Ad platforms find your next customer by learning from your current ones. Every click, visit, add to cart, phone call, store visit, and purchase is a lesson. Most of it never arrives. For brands selling across more than one domain, even more of it goes missing.
What is multi-domain conversion tracking?
Multi-domain conversion tracking means following a shopper across two or more domains owned by the same brand, such as brand.ae and brand.sa. It also means delivering their conversions to ad platforms with the original click and customer identifiers still attached. Browsers store cookies and consent separately for each domain, so by default every new domain treats the same shopper as a stranger. A complete setup ends with one purchase per order, matched to a real person and credited to the ad that started the journey.
How hard this is depends on how your markets are structured. Subfolders like brand.com/ar and brand.com/en share one domain, so they share cookies and consent. Subdomains like sa.brand.com can read cookies set on the parent domain. Country domains and separate stores per market share nothing. Those are the setups where conversions quietly disappear.
Why a second domain breaks the signal ad platforms learn from
Each platform places its identifiers on the domain where the shopper lands. Meta sets _fbp and _fbc, TikTok stores ttclid and _ttp, Google relies on gclid, and Snap uses ScCid. When a shopper clicks a Snap ad, lands on brand.com, and completes checkout on brand.sa, the purchase fires on a domain where none of those identifiers exist. Meta and TikTok then receive a purchase they can't connect to a click. Google may not count the conversion at all.
Consent adds a second break. A shopper who accepted the banner on brand.com sees it again on brand.sa. Until they accept a second time, tags run in denied mode. Events in that window either go missing or arrive with fewer identifiers.
None of this shows up as an error. Your reports just get quieter than reality, and each platform's model learns from fewer, weaker examples of who actually buys.
Configured multi-domain tracking is not the same as complete conversion data
Most multi-domain guides stop at configuration. They cover turning on the Google linker, appending click IDs to cross-domain links, and giving each domain a first-party hostname on one server container. All of that is necessary. It tells you data can flow, but it doesn't tell you what landed. Three gaps regularly survive a correct setup.
Duplicate purchases across domains
A checkout hosted on a shared domain can fire a purchase alongside the storefront. A shopper can also reload the confirmation page. Either way, one order reaches the platform twice. Meta and TikTok can only deduplicate when every domain generates the same event_id from the order ID in the same way. If one market builds it differently, duplicates pass straight through. For more on how platforms handle this, see how event deduplication works across browser and server events.
Purchases that arrive without the click
Link decoration only works on links. Geo redirects, language pop-ups, and age gates often run in JavaScript, so they skip the decoration step, and the click ID is dropped before the shopper ever sees the second domain. The purchase arrives with an email and phone number but no click. The platform may still match the person but can't credit the ad, so the purchase does less to teach the model which ads work. Match rate on that domain also falls. If you see that, start with what to check when your Meta match rate drops.
Events that mean different things on each domain
Suppose brand.ae sends purchase value in AED including VAT, while brand.sa sends SAR excluding it. Or one storefront uses product IDs that don't match the catalog connected to Meta or Snap. The event names look identical, but the data behind them doesn't. Catalog ads lose the link between purchase and product, and value-based bidding learns from numbers that can't be compared.
Configuration is an input. Signal completeness is the outcome, and the outcome has to be measured.
Why GCC brands run into this more than most
Country domains are common in the GCC. Many brands run .ae and .sa storefronts next to a .com. Shoppers trust local domains, and legal entity, payment providers, and VAT often differ by market (5% in the UAE, 15% in Saudi Arabia). Arabic and English versions sometimes live on separate subdomains or even separate stores.
Shoppers also move between those domains more often than brands expect. A Saudi shopper who taps a TikTok ad might land on the .com and be geo-redirected to the .sa. That redirect is exactly the kind of JavaScript hop that drops the click ID.
Then there's the purchase itself. Cash on delivery is still common, and many orders are confirmed by phone or WhatsApp. The final purchase therefore happens in a backend system, not in the browser on either domain. It has to be sent server-side with the original click and customer identifiers attached, or it reaches Google, TikTok, Snap, and Meta as an orphan. Our guide to server-side GTM considerations for GCC e-commerce brands covers the infrastructure side of this.
A multi-market GCC store needs multi-domain tracking that holds up from the ad click through to the delivered order.
How to check whether your multi-domain tracking is complete
Run these checks for each domain separately. Totals across domains hide exactly the losses you're looking for.
- Orders versus reported purchases. For the same week, compare backend orders per market with purchases reported by each platform. A gap of more than a few percent on one domain points to missing or duplicate events there.
- Click ID coverage. Check what share of purchase events carries a click ID (fbc, ttclid, gclid, ScCid). Compare shoppers who bought on their landing domain with those who switched domains. A large difference means identifiers are being lost in transit.
- Match rate per domain. In Meta Events Manager, compare Event Match Quality across data sources. Do the same in TikTok Events Manager and Snap's event quality reporting.
- Duplicate orders. Check whether any order ID appears more than once in what each platform received.
- Value and currency. Compare event values against the order record for each market.
For the wider version of these checks, see how to know if your server-side tracking is actually working and what drives event quality score down. Repeat these checks after every new domain, redirect rule, or consent change, because each one can break a setup that was previously complete.
This is the gap Journify's Validate and Enrich layers are built for. Validate catches duplicate and inconsistent events before they reach a platform. Enrich adds the identifiers that let a purchase on a second domain still match a real person.
FAQ
Does server-side tracking fix multi-domain tracking on its own?
No. Server-side tracking keeps cookies first-party on each domain, but it doesn't carry a click ID from one domain to another. That still requires link decoration or server-side identity stitching.
Do subfolders need a cross-domain setup?
No. Subfolders share one domain, so they share cookies and consent. Subdomains can share cookies set on the parent domain. Country domains and separate stores are the setups that need cross-domain work.
Can consent be shared across domains?
Sometimes. The domains generally need to belong to the same company, and the banner has to name them. Confirm with your legal team, especially for separate legal entities and under Saudi Arabia's PDPL.
How do I know if my multi-domain setup is losing conversions?
Compare backend orders per market with purchases reported by each ad platform for the same period. A consistent gap on one domain is the clearest sign.
Multi-domain tracking is only finished when every order reaches every platform once, with its click and its customer attached. Measure that per domain, and fix the gaps before the platforms learn from them.
Book a call to see what your domains are actually sending.
