters)
Feeding the CBONew ads duplicated in two to three times a week
Campaign namesMarket and product name in every campaign name
Creative upload campaignOne inactive campaign per product, per ad account, per market

The new ads are competitors' ads that we copied and translated, plus iterations made in-house. Feeding the CBO constantly is what keeps creative fatigue down.

Don't skip the naming convention. It looks like a detail, but it is what makes automation possible later: a script can only find and move ads if everything is named consistently.

Step 4 — How do we upload new ads without touching the scaling campaign?

With the inactive creative upload campaign:

  1. Whoever launches ads always uploads ready-to-go ads into this inactive campaign.
  2. The media buyer duplicates them into the scaling CBO.
  3. Every ad that has been duplicated is turned off in the upload campaign, so everyone knows it has already been used.

We have since automated ad launching completely. Nobody launches ads by hand for us anymore, and this upload campaign is what made the automation simple.

Step 5 — How do we test a new COD product?

We use two protocols, depending on whether we have stock.

Protocol A: test without stock. We test the product in our four best markets: Hungary, Greece, Bulgaria and Slovakia. Yours may be different, depending on your niche. Budget: $20 to $30 for two or three days. You see very fast whether the product is a winner. If it works in three or four markets, we order the stock.

Protocol B: launch when the stock lands. Once the validated product arrives at the warehouse, we create a campaign with around 20 different creatives and launch as fast as possible.

Our testing rules:

  • Order a small first batch: around 1,000 pieces, not 10,000. It also lets you measure the return rate before taking a big risk, because some products come back far more than expected. See how to reduce your COD return rate.
  • Cut after two days if results are bad. Sometimes one day is enough to see it, but we let it spend two days, because everyone has a bad day.
  • Check more than the ads. A bad test can also mean the website, the offer or the order form is not working. Look at it in Ads Manager, or with Claude connected through MCP.
  • Recover your test buyers. People who ordered while there was no stock are not lost. When the stock lands, send each of them an SMS with a promo code so they buy again. Email works too, but SMS works better. It brings the cost of testing down.

To pick your own test markets, start with the best cash-on-delivery markets in Europe.

Step 6 — How do we scale a product that works?

Copy winners across markets. Winning ads are usually universal: an ad that wins in Hungary tends to win in Slovakia, and the other way round. Not always, but most of the time. When you add new ads two or three times a week, you will forget to put a Hungarian winner into the Slovak CBO. So we have a process that scans the campaigns by ad ID and duplicates every winner missing from a market. It also stabilises results. Again: this only works with a naming convention.

Duplicate the CBO with a new page and a new funnel. This is how we sometimes double, even triple, revenue. Duplicate the scaling CBO with the same ads, but a new Facebook page and a new funnel: the URL you send people to, for example a listicle instead of the product page. A new page reaches a new audience, and Meta also analyses the landing page, so the two CBOs can run at the same time, at the same budget and the same ROAS. It does not always work: sometimes the new CBO competes with the original one. Check your metrics. For us it works pretty well.

Put the budget where the profit is. Apply 80/20 to markets. If a market is not very profitable, move the spend to the markets that bring the most profit. Always.

Step 7 — How do we keep tracking accurate on Meta?

The algorithm is only as good as the data you give it. If 20% of your sales are not tracked in Meta, it does not know which customers it acquired. We use a dedicated tracking app that can connect several pixels to one store, which matters because we run one pixel per market. When we connected every pixel through our COD order form app instead, 10 to 20% of sales never showed up in Ads Manager.

Our weekly routine

WhenWhat
Every dayCheck net margin, so you know whether the day was profitable
Two to three times a weekDuplicate new ads (copied, translated, iterated) from the upload campaign into each scaling CBO
Two to three times a weekTurn off the duplicated ads in the upload campaign
RegularlyRun the cross-market check: every winner present in every market's CBO
RegularlyMove budget from weak markets to the most profitable ones
When testing$20-30 a day for two to three days in four markets, cut bad tests after two days

At nine markets, even a simple structure means a lot of repetitive work. We connected Claude to the ad accounts through MCP to manage campaigns and duplicate ads, and we built an internal web app to manage all our brands. We used to have a big team, and it was a mess. A small team with high revenue works much better. If automation is not your strength, hire an AI developer, or hire media buyers and give them written processes.

Common mistakes

  • One ad account for every market. One ban stops everything.
  • Creating ad accounts, pages or pixels in the running BM. One disabled account and the BM itself is at risk.
  • A testing CBO plus a scaling CBO in every market. It eats the time you need for everything else.
  • No naming convention. You can't automate, and you can't find cross-market winners.
  • Ordering 10,000 pieces before you know the return rate.
  • Blaming the ads for every bad test. Check the site, the offer and the order form.
  • Losing 10 to 20% of sales in tracking. The algorithm optimises on what it sees.
  • Spending your own days in Ads Manager. Your time is worth more on new markets and new products.

Your next steps

  1. Split your setup: one business manager per brand, one ad account per market, all shared into a clean BM.
  2. Rename every campaign with market and product name, then create one inactive upload campaign per product and market.
  3. Merge your testing and scaling campaigns into one scaling CBO per product and market, fed two to three times a week.
  4. Test your next product at $20-30 a day for two to three days in your four best markets, and order a small first batch.
  5. Set up daily net margin tracking, then read scaling cash on delivery without marketing skills to free your own time.

We break this down on camera on our YouTube channel: here's my cash on delivery media buying structure (+$320k last month).

COD Confirm

Ads decide how many orders come in; confirmation decides how many get delivered. We built COD Confirm so that part runs without us: it checks every address and sends every customer an SMS in their language before the order reaches your warehouse, and it works with EU-Shipment (EuShip), PrimeCOD, FHB and HelpShip. See pricing

Confirm every COD order before it ships

Address check, one SMS in the customer's language, and a call when needed. You pay only for confirmed orders.

Keep reading