Facebook says it drove the sale. Google says it did. The truth? Both helped. Attribution models try to solve this, but they're all flawed. Here's how to navigate it.

The Attribution Problem: Who Gets Credit?

Customer journeys are complex:

Example Journey

Sees Facebook ad → Clicks
Visits website → Doesn't convert
Goes to Google → Searches your brand
Clicks Google ad → Visits again
Gets email → Opens and clicks
Returns via direct → Converts

Question: Who gets credit? Facebook? Google? Email? Direct?

Linear, Time-Decay, and U-Shaped Models

Different models credit differently:

First-Click Attribution

100% credit to first touchpoint
Good for: Understanding what creates awareness
Problem: Ignores all other touchpoints

Last-Click Attribution

100% credit to last touchpoint
Good for: Understanding what closes deals
Problem: Ignores awareness and consideration

Linear Attribution

Equal credit to all touchpoints
Good for: Fair distribution
Problem: Doesn't weight importance

Time-Decay Attribution

More credit to recent touchpoints
Good for: Valuing recency
Problem: Still arbitrary

U-Shaped Attribution

40% to first touch, 40% to last touch, 20% to middle
Good for: Valuing awareness and conversion
Problem: Still arbitrary weighting

The Bias of Platform Data (Why FB Claims 100% Credit)

Platforms are biased:

Facebook's View

Sees: User clicked ad → Later converted
Claims: "We drove the conversion"
Reality: They helped, but weren't the only factor

Google's View

Sees: User searched brand → Converted
Claims: "We drove the conversion"
Reality: They helped, but user was already aware

The Truth

Both helped. Attribution is always incomplete. Use multiple models and make decisions with imperfect data.

Tools for Unbiased Attribution

Use tools that see across platforms:

Attribution Tools

Google Analytics 4: Cross-channel attribution
HubSpot: Marketing attribution reporting
Ruler Analytics: Call and form attribution
Triple Whale: eCommerce attribution
Custom models: Build your own in BigQuery

Making Budget Decisions with Imperfect Data

You'll never have perfect attribution. Here's how to decide:

1. Use Multiple Models

Look at first-click (awareness)
Look at last-click (conversion)
Look at linear (fair distribution)
See patterns across models

2. Test Incrementally

Pause a channel for 2 weeks
See if conversions drop
If yes, that channel matters
If no, maybe it doesn't

3. Focus on Business Outcomes

Don't optimize for attribution
Optimize for revenue
If revenue increases, keep doing it
Attribution is a guide, not the goal

Conclusion

Attribution is always imperfect. Every model has bias. Use multiple models, test incrementally, and focus on business outcomes, not perfect attribution. The goal isn't perfect data—it's better decisions.