Attribution
Shopify attribution models explained
An attribution model decides which touchpoint gets credit for a sale. There's no model that's objectively correct — each one answers a slightly different question, and picking one is a deliberate trade-off, not a technical detail.
The same journey, five different answers
Take one customer: they see a Meta ad, ignore it, see it again a week later, click through, browse without buying, then find the store again through a Google search three days after that and purchase. Ask five different attribution models who gets credit for that sale, and you get five different answers — none of them wrong, because none of them are describing an objective fact. They're describing a rule you chose to apply.
The five models, side by side
Rewards: The first touchpoint that brought the customer into your world.
Blind spot: Ignores everything that happened after — including the touchpoint that actually closed the sale.
Rewards: The final touchpoint before purchase.
Blind spot: Rewards the closer, not the opener. A discovery ad gets zero credit if a branded search closed the sale.
Rewards: Every touchpoint in the journey, equally.
Blind spot: Treats a passing glance the same as a deliberate return visit — spreads credit without judging which touch actually mattered.
Rewards: The first and last touch most heavily, with a smaller share to the middle.
Blind spot: The 40/20/40 (or similar) split is a modelling choice, not a measured fact about your customers.
Rewards: Whatever touchpoints the underlying model finds statistically associated with conversion.
Blind spot: Needs enough order volume to be statistically meaningful, and the model is a black box you can't fully audit.
Click-through, view-through, and the window around them
Underneath whichever model you pick sits two more decisions: whether an ad view (not just a click) counts as a touchpoint, and how long after that touchpoint a purchase still counts. A 7-day click / 1-day view window will produce a noticeably different attributed-revenue number than a 28-day click / 1-day view window applied to the exact same orders — without either being more "accurate." Longer windows generally attribute more revenue to advertising; shorter windows attribute less. Neither is measuring something different about your customers — they're just drawing the line at a different point.
Why Meta and Google rarely agree
Meta and Google each apply their own attribution model, on their own window, using only the touchpoints they can see on their own platform. Meta doesn't know about the Google search that happened between the ad view and the purchase. Google doesn't know about the Instagram scroll. Each platform reports as if its own touchpoints were the whole story, because from where it's sitting, they're all it can observe. That's the root cause of most Meta-vs-Google attribution disagreements — not a bug in either platform, but a structural limit of single-platform measurement.
There's no universally "right" model to pick
The honest framing is that an attribution model is a lens, not a measurement instrument. Last-click tends to undervalue top-of-funnel discovery spend. First-click tends to overvalue it. Data-driven models are the most balanced in theory but require enough conversion volume to be statistically stable, and their internal logic isn't fully auditable. What matters more than picking the "correct" model is picking one, being consistent about which one you're looking at, and not comparing numbers produced under different models as if they were the same metric.
See your orders under a consistent model
Attribix applies one clear methodology across Meta and Google, so you're comparing campaigns on the same terms.