Every Shopify merchant running paid traffic eventually hits the same wall: Meta, Google, GA4, and Shopify all report a different number of "sales" for the same week. A big part of that gap is the attribution model doing the counting. This guide walks through last-click vs data-driven attribution on Shopify with real arithmetic, so you can decide which lens to trust before you cut a campaign.
What each model actually does
Last-click attribution gives 100% of an order's credit to the last channel the shopper clicked before buying. If someone discovered you on Instagram, came back through a Google search, then finally clicked an email and bought, last-click hands the whole sale to email. Everything upstream gets zero. This is Shopify's native default — last non-direct click — and it is what your Shopify Analytics "Sessions by referrer" and order reports use.
Data-driven attribution (DDA) splits that same order into fractions. Using machine learning, it might credit 0.35 of the conversion to Google Ads, 0.25 to organic search, and 0.40 to email, per an example from WeltPixel. GA4 made DDA its default model in November 2023, replacing last-click as the out-of-the-box setting. So the moment you look at GA4 channel reports, you are usually looking at fractional credit — while Shopify beside it is still awarding whole conversions.
That single design difference is why channel rows never line up between the two tools. It is the same reason your platform ROAS overstates real profit.
Why the two models disagree on your store
Think of the disagreement in two buckets.
The first is credit distribution, covered above: whole credit to one channel versus fractions spread across many. The second is what each system can even see. Shopify records every completed checkout server-side, so it owns 100% of your orders. GA4's data-driven model runs on client-side JavaScript that ad blockers, cookie-consent declines, and closed tabs quietly break — field estimates put ad-blocker and consent-affected traffic at roughly 10–25% of users, according to Audiense/Elevar. GA4 typically reports fewer purchases than Shopify as a result; a 15–30% gap is considered normal per BlueFrog.
Meta adds a third wrinkle. On its default 7-day-click / 1-day-view window (Jon Loomer), Meta claims sales a shopper only saw an ad for, not just clicked. A 20–35% gap between Meta-reported purchases and Shopify orders is normal on that window, according to Vaizle. So you are never comparing two numbers — you are comparing four different attribution philosophies at once, which is the whole subject of reconciling your ecommerce data.
A worked example: one week, four numbers
Say you sell enamel mugs. In one week you get 100 real orders, and each order is $40 in product plus $5 shipping plus $4 tax, so $49 total per order. Ground truth: 55 buyers clicked a Meta ad within the last week, 15 only saw one, 10 clicked a Google ad last, and 20 arrived via organic search or typed you in directly.
Here is how last-click sorts those 100 orders in Shopify:
- Facebook: ~55 orders (the shoppers whose last click was a Meta ad)
- Google: ~10 orders
- Search / Direct / Other: ~35 orders
Notice the 15 view-through buyers are not credited to Facebook here. They clicked nothing, so last-click files them under whatever they actually clicked last. Clean, simple, and it always sums to your real 100 orders.
Now run the same week through data-driven attribution in GA4. Suppose GA4 loses about 20 buyers to blockers and consent declines, then models some back, landing near 72 recorded purchases. Under DDA it might report "Paid Social" at ~48 conversions, "Paid Search" at ~10, and "Organic" at ~14 — fractions that reflect assisting touches last-click threw away. No single row matches Shopify's 55, and the total does not match either.
Which number is right? For how many sales happened, only Shopify's 100 is right. For how much each channel helped along the way, DDA's spread is arguably closer to reality than last-click's winner-take-all. They answer different questions.
The number both models ignore: profit
Here is what neither last-click nor data-driven attribution tells you — whether that order made money. Attribution divides credit for revenue; it never touches cost.
Keep the mug example. Say your landed product cost from Printify is $14, and Shopify Payments takes roughly 2.9% + 30¢ per transaction on the Basic plan, per Webgility. On a $49 order that fee is about $1.72. If your Meta ad spend works out to $12 per order, the math is:
$49 revenue − $14 product − $1.72 fee − $5 shipping label − $12 ad spend = $16.28 profit
Now the attribution question changes shape. Last-click told you Meta drove 55 orders; data-driven told you paid social assisted ~48. But your decision is not "how many did Meta earn credit for" — it is "did the orders Meta influenced clear $16.28 after everything, or did the ones needing three touches to close cost more in ad spend than they returned?" Attribution models cannot answer that, because they never see product cost, delayed conversions, refunds, or the ~$15 chargeback fee per dispute that Webgility documents.
Which model should you use on Shopify?
Match the model to your volume and your question.
Use last-click as your default when order volume is modest or you want a stable, auditable source for "which channel got the final click." Google recommends at least 400 conversions per month for reliable data-driven model training, per WeltPixel; below that floor, DDA's machine learning has too little signal and its output gets noisy. Last-click never has that problem — it is a fixed rule.
Layer data-driven attribution on top once you clear that conversion threshold and you are actively running multi-channel campaigns where upper-funnel ads seed sales that close on branded search or email. DDA will surface the assisting channels last-click zeroes out, which can change how you value prospecting versus retargeting.
Trust Shopify's order count for revenue, always. Whatever GA4's data-driven view says about channels, your Shopify total sales — after refunds — is the source of truth for how much money came in. If you want to go deeper on model selection, compare the tradeoffs of dedicated attribution modeling tools. And if your channel reports look empty or misattributed, the culprit is often Shopify UTM parameters not showing up, not the model itself.
Where PodVector fits
PodVector connects your Shopify, Meta Ads, Google Ads, Printify, and Printful accounts and computes true per-order profit — the $16.28 line, not the attributed-revenue line. It is not a dashboard and not another attribution model; it reconciles the data you already have into what each order actually earned after product, fees, shipping, and ad spend.
Victor, PodVector's AI employee, reads that reconciled picture and proposes moves you approve, taking action on the Shopify side. Victor does not touch your ad account — he reads your ad data and hands you the profit-aware next step. Connect your store and see per-order profit.
FAQs
Is Shopify last-click or data-driven by default?
Shopify's native attribution is last non-direct click — 100% of an order's credit goes to the last channel the shopper clicked. This is independent of GA4. Even if your GA4 property uses data-driven attribution (its default since November 2023, per WeltPixel), Shopify's own reports keep using last-click, which is why the two tools disagree.
Why does data-driven attribution show fractional conversions?
Because it splits one sale across every touchpoint that contributed. Instead of awarding a whole conversion to the final click, DDA might assign 0.35 to Google Ads, 0.25 to organic, and 0.40 to email for a single order, as WeltPixel illustrates. The fractions sum to one real conversion; they just distribute the credit.
Will switching attribution models change how much money I actually made?
No. Attribution only reassigns credit for revenue that already happened. Your Shopify total sales and your bank payout are unchanged whether you use last-click or data-driven. Switching models can change which channel looks responsible — and therefore where you spend next — but not the dollars in the register.
How many conversions do I need for data-driven attribution to be reliable?
Google recommends at least 400 conversions per month for reliable model training, according to WeltPixel. Below that, the data-driven model has too little signal and its output can be noisier than a simple rule like last-click. Higher-volume stores get the most benefit from DDA.
Which model tells me if a campaign is profitable?
Neither. Both last-click and data-driven attribution only divide revenue credit; neither subtracts product cost, transaction fees, shipping, refunds, or ad spend. To know whether a campaign made money, you need per-order profit reconciled across your store and ad platforms — which is exactly what PodVector computes.