You check AppLovin's dashboard and it shows a 4.2x ROAS. Solid. Then you check Shopify and the order count for that same period doesn't come close to matching what AppLovin says it drove.
Founders see this every week and assume something's broken. It isn't.
This is a structural difference in how AppLovin's attribution model counts a conversion versus how Shopify records an order. AppLovin is modeling credit. Shopify is logging transactions. Those are two different jobs, and they were never going to produce identical numbers.
This article breaks down how AppLovin attribution works for Shopify brands, mechanically, so you can judge performance with the right context instead of panicking every time the numbers diverge, or worse, trusting AppLovin's dashboard as gospel because it's the one showing the bigger win.
AppLovin's Attribution Model, Explained
AppLovin uses last-click (sometimes called last-touch) attribution inside its own network, AXON. If a user interacts with an AppLovin ad and later converts, AppLovin's system gives full credit to itself, regardless of what other channels that user touched along the way. No splitting credit with Meta, Google, or organic. Whoever AppLovin thinks was last wins the whole conversion.
That's different from a multi-touch model, which spreads credit across every touchpoint in a user's path.
Attribution windows matter here too. AppLovin typically applies something like a 1-day view-through window and a 7-day click-through window, though these are configurable per campaign and defaults can shift [VERIFY exact current defaults in-platform]. A longer window means more conversions get pulled into AppLovin's bucket, even ones that happened days after the ad interaction and might have happened anyway.
Then there's the matching problem. A lot of AppLovin's traffic is mobile and app-driven, landing on a Shopify storefront from an in-app environment. Device-level and probabilistic matching fill the gaps where a clean, deterministic click-to-purchase link doesn't exist. So AppLovin's reported conversions are a mix of two things: conversions it directly and verifiably drove, and conversions it's modeled as probably related. The dashboard doesn't separate the two. It just shows you one number, and that number tends to run high.
Why Shopify Data and AppLovin Data Rarely Match 1:1
Shopify has no attribution logic at all. It counts an order the second checkout completes. That's it. No modeling, no windows, no probabilistic matching, just a raw transaction log.
That gap alone explains most of the mismatch you'll see between AppLovin and Shopify.
Cross-device behavior makes it worse. Someone scrolls past an AppLovin ad on their phone, doesn't click, then opens a laptop later and buys directly. AppLovin might still credit itself through view-through modeling. Shopify just sees a completed order with no clean attribution thread back to that ad at all.
Then add iOS 14.5 and App Tracking Transparency into the mix. Since Apple locked down device-level tracking, AppLovin (like every ad network) lost a chunk of its deterministic signal. The response across the industry was to lean harder into modeled, probabilistic attribution to fill the hole. That's not a criticism unique to AppLovin, every major ad platform did the same thing, but it does mean the "conversions" you see today carry more estimation than they did in 2020.
Here's the pattern that shows up constantly: AppLovin's dashboard claims 50 conversions for a campaign. You pull Shopify orders for the matching UTM tag over the same window, and you find 30. Not because AppLovin is lying. Because 20 of those 50 were modeled assumptions that never had a matching Shopify transaction to begin with.
Where GA4 Adds a Third, Different Number
Now bring GA4 into the room, and you've got a third opinion.
GA4 runs data-driven attribution by default, splitting conversion credit across multiple touchpoints in a user's path instead of handing 100% to whichever channel touched them last. So if a user saw an AppLovin ad, then later clicked a Google search ad, then converted, GA4 might give partial credit to both. AppLovin, using last-click logic on its own network, would claim the whole thing.
GA4 also depends on its own tracking layer, UTMs and GA4 tags, which doesn't always line up cleanly with AppLovin's SDK-based reporting. Different data collection methods, different identity resolution, different default logic. If you're comparing GA4's funnel data against AppLovin's dashboard expecting them to agree, you're going to be disappointed every time.
None of these three tools are wrong. AppLovin is correct by its own last-click logic. Shopify is correct because it's a ledger. GA4 is correct by its own multi-touch model. Three legitimate systems, three different numbers, same 7-day window. That's just how attribution works when each platform builds the rules to make itself look good, intentionally or not.
How to Sanity-Check AppLovin's Numbers
Start by comparing AppLovin's reported revenue against Shopify's actual revenue for the same campaign or UTM tag, over a rolling 7-day window. Not real-time. Real-time attribution data is the least reliable version of itself, it firms up over several days as delayed conversions and modeled matches settle.
If you want the real answer, run a holdout test. Pause AppLovin spend for a short stretch and watch what happens to total Shopify orders. If revenue barely moves, a chunk of what AppLovin was "driving" was going to happen anyway. This is the closest thing to ground truth you'll get, and it's more honest than any dashboard metric.
Check your attribution window settings too. If AppLovin's default window is 7 days click-through but your average customer takes 12 days to decide, you're either missing real conversions or, more likely, catching conversions that had nothing to do with the ad. Match the window to your actual consideration cycle, not whatever AppLovin ships with by default.
And stop leaning on platform-reported ROAS as your main scoreboard. Blended CAC, meaning total ad spend divided by total Shopify orders across all channels, is a far more honest number. It doesn't care which platform wants credit. It just tells you what it actually cost to get a customer.
Getting a Single Source of Truth Across AppLovin, Shopify, and GA4
The real fix isn't picking which dashboard to trust. It's stopping the practice of trusting any single dashboard in isolation.
Pull AppLovin's spend and conversion data, Shopify's order data, and GA4's funnel data into one warehouse-backed view. Once everything sits on the same timestamps with consistent matching logic, the "which number is right" argument mostly disappears, because you're looking at one reconciled dataset instead of three competing exports.
That's what Trivas builds on Amazon Redshift: Shopify order data and ad platform data sitting side by side, matched consistently instead of eyeballed across browser tabs. On top of that, the Wingman AI layer flags attribution discrepancies automatically, so instead of a founder manually cross-referencing AppLovin, Shopify, and GA4 every Monday morning, the system surfaces the gap for you.
AppLovin's number is a modeled estimate, not a ledger. Read it as directional, not absolute, and you won't get blindsided when Shopify's numbers come in lower.
Track blended CAC and actual Shopify revenue as your anchor metrics. Let AppLovin's dashboard be a secondary signal, useful for spotting trends, not the number you build your budget decisions around.
Revenue growth leader and co-founder driving Trivas's commercial strategy. Om has led the product vision and execution from scratch. With a strong background in SaaS sales and GTM strategy, Om bridges product innovation with real-world customer needs.
Continue Reading
explore more insights
Ecommerce Analytics for Barcelona Brands: A Buyer's Guide Before Shoptalk Europe
3 min read
Attribution Modeling and Customer Journey Analysis
3 min read
Why Brands Switch from Triple Whale: 8 Honest Reasons
.d53b12e5.png&w=3840&q=75 "Trivas.ai")
