Skip to main content

E-commerce

E-commerce Attribution Models: Which to Trust

Last-click? Multi-touch? Data-driven? With iOS 14+ and cookie deprecation, attribution is messier than ever. Here's a framework for what to trust and what to ignore.

Share:

Attribution is the messy art of figuring out which marketing touchpoint deserves credit for a sale. With iOS 14+, cookie deprecation, and platform fragmentation, this is harder than ever in 2026.

Here's the framework for thinking about attribution.

TL;DR

  • Different attribution models give different answers. None is "right."
  • Use multiple sources and triangulate.
  • For platform optimization: trust the platform's reported number.
  • For business decisions: blend Shopify-reported, GA4, and platform data.
  • Most accuracy lost to iOS 14+: 25–40%. CAPI recovers most.

Common attribution models

Last-click

All credit to the last click before conversion.

Pros: simple, clear.

Cons: ignores upstream touches (e.g., Meta ad created awareness, Google capture got the click).

First-click

All credit to first touchpoint.

Pros: rewards top-of-funnel discovery.

Cons: ignores conversion-driving last touch.

Linear

Equal credit across all touchpoints.

Pros: fair-ish.

Cons: ignores fact that some touches matter more than others.

Time decay

More credit to recent touches.

Pros: reflects how memory works.

Cons: still arbitrary weighting.

Position-based

40% first, 40% last, 20% middle.

Pros: balances awareness and conversion.

Cons: arbitrary 40/40/20.

Data-driven

ML-driven attribution based on actual customer paths.

Pros: most accurate.

Cons: requires sufficient data; opaque.

What each platform uses

Meta (Facebook/Instagram)

  • Default: 7-day click + 1-day view.
  • Available: 1-day, 7-day, custom.
  • Default: Data-driven attribution.
  • Available: Last-click, first-click, linear, time decay, position-based.

Klaviyo

  • Last-click email/SMS attribution.
  • 5-day click, 1-day open windows (default).

Shopify

  • Source of truth for actual orders.
  • Reports referrer/UTM source per order.

GA4

  • Default: Data-driven.
  • Available: Last-click, first-click.

The reality

Each source attributes different conversions to itself. If you sum up Meta + Google + Klaviyo + organic claims, you'll often see 130–150% of your actual revenue. That's because they all claim partial credit.

Don't add them up. Use Shopify as source of truth for actual revenue.

How to think about attribution

Different decisions need different data.

Decision: optimize Meta campaigns

Use Meta's reported numbers. Even if attribution is partial, the algorithm optimizes against it.

Decision: optimize Google campaigns

Use Google's reported numbers, with same logic.

Decision: budget allocation across channels

Use Shopify's UTM data + GA4 multi-touch.

Decision: assess true ROI

Compare ad spend to total revenue (Shopify), accounting for organic and direct.

Triangulation framework

Track 4 numbers:

1. Shopify revenue (source of truth).

2. Meta-attributed revenue (per Pixel/CAPI).

3. Google-attributed revenue (per Ads conversion).

4. Klaviyo-attributed revenue (email/SMS).

For each marketing source, calculate:

  • Platform-reported.
  • UTM-based (orders with that UTM).

Common pattern: platform reports 30–40% MORE than UTM data. That's view-through attribution.

For decisions: trust UTM data. For optimization: let platforms use their preferred attribution.

iOS 14+ impact

Apple Tracking Transparency reduced cross-app attribution. Result:

  • 30–50% of iOS Pixel events are lost without CAPI.
  • Some attribution shifts to "view-through."
  • Lookalikes degrade (less seed data).

CAPI recovers most of this. Without CAPI: you're flying blind on iOS.

Third-party cookies are being phased out. Affects:

  • Display Network retargeting (limited).
  • Cross-site attribution.
  • Some custom audience tools.

First-party data (your own customer list, server-side tracking) is the future. Build:

  • Email + SMS opt-ins.
  • Customer accounts on your site.
  • Server-side conversion tracking (CAPI for Meta, gtag/GTM for Google).

Multi-touch attribution tools

For accounts spending $30K+/month:

Triple Whale

Pulls Meta + Google + Klaviyo + Shopify into one dashboard with multi-touch attribution.

Northbeam

Similar; more analytical.

Wicked Reports

Older but still capable.

Cost: $200–$1,000/month.

For smaller accounts: free Looker Studio dashboards work.

Post-purchase survey

The single best zero-tech attribution method:

On the order confirmation page, ask:

"How did you hear about us?"

☐ Facebook/Instagram

☐ Google search

☐ Friend/family

☐ Other

Compare to your platform-reported attribution. Often surfaces:

  • Channels under-credited (word of mouth).
  • Channels over-credited (view-through).

Use this data to calibrate your trust in each platform.

Common attribution mistakes

1. Trusting one source

Believing Meta's reported ROAS while Shopify shows half. Triangulate.

2. Adding up channel-claimed revenue

You'll over-count. Sum will exceed Shopify's actual.

3. Ignoring view-through

Meta especially relies on view-through. Some legitimate, some inflated.

4. Not running CAPI

Losing 25–40% of Pixel attribution unnecessarily.

5. Switching attribution models in dashboards constantly

Each switch resets your reference. Pick a model per decision type and stick.

What I do

For my client accounts:

  • Daily checks: platform-reported numbers (Meta, Google, Klaviyo).
  • Weekly review: Shopify + GA4 + platform-reported, look for divergence.
  • Monthly strategic: post-purchase survey + UTM attribution + LTV per channel.

This combo gives optimization data + business reality.

Want help with attribution setup?

Attribution clarity is competitive advantage. My Shopify Expert service includes attribution setup and reporting. Or learn the system in the free Shopify videos on YouTube.

Related reading:

Back to all articles

Apply it to your account

Apply this to your own account

Share your website, goal and current setup. I will tell you whether it is a fit, then we agree on the scope before work starts.