What Is Attribution?
Marketing attribution is the process of assigning credit for a conversion to the marketing touchpoints that led to it. In practice, every ad platform, analytics tool, and CRM uses a different attribution model and window — which is why the same purchase gets counted three times across dashboards and why 'ROAS' means five different things.
How the models differ
A customer sees a Meta ad on Monday, clicks a Google Search ad on Wednesday, gets a retargeting email on Friday, then buys on Saturday. Last-click gives 100% credit to Email. Last-non-direct-click gives 100% to Google. Linear splits credit evenly across all three. Data-driven (GA4 default) uses machine learning to weight each touch. Meta's own attribution gives full credit to itself. None are 'right' — they measure different things.
Benchmarks
- Common models: last-click, first-click, linear, time-decay, position-based (U-shaped), data-driven.
- Common windows: 1-day-click, 7-day-click, 28-day-click, 1-day-view, 7-day-click + 1-day-view.
- Meta default (2026): 7-day-click + 1-day-view.
- Google Ads default: data-driven with 30-day-click.
Why it matters
Attribution decides how you allocate budget. Under last-click, retargeting looks like a hero and prospecting looks broken. Under data-driven, prospecting gets more credit. Switch models and the same account looks like a different business. That's why serious teams supplement platform attribution with MER, geo-lift tests, and incrementality studies — they're the truth checks.
Common mistakes
- 1.Summing Meta ROAS + Google ROAS + TikTok ROAS. They all overclaim; the sum is fantasy.
- 2.Assuming last-click reflects buyer behaviour. It reflects the last cookie the browser allowed.
- 3.Ignoring view-through conversions. Sometimes they're real lift; sometimes they're the platform padding numbers.
- 4.Changing attribution windows mid-quarter without recalibrating targets. Every dashboard breaks.
Put Attribution to work
Free calculators
Related services
FAQs about Attribution
Which attribution model is best?
None in isolation. Use data-driven or position-based as the daily working model, then reconcile monthly with MER and periodic incrementality tests. Single-touch models (last-click, first-click) are diagnostic, not decisional.
Why does Meta report more sales than my Shopify?
Meta uses 7-day-click + 1-day-view and cross-device matching. Shopify uses last-click and only sees users who converted on-site in one session. A 20–60% gap is normal; > 100% suggests broken pixel or view-through overweight.
What's the difference between attribution and incrementality?
Attribution divides credit for conversions that happened. Incrementality asks 'would those conversions have happened anyway if we hadn't run ads?' The answer is usually 'some yes, some no' — and only incrementality tests can tell you which.
Is data-driven attribution reliable?
It's the best model most teams can implement, but it's a black box and requires enough conversion volume to train (Google's floor: 300 conversions in 30 days per campaign). Below that, it silently reverts toward last-click.
Related terms
Measures the lift ads caused vs what would have happened anyway.
Total revenue ÷ total ad spend — the blended, attribution-free ROAS.
Revenue attributed to ads ÷ ad spend — the fastest efficiency read.
Ad spend divided by conversions — the price of one action.
Incrementality test that turns spend on/off by region to isolate impact.
Statistical model measuring channel contribution using historical data.
Conversion from a user who saw but didn't click an ad.