Attribution determines which channels get credit for conversions, which get additional funding, and which get cut. Choose the wrong attribution model and you will systematically defund your best-performing channels while over-investing in the ones that merely capture demand others created.
This isn't an academic exercise. Attribution model selection directly impacts millions of dollars in media allocation for most DR advertisers. Get it wrong and you'll optimize yourself into a corner — cutting demand generation while doubling down on demand capture until there's no demand left to capture.
The Major Attribution Models
Last-Touch Attribution
The simplest and most common model. 100% of conversion credit goes to the last touchpoint before the conversion. Someone sees a TV ad, Googles your brand, clicks a paid search ad, and submits a form. Under last-touch, paid search gets all the credit.
Pros: Simple to implement. Unambiguous. Every conversion has exactly one source.
Cons: Massively undervalues awareness channels like TV, radio, and display. Over-credits capture channels like branded search and retargeting. Creates a perverse incentive to cut the top-of-funnel activity that feeds the bottom of the funnel.
When it makes sense: Almost never as your primary model. But it's useful as one lens among several.
First-Touch Attribution
The opposite of last-touch. 100% of credit goes to the first touchpoint. That same customer: TV ad gets all the credit, and the paid search click gets none.
Pros: Highlights demand generation channels. Answers "what introduced this customer to us?"
Cons: Ignores everything that happened after the first touch. A customer might have 15 interactions with your brand before converting — first-touch pretends the other 14 didn't matter.
When it makes sense: As a supplementary view for evaluating top-of-funnel channel performance. Never as your sole model.
Linear Attribution
Equal credit across all touchpoints. TV ad, display impression, social media click, branded search — each gets 25%.
Pros: Fair and democratic. Acknowledges the full customer journey.
Cons: Treats a passive display impression the same as an active paid search click. Not all touchpoints contribute equally, and pretending they do leads to mediocre optimization decisions.
When it makes sense: Better than last-touch for channels like TV and display. A reasonable default if you're unsure.
Time-Decay Attribution
More credit to touchpoints closer to the conversion. Recent interactions are weighted more heavily than early ones. A touchpoint 30 days before conversion gets a fraction of the credit that a touchpoint 1 day before conversion gets.
Pros: Acknowledges the full journey while recognizing that recent interactions are more influential. Balances demand generation credit with conversion catalyst credit.
Cons: Still somewhat arbitrary — the decay rate is a parameter you set, not a truth you discover. Different decay rates produce different conclusions.
When it makes sense: Our recommended starting point for most DR advertisers. It's the best balance of simplicity and accuracy before you have enough data for data-driven models.
Data-Driven Attribution
Machine learning analyzes your actual conversion paths and assigns credit based on the statistical contribution of each touchpoint. Instead of predetermined rules, the algorithm discovers which interactions actually influence conversions.
Pros: Most accurate model available. Based on your actual data, not theoretical assumptions. Captures interaction effects between channels that rule-based models miss entirely.
Cons: Requires significant data volume — typically 300+ monthly conversions minimum. Can be a black box if you don't understand the underlying methodology. Sensitive to data quality issues.
When it makes sense: For any advertiser with sufficient conversion volume. This should be your goal model — everything else is a stepping stone.
The Attribution Model You Should Never Use Alone
Last-touch attribution applied to budget decisions has killed more good campaigns than any competitor. Here's the pattern we see repeatedly:
This death spiral is entirely preventable with proper attribution. Never make budget decisions on last-touch alone.
Which Model Should You Choose?
Start with time-decay if you're running fewer than 300 monthly conversions. It's imperfect but dramatically better than last-touch for multi-channel programs.
Migrate to data-driven once you have the conversion volume. Most major platforms (Google Ads, Meta) offer data-driven attribution natively now.
Always maintain multiple attribution views simultaneously. Look at your performance through first-touch, last-touch, and time-decay lenses. Where they agree, you have a clear signal. Where they disagree, you have a channel interaction worth investigating.
Attribution and Channel Strategy
Your attribution model directly impacts how you evaluate every channel in your mix:
Understanding these biases doesn't mean ignoring these channels — it means allocating budget based on true contribution rather than measurement artifacts. That's the difference between a media plan that looks good on paper and one that actually maximizes revenue.
