by Om RathodThird-party cookies have been dying a slow death for years, and 2024-2025 finished the job for most Shopify brands. If you're still trying to measure ROAS without cookies on Shopify using the same pixel setup you had in 2019, your numbers are probably wrong. Not slightly off. Wrong enough to misallocate real ad budget.
This guide walks through what's actually broken, what replaces cookie-based tracking, and how to calculate ROAS on Shopify using data you actually own.
Three things happened in quick succession. Apple's iOS 14.5+ App Tracking Transparency requirement made users opt out of cross-app tracking by default. Safari's Intelligent Tracking Prevention (ITP) has been quietly killing third-party cookies in that browser since 2017. And Chrome, which still commands the largest browser market share, has been phasing out third-party cookies too.
The result: pixels can't follow a shopper from an Instagram ad to your Shopify checkout the way they used to.
So platforms guessed. Meta and Google both moved to modeled conversions, filling gaps with statistical estimates instead of observed data. That's why platform-reported ROAS numbers have gotten inflated and inconsistent, and why two ad platforms will each claim credit for the same sale.
For Shopify merchants, the practical impact is simple and expensive. Every ad platform overcredits itself. Meta says a campaign drove the sale. Google says its Performance Max campaign drove the same sale. Add both up and you'll think you're profitable on channels that are actually break-even.
That's the real question this page answers: if cookies aren't doing the attribution math anymore, what is?
Cookieless ROAS measurement means calculating ad efficiency using data sources that don't depend on third-party pixels or cross-site cookies to track a user's path.
Three data types replace what cookies used to do:
First-party order data
Server-side conversion APIs
Platform-level aggregated reporting
The formula hasn't changed. ROAS is still revenue attributed to ads divided by ad spend. What's changed is that the "attributed to ads" part got a lot harder to prove. The math was never the problem. Knowing which dollar of revenue to credit to which dollar of spend is.
Start with the numerator: actual order revenue from Shopify, not the conversion count a platform reports. Pull it straight from your Shopify orders, not from a pixel's version of events.
For the denominator, pull ad spend directly from the Meta, Google, and TikTok ad account APIs. Don't eyeball it from a dashboard. API-pulled spend numbers match what you're actually billed, which sounds obvious until you've reconciled a spreadsheet that didn't.
UTM parameters are the connective tissue holding this together. If your campaign naming is inconsistent, "revenue by source" turns into a mess of near-duplicate labels that nobody can total up cleanly. Set a naming convention once (source, medium, campaign, and stick to it) and every future report gets easier.
Once you've got clean revenue and clean spend, calculate a blended ROAS across all channels combined. This is your sanity check. If Meta says 4x, Google says 3.5x, and your blended number across Shopify orders and total spend is 2.1x, you know overlap is happening somewhere. Run your current numbers through a ROAS calculator before you trust any single platform's self-reported figure.
Shopify itself is your best source of ground truth. Order value, discount code usage, landing page, and referrer at time of purchase all live in your store's backend, no cookie required.
Server-side conversion APIs are the second piece. Meta's Conversions API and Google's Enhanced Conversions send hashed customer data (email, phone, order value) directly from Shopify's backend to the ad platform. Browser-level blocking never touches it because the data never routes through the browser in the first place.
GA4 adds a cross-channel view on top of that, especially when paired with consent mode and server-side tagging. It's not perfect, GA4's event model has its own quirks, but it fills gaps that a single ad platform's dashboard won't show you. If you haven't set this up yet, it's worth doing properly rather than leaving GA4 on its default client-side config: check GA4 integration for the setup.
Then there's email and SMS. Klaviyo and similar platforms attribute revenue from owned channels using their own send and click data, which never depended on third-party cookies to begin with. If a meaningful chunk of your revenue comes from flows and campaigns, that's attribution you can trust without any of the cookie-deprecation headaches.
Marketing mix modeling (MMM) is the most cookie-independent option available. It doesn't try to track a person, it looks at aggregate spend and revenue over time and works out statistical relationships between them. No pixel, no user-level match required.
First-party last-touch attribution, using Shopify's own referrer and UTM data, is the other end of the spectrum. It's less precise, since it only captures the last thing before a purchase, but it's still directionally useful and a lot cheaper to set up than a full MMM.
Most Shopify brands don't need to pick one. Use first-party last-touch for day-to-day tactical calls (which ad creative is pulling its weight this week), and lean on MMM or incrementality testing for the bigger budget-level decisions (should we shift 20% of spend from Meta to TikTok).
Honestly, chasing a single "correct" attribution model is the wrong goal now. Nothing gives you certainty anymore. The goal is triangulating a few imperfect signals instead of trusting one dashboard blindly.
SGE and AI Overviews blend organic and paid search results into one compressed experience. That makes click-based attribution even noisier than it already was, because the click itself is a less reliable signal of intent than it used to be.
The shift underway is toward measuring outcomes at the account or campaign level instead of trying to reconstruct an individual's click path. Google's own reporting has been nudging in this direction for a while, and it's only going to accelerate as SGE rolls out further.
This is exactly why first-party revenue data from Shopify becomes the more stable ground truth. Search UX will keep changing. Your order data won't. Anchoring ROAS calculations to what actually got sold, rather than to a click trail that search engines keep rearranging, is the more durable approach.
Yes, and doing it manually is genuinely painful past a certain order volume.
Trivas pulls Shopify order data and ad platform spend data into one warehouse, built on Amazon Redshift, and calculates blended ROAS without depending on browser-level tracking at all. It's the same underlying data described above (first-party orders, API-pulled spend), just automated instead of hand-assembled.
That replaces the manual reconciliation most teams are stuck doing between Shopify exports and ad account reports, which for a lot of brands eats a few hours a week that could go somewhere more useful. Once it's connected, you're looking at a live dashboard instead of a spreadsheet that's already stale by the time you finish it. That live view is part of what BI reporting is built around.
One thing worth pointing out: this setup doesn't force you to pick between MMM, MTA, or last-click. The underlying data connections stay the same no matter which attribution model you layer on top, so you're not locked into one approach.
The fastest way to start is installing Trivas AI on the Shopify App Store, which connects your order data cleanly without the manual export step most teams are still doing by hand.
Three things to do this week, in order. Connect your Shopify order data so revenue numbers are coming from actual orders, not platform-reported conversions. Connect ad account spend via API for Meta, Google, and whatever else you're running. Standardize your UTM tagging so the two data sets actually line up.
Before you automate anything, run your current blended numbers through a free ROAS calculator. It's a quick way to see how far off your platform-reported ROAS is from reality, and it'll tell you how urgent this project actually is.
If you're just starting to explore what a Shopify-specific setup looks like, it's worth reading through before committing to anything. And if you want more of this kind of breakdown as tracking keeps shifting, our newsletter covers the changes as they happen rather than after the fact.
explore more insights
3 min read
3 min read
3 min read
.d53b12e5.png&w=3840&q=75 "Trivas.ai")