Cross-Channel Attribution Explained: What Actually Works

Your Meta dashboard says it drove 100 conversions. Google Ads claims 80. Your Shopify store shows 90 total orders.

100 + 80 = 180. But you only had 90 orders.

Welcome to the attribution problem. Every platform claims credit, and the math never adds up. This guide explains why—and what you can actually do about it.

Attribution answers the question: "Which marketing touchpoint caused this sale?"

The problem: A customer might:

• See a Facebook ad on Monday
• Google your brand on Wednesday
• Click a retargeting ad on Thursday
• Open an email on Friday
• Type your URL directly and buy on Saturday

Which channel "caused" the sale? All of them claim credit.

• Each platform only sees its own touchpoints
• Each platform attributes based on its own rules
• No platform has the complete customer journey
• Result: Overcounting everywhere

Ad platforms are incentivized to show you high ROAS so you keep spending:

• Meta: "We touched this customer, we get credit"
• Google: "We touched this customer, we get credit"
• Both: "Trust our numbers, keep spending"

With different windows, platforms count the same sale multiple times.

Meta's "1-day view" is especially controversial:

• Customer scrolls past your ad (counts as "view")
• Customer buys within 24 hours through any channel
• Meta claims credit for the sale
• This massively inflates Meta's reported conversions

100% credit to the last touchpoint before purchase.

• Pros: Simple, clear
• Cons: Ignores awareness/discovery channels

100% credit to the first touchpoint.

• Pros: Values discovery
• Cons: Ignores conversion channels

Equal credit to all touchpoints.

• Pros: Acknowledges full journey
• Cons: Treats all touches equally (probably wrong)

More credit to touchpoints closer to conversion.

• Pros: Balances discovery and conversion
• Cons: Arbitrary decay rate

Algorithmic model based on your data.

• Pros: Theoretically optimal
• Cons: Black box, requires volume, still overcounts

Bottom line: No model is "correct." Each is a different lens on the same incomplete data.

iOS14's App Tracking Transparency made attribution even harder:

• ~80% of iOS users opted out of tracking
• Meta lost visibility into post-click behavior
• Conversion data is now modeled (estimated), not measured
• Attribution windows were shortened

• Platform-reported conversions are now estimates
• Accuracy varies wildly by account and vertical
• Some advertisers see 30-50% underreporting
• Others see overcounting (modeled conversions too aggressive)

Trust in platform numbers has declined significantly since 2021.

Instead of per-channel ROAS, look at total revenue ÷ total ad spend.

If MER stays healthy as you scale, your marketing mix is working—regardless of what each platform claims.

Turn off a channel for a period, measure the impact:

• Pause Meta ads for 2 weeks
• If revenue drops 30%, Meta was driving ~30% incrementally
• If revenue barely moves, Meta was overclaiming

Requires patience and willingness to sacrifice short-term.

Run ads in some regions, not others. Compare sales lift.

"How did you hear about us?" isn't perfect but provides directional signal.

If Meta says ROAS dropped 20% week-over-week, the direction is probably right even if the absolute number is wrong.

• Use blended ROAS/MER as primary metric
• Don't over-optimize based on platform numbers
• Run occasional incrementality tests (pause for 1-2 weeks)

• Implement post-purchase surveys
• Track MER and per-channel ROAS (directional)
• Use a unified analytics tool to see total picture
• Run quarterly incrementality tests

• Invest in MMM (marketing mix modeling)
• Run ongoing geo/holdout tests
• Build internal attribution model
• Treat platform numbers as one input, not truth

The key insight: Attribution will never be perfect. Make peace with directional accuracy and focus on overall business health.