The Shopify media buying playbook
How media buyers run paid ads for a Shopify store in 2026, stage by stage, with the numbers to decide by and a calculator for cash-on-delivery stores in Egypt.
The recipe in one line
Know your break-even, track every order server-side, feed the algorithm many different creatives, test cheaply, move winners into one broad scaling campaign, cut losers by rule, and repeat weekly. In 2026 the creative is the targeting: Meta's ad system (Andromeda) finds buyers from what the ad shows, not from interest boxes.
1. Mathbreak-even ROAS and CPA
2. Trackingpixel + Conversions API
3. Creativedistinct concepts weekly
4. Structuretest, scale, retarget
5. Test and killrules, not feelings
6. Scalebudget, duplicates, iterations
7. Loopdaily, weekly, monthly
| Stage | Goal | What you do | Output | Time to set up |
|---|---|---|---|---|
| 1. Math | Know what you can pay per order | Margin per order, break-even ROAS and CPA, target CPA, COD adjustment | One sheet of numbers | 1 hour |
| 2. Tracking | Meta and TikTok see every purchase | Shopify apps for Meta (Maximum data sharing), TikTok and Google; UTMs; domain check | Events firing, deduplicated | Half a day |
| 3. Creative | Enough different ads to learn from | Angle matrix, 4 to 6 concepts per test, Arabic hooks | A creative tracker | 2 to 3 days for the first batch |
| 4. Structure | A clean account | Testing campaign, Advantage+ sales campaign, retargeting with a cap | 3 campaigns | 2 hours |
| 5. Test and kill | Find winners cheaply | One concept per ad set, judge by rules, automated rules in Ads Manager | A test log | Ongoing, 3 to 7 days a round |
| 6. Scale | More orders at the same CPA | Budget steps, duplicates, new hooks on winners | Growing spend | Ongoing |
| 7. Loop | Keep it healthy | Daily checks, weekly creative and margin review, monthly offer test | A weekly report | 30 minutes a day |
1Math before money
Every rule later in the playbook keys off one number: the target CPA. Work it out first.
| Number | Formula | Why it matters |
|---|---|---|
| Margin per delivered order | price − product cost − shipping you pay − payment or COD fee | What one delivered order earns before ads |
| Delivered share | confirmation rate × delivery rate | COD orders that turn into money |
| Profit per placed order | delivered share × margin − shipped share × failed share × return cost − confirmation cost | What a Meta "Purchase" is really worth on COD |
| Break-even CPA | = profit per placed order | Pay more than this per purchase and you lose money |
| Break-even ROAS | order value ÷ break-even CPA | The ROAS Meta must show for you to break even |
| Target CPA | break-even CPA × (1 − profit share you want) | The number every test is judged against |
| MER (blended) | all store revenue ÷ all ad spend | The truth check above platform-reported ROAS |
On cash on delivery, Meta counts a "Purchase" when the order is placed, before anyone confirms or pays. Judge by cost per delivered order, not by Meta's ROAS.
Break-even calculator
The inputs below are example numbers only, to show how it works. Replace every one with your real prices, costs and rates. Amounts in EGP, one item per order.
2Tracking on Shopify
| Piece | How | Check |
|---|---|---|
| Meta pixel and Conversions API | Install the Facebook & Instagram app, connect the business portfolio, ad account, pixel and catalog. Set data sharing to Maximum: pixel plus Conversions API, server to server, not blocked by ad blockers. | Events Manager shows browser and server events, deduplicated |
| Events tracked | PageView, ViewContent, Search, AddToCart, InitiateCheckout, AddPaymentInfo, Purchase come from the app | Test one order end to end |
| UTMs | On every ad: utm_source=facebook&utm_medium=paid&utm_campaign={{campaign.name}}&utm_content={{ad.name}} | Shopify's marketing reports show the campaign |
| TikTok | The TikTok app on Shopify installs the pixel, the Events API and the catalog. TikTok says the app is not offered in every market: check it lists Egypt. | TikTok Events Manager |
| Google & YouTube app: Merchant Center feed for Shopping and Performance Max | Products approved in Merchant Center | |
| Delivered orders | Tag orders in Shopify when confirmed and when delivered; cancel unconfirmed ones so reports stay true | Weekly delivered count per campaign |
Privacy: Enhanced and Maximum send name, location, email and phone to Meta for matching. Keep your privacy policy current.
3Creative: the real targeting
Meta treats ads that look or feel alike as one creative. A new concept means a new persona, angle, offer or format, not a new color or font.
Angle matrix for a garment store
| Format | What it shows | Hook idea (first 3 seconds) |
|---|---|---|
| Try-on video | A real person wearing it, turning, moving | "Ordered my size, here is the fit" |
| Fabric close-up | Weight, stitching, print quality, after a wash | Stretch or scratch the print on camera |
| Use case | Team shirts, family event, company uniform, group trip | "30 shirts for our team in 5 days" |
| Design it yourself | Screen recording of customizing the design | "Your name on it in 20 seconds" |
| Offer static | Price, bundle, free delivery, pay on delivery | "Pay when it reaches you" |
| Social proof | Reviews, unboxing, WhatsApp screenshots (with consent) | A customer's own words |
- Per test: 4 to 6 genuinely different concepts rule of thumb. Sources range from 4 to 20.
- Inside one concept: the 3:2:2 pattern, 3 visuals, 2 primary texts, 2 headlines in one flexible ad rule of thumb.
- Language: Egyptian Arabic text on screen and in voice; English as a second test.
- Cadence: new concepts every week; refresh when frequency climbs and CTR falls.
4Account structure
| Campaign | Type | Budget share | Audience | What goes in |
|---|---|---|---|---|
| Testing | Sales, budget per ad set | 20 to 30% rule of thumb | Broad: country, age, language | One concept per ad set, so results stay readable |
| Scaling | Advantage+ sales campaign (was "Advantage+ shopping") | 60 to 70% rule of thumb | Broad, Meta decides | Winners from testing, moved by post ID to keep their likes and comments |
| Existing customers and visitors | Separate ad set with a spend cap, plus catalog ads | 10 to 20% rule of thumb | Customer list, site visitors, add-to-carts | Reviews, new arrivals, reorder offers |
One source reports new Advantage+ sales campaigns no longer cap spend on existing customers, so exclude your customer list there and give customers their own capped ad set. Check this in Ads Manager.
5Test and kill by rule
An ad set needs about 50 optimization events in 7 days to leave the learning phase; big edits restart it. Most small stores can't afford 50 a week per ad set, so judge early signals first and purchases second.
| When | Rule | Action |
|---|---|---|
| Spend reaches 1× target CPA | No add-to-cart, or CTR far below your account's usual | Pause the ad rule of thumb |
| Spend reaches 2× target CPA | No purchase | Pause the ad or ad set rule of thumb |
| After 3 to 7 days | CPA above 1.3× target | Pause; note what it taught you rule of thumb |
| After 3 to 7 days | CPA at or under target, at least a few purchases | Graduate to the scaling campaign |
| During learning | Any urge to edit | Wait. Edits to budget, audience, event or creative can reset learning |
Put the pause rules into Ads Manager's automated rules (free) so they run while you sleep.
6Scale
| Lever | How | Stop when |
|---|---|---|
| Vertical (budget) | +20 to 30% every 2 to 3 days while CPA holds rule of thumb | CPA rises above target for 2 days |
| Horizontal (duplicates) | Copy a winner into a new ad set or campaign instead of one big budget jump | Duplicates compete and CPM rises |
| Creative iteration | New hooks, new first 3 seconds, new person on a winning body | The angle stops beating the original |
| New markets | Same winners in Gulf countries with card payment | Delivery or payment economics differ: redo the math |
| New channels | TikTok Smart+ web campaigns, Google Performance Max | Blended MER falls below break-even |
TikTok's GMV Max runs only for TikTok Shop sellers; a Shopify store uses web campaigns with the TikTok pixel.
7The operating loop
| Cadence | Check | Act | Time |
|---|---|---|---|
| Daily | Spend pacing, CPA per ad set, kill rules fired, yesterday's confirmation rate | Pause losers, call or WhatsApp unconfirmed orders | 15 to 30 min |
| Every 2 to 3 days | Winners holding target CPA | Budget step on scaling campaign | 10 min |
| Weekly | Cost per delivered order, MER, frequency and CTR trend per ad, creative tracker | Launch 4 to 6 new concepts, retire tired ads, write the weekly report | 2 to 3 hours |
| Monthly | Margin after returns, repeat buyers, ROAS by product | Test an offer (bundle, free delivery), drop products that lose money after returns | Half a day |
Metrics: what each one tells you
| Metric | Formula | Reads as |
|---|---|---|
| Hook rate | 3-second video plays ÷ impressions | Do the first 3 seconds stop the scroll |
| Hold rate | ThruPlays ÷ 3-second plays | Does the video keep them |
| Link CTR | link clicks ÷ impressions | Does the ad make them want to see more |
| CPM | spend ÷ impressions × 1,000 | How expensive the audience is |
| Add-to-cart rate | add-to-carts ÷ link clicks | Does the product page sell |
| Conversion rate | purchases ÷ link clicks | Does checkout work |
| CPA | spend ÷ purchases | Against target CPA |
| ROAS | purchase value ÷ spend | Against break-even ROAS (platform view) |
| MER | all revenue ÷ all ad spend | The business view, all channels |
| Frequency | impressions ÷ reach | Rising with falling CTR means fatigue |
| Confirmation rate | confirmed ÷ placed orders | Order quality from each campaign (COD) |
| Delivery rate | delivered ÷ shipped | Refusals and failed deliveries (COD) |
| Cost per delivered order | spend ÷ delivered orders | The real CPA on COD |
No universal benchmark is given: compare each ad against your own account's median.
Egypt notes
| Topic | What the sources say | What to do |
|---|---|---|
| Cash on delivery | Estimates put COD at 40 to 70% of Egyptian online orders | Build the COD rows of the calculator first |
| Returns | Unconfirmed COD sees 30%+ returns; confirming every order brings it to 10 to 15% (vendor claims); other sources say 15 to 35% | Confirm every order by call or WhatsApp before shipping |
| Measure | Some traffic sources bring more refusals than others | Track confirmation and delivery rate per campaign |
| VAT on ads | Digital advertising is taxable at 14%; non-resident platforms collect VAT from unregistered advertisers | Add your VAT number in Meta billing if registered; budget 14% on top if not |
| Channels | Facebook and Instagram bring most early traction for Egyptian stores | Start on Meta; add TikTok once Meta works |
Build it, step by step
| When | Step | Done when |
|---|---|---|
| Day 1 | Fill the calculator with real numbers; write down target CPA and the kill thresholds | One sheet with 5 numbers |
| Day 1 | Install the Facebook & Instagram app, Maximum data sharing, catalog sync, UTMs | A test order shows as a server Purchase |
| Days 2 to 4 | Shoot 4 to 6 concepts from the angle matrix; 3:2:2 variants each | Creative tracker filled |
| Day 5 | Launch the testing campaign; add automated pause rules | Spend running, rules active |
| Days 8 to 12 | Apply the test rules; graduate winners into an Advantage+ sales campaign; add the capped customer ad set | 3 campaigns live |
| Week 3 on | Run the operating loop; new concepts every week | Weekly report on delivered CPA and MER |
Sheets to keep
- Economics: the calculator's inputs and outputs, per product.
- Creative tracker: concept, angle, format, hook, launch date, status.
- Test log: ad set, spend, purchases, CPA, decision, lesson.
- Daily dashboard: spend, orders, confirmed, delivered, cost per delivered order, MER.
Paid helpers if you outgrow Ads Manager: attribution tools (Triple Whale, Northbeam), rule engines (Madgicx, Revealbot), AI buyers with approval gates (Superscale, Agency AI from $59 a month). Prices beyond Agency AI were not checked.
Sources
- Meta: about the learning phase
- Shopify: Facebook data sharing levels
- TikTok on Shopify
- TikTok: Product GMV Max
- TikTok: advertising a Shopify store
- Confect: Meta Andromeda 2026
- AdsUploader: Andromeda, entity IDs vs volume
- AdScale: creatives per ad set
- Advantage+ shopping is now Advantage+ sales
- Advantage+ sales campaigns guide
- AdManage: Advantage+ 2026
- 1ClickReport: Advantage+ settings
- Code3: learning phase and budget steps
- RocketShip HQ: the 50-conversion rule is dated
- FirstPier: creative testing methods
- Podcast: the 3:2:2 method
- EasySell: COD in Egypt
- Whapi: COD e-commerce in MENA
- Tassyir: how COD works
- Chambers: Egypt VAT on digital ads
- Superscale: AI media buying tools
- Agency AI on the Shopify App Store
- Triple Whale: media buyer ebook