Multi-Touch Attribution Models Compared: Which One Should You Use?

Last click, first click, linear, time decay, data-driven — each model tells a different story about your marketing. Here's when to use each and why they disagree.

attributionGA4multi-touchmarketing measurementROASanalytics

A customer sees your Instagram ad on Monday. Clicks a Google ad on Wednesday. Opens your email on Friday. Buys on Saturday via a direct visit.

Which channel gets credit for the sale? The answer depends entirely on which attribution model you use — and each model gives a completely different answer.

The Six Models

1. Last Click (Google Ads Default)

100% credit to the last touchpoint before conversion.

In our example: Direct visit gets all credit. Your $5,000 in Google Ads and $3,000 in Meta Ads show zero attributed revenue.

When it’s useful: Never as the only model. It systematically undervalues awareness and consideration channels.

When it misleads: Almost always. It credits the last step of a multi-step journey, ignoring everything that brought the customer there.

2. First Click

100% credit to the first touchpoint.

Our example: Instagram ad gets all credit. Email and Google Ads show nothing.

When it’s useful: Understanding which channels drive initial awareness. Good for top-of-funnel analysis.

When it misleads: Overvalues channels that introduce but don’t convert. You might pour budget into awareness campaigns that never close deals.

3. Linear

Equal credit split across all touchpoints.

Our example: Instagram (25%), Google Ads (25%), Email (25%), Direct (25%).

When it’s useful: Fair representation when all channels contribute equally. Good baseline for comparison.

When it misleads: Assumes every touchpoint matters equally. A random Display ad impression isn’t worth the same as the retargeting email that closed the deal.

4. Time Decay

More credit to touchpoints closer to conversion. Exponentially weighted.

Our example: Direct (40%), Email (30%), Google Ads (20%), Instagram (10%).

When it’s useful: Businesses with short consideration cycles (impulse purchases, low-cost products). Rewards the channels that push people to buy.

When it misleads: Undervalues the awareness stage for high-consideration products (B2B, luxury, enterprise software) where the first touch is critical.

5. Position-Based (U-Shaped)

40% to first touch, 40% to last touch, 20% split across middle.

Our example: Instagram (40%), Direct (40%), Google Ads (10%), Email (10%).

When it’s useful: Balances awareness and conversion. Recognizes that introducing a customer and closing the deal are both important.

When it misleads: Still arbitrary. Why 40/40/20? What if the middle touchpoints are what actually convinced the customer?

6. Data-Driven (GA4 + Google Ads Default)

Uses machine learning to assign credit based on actual conversion data.

Google analyzes all converting and non-converting paths. Touchpoints that appear more frequently in converting paths get more credit.

Our example: Depends on your data. If Google Ads clicks appear in 80% of converting paths but only 30% of non-converting paths, Google Ads gets proportionally more credit.

When it’s useful: When you have enough data (300+ conversions/month). It’s the most accurate model because it’s based on YOUR customers, not a theoretical framework.

When it misleads: Black box — you can’t see the exact weighting. Requires significant conversion volume. Biased toward Google channels (Google’s model naturally sees more Google touchpoints).

The Comparison Table

ModelBest ForWeaknessMin Data Needed
Last ClickNone (legacy)Ignores everything except final touchAny
First ClickAwareness analysisIgnores conversion driversAny
LinearBaseline comparisonAssumes equal contributionAny
Time DecayShort sales cyclesUndervalues awarenessAny
Position-BasedBalanced overviewArbitrary weightingAny
Data-DrivenMature advertisersBlack box, needs volume300+ conversions/month

Which Model Should You Use?

For Google Ads Optimization

Use data-driven (it’s the default for Smart Bidding). Google’s algorithm needs conversion signals to optimize bids. Data-driven gives it the most accurate picture.

If you don’t have 300+ conversions/month, Google falls back to a simplified model. Focus on getting your conversion tracking solid first.

For Cross-Channel Budget Allocation

Use position-based or time decay in GA4 for reporting. These give reasonable credit to both acquisition and conversion channels.

Never use last click for budget allocation — it tells you to cut the channels that generate demand (social, display, content) because they don’t show up as the last click.

For Executive Reporting

Use blended ROAS (total revenue / total ad spend). Attribution models are approximations. Your finance team doesn’t care which channel “gets credit” — they care whether total revenue exceeds total spend.

Read our ROAS calculator guide for the math.

Why Platforms Disagree

Google and Meta always show different numbers. This isn’t a bug — they use different attribution:

PlatformDefault ModelDefault Window
Google AdsData-driven30-day click, 1-day view
GA4Data-driven (cross-channel)Session-based
Meta7-day click, 1-day view7-day click, 1-day view
TikTokLast click28-day click, 1-day view

A single purchase can be claimed by ALL platforms simultaneously. This is why platform ROAS doesn’t add up.

How to Configure Attribution in GA4

GA4 → Admin → Attribution settings:

  1. Reporting attribution model: Choose data-driven (default) or select a rule-based model
  2. Lookback window: 30 days for acquisition events, 90 days for all other events
  3. Paid/organic channels: Include or exclude organic touchpoints

Changes apply to reports retroactively — they recompute historical data with the new model.

The Real Answer: Use Multiple Views

No single model tells the complete story. Here’s the practical approach:

  1. Google Ads: Use data-driven for bidding optimization (let it ride)
  2. GA4 reporting: Use position-based for cross-channel comparison
  3. Budget decisions: Use blended ROAS as the final check
  4. Monthly review: Compare platform-reported conversions to actual revenue — if the gap is growing, your tracking needs attention

When Attribution Doesn’t Matter

For most businesses under $20K/month ad spend:

  • You’re running 1-2 channels
  • The attribution model barely changes the numbers
  • Your time is better spent on creative testing and audience optimization
  • Fix your conversion tracking — accurate data matters more than the model

Attribution models are a scaling problem. When you’re spending across 5+ channels and need to justify budgets, the model matters. At smaller scale, just make sure you’re tracking correctly and the ROAS is above break-even.