What is Marketing Attribution? Models Explained
Short answer: Attribution assigns credit for a conversion to the marketing touchpoints that led to it. The model you choose changes which channels look profitable — so it directly affects where you spend. No model is perfect; the trick is understanding each one’s bias.
Why attribution matters
A customer rarely converts on a single touch. They might see a Facebook ad, read a blog post, click a Google search ad a week later, then finally buy after an email. Which of those gets credit for the sale? That decision — attribution — determines which channels look worth funding and which look like a waste. Get it wrong and you’ll cut the channels that actually drive growth.
The main attribution models
Last-click attribution
100% of the credit goes to the final touchpoint before conversion. Simple and still the most common, but it systematically undervalues awareness and mid-funnel channels (like Display, social, and blog content) that start the journey.
- Best for: simple funnels, last-touch-heavy businesses
- Bias: over-credits bottom-of-funnel channels like branded search
First-click attribution
100% of credit goes to the first touchpoint. Useful for understanding what drives initial awareness, but it ignores everything that closes the deal.
- Best for: evaluating top-of-funnel discovery channels
- Bias: over-credits awareness channels, ignores conversion drivers
Linear attribution
Credit is split equally across every touchpoint. Fairer than single-touch models, but treating a minor blog visit the same as the final closing click isn’t realistic either.
Time-decay attribution
Touchpoints closer to the conversion get more credit than earlier ones. A reasonable compromise that reflects that recent interactions tend to matter more — though it still undervalues early awareness.
Position-based (U-shaped) attribution
Typically 40% credit to the first touch, 40% to the last touch, and 20% split among the middle. Good when both discovery and closing matter and you want to recognise both ends of the journey.
Data-driven attribution
Uses machine learning to assign credit based on the actual conversion patterns in your account. It’s the most accurate model when you have enough conversion volume — it learns which touchpoints genuinely move people toward converting. Now the default in Google Ads.
- Best for: accounts with sufficient conversion volume
- Requirement: needs a minimum number of conversions to model reliably
The attribution problem nobody fully solves
Every platform claims credit for conversions using its own attribution. Google and Facebook will both report that they drove the same sale — so if you add up platform-reported conversions, you’ll double-count. This is why savvy marketers cross-check platform numbers against a blended, top-down metric.
MER (marketing efficiency ratio) sidesteps attribution entirely: total revenue ÷ total ad spend. It can’t be gamed by attribution windows because it ignores them. See ROAS vs MER and blended ROAS explained.
Common mistakes
- Trusting platform ROAS at face value. Each platform over-credits itself. Always sanity-check against blended revenue.
- Using last-click and then cutting “unprofitable” awareness channels. Those channels may be starting journeys that last-click never credits.
- Switching models constantly. Pick a model, understand its bias, and compare performance consistently over time.
FAQ
Which attribution model should a small business use?
Start with data-driven attribution if your platform offers it and you have enough conversions. If not, use last-click but cross-check against MER so you don’t cut awareness channels by mistake.
What is an attribution window?
The time period during which a touchpoint can receive credit for a conversion — e.g. a 30-day click window means a click counts if conversion happens within 30 days. Longer windows credit more touchpoints.
Does attribution affect Smart Bidding?
Yes. Google’s Smart Bidding optimises toward the conversions credited by your attribution model. A better model gives the algorithm better signals to optimise against.