Skip to main content
Attribution model analysis is essential for media, creative, and budget decisions. Below are complete tables and examples maintaining the original logic, now including: Linear, U-Shaped, and Last Click Non Direct & Organic.

​
How to use models in campaign analysis? (guide table)

​
When to use each model (quick best practices)

​
Last Click Paid

Use for: measuring direct performance of paid campaigns; evaluating bottom-of-funnel ads; prioritizing channels with financial return without organic/direct bias.
Essential metrics: Attributed Sales/Revenue, ROAS, ROI, CPA.

​
Last Click Non Direct & Organic

Use for: reducing noise from Direct/Organic visits in final moments; highlighting the last “active” effort from media/channel.
Essential metrics: Attributed Sales/Revenue, ROAS, non-organic session conversion rate.

​
Linear

Use for: measuring collaboration between channels in multi-touch strategies; comparing “share” of influence between touchpoints.
Essential metrics: Linear Revenue, % participation per channel, average journeys per conversion.

​
U-Shaped

Use for: balancing acquisition and closing, valuing awareness + conversion; useful in long funnels with multiple touchpoints.
Essential metrics: First/Last Revenue (weighted), middle-of-funnel Revenue, CAC/CPA per stage.

​
Markov

Use for: understanding real impact of each channel across the entire journey; justifying awareness investments; complementing First/Last analyses.
Best practices: combine with First/Last for balanced view; use to explain discrepancies between models.

​
Practical examples with comparative tables

The following numbers are illustrative to demonstrate readings and decisions.

​
1) Top of Funnel attracting new customers

Scenario: Beauty e-commerce seeking awareness and qualified traffic.
Strategy: Instagram/TikTok for top; Google for active demand; Facebook Remarketing. Reading:
  • First Click: Instagram/TikTok lead acquisition.
  • Last/Last Paid/Non Direct & Organic: Google + Remarketing close.
  • Assisted/Linear/U-Shaped: Instagram/TikTok influence the journey; U-Shaped highlights the importance of first and last touch.
Actions: increase budget on Instagram/TikTok (acquisition); maintain/optimize Google and Remarketing (closing); test top-of-funnel creatives to increase final conversion rate.

​
2) High investment with low perceived return

Scenario: Electronics store invests $50,000/month in Google, but revenue doesn’t keep up.
Objective: Improve ROAS/ROI, reallocate budgets according to real contribution. Reading:
  • Google strong in First, weak in Last/Last Paid → generates traffic but doesn’t close.
  • Facebook closes more and has higher ROAS.
  • E-mail converts, but doesn’t appear in Last Paid (not paid media).
Actions: reduce part of Google budget and reallocate to Facebook; improve targeting/keywords (bottom of funnel); create paid remarketing to capture traffic generated by Google.

​
3) Campaign performing well and ready to scale

Scenario: Edtech launched new course; Facebook and Google performing.
Objective: Identify scale potential while maintaining efficiency. Reading:
  • Facebook drives top and support (Assisted/Linear); Google closes sales (Last/Last Paid/Non Direct & Organic).
  • YouTube contributes, but with higher CPA.
Actions: increase budget on Facebook (acquisition); scale Google with bottom-of-funnel campaigns; optimize YouTube creatives to reduce CPA.

​
Additional cross-analysis tips

  • Compare First vs Last/Last Paid/Non Direct & Organic to separate attraction from closing.
  • Use Linear and Assisted to measure journey collaboration.
  • Use U-Shaped when you want to strategically reflect the weight of discovery and conversion.
  • Bring Markov to justify investments and explain discrepancies between rule-based models (First/Last/Linear/U-Shaped) and real impact.

​
Final Summary (checklist)

  • First Click → new customer acquisition.
  • Last Click → final conversions.
  • Last Click Paid → direct performance of paid media.
  • Last Click Non Direct & Organic → removes Direct/Organic noise at last touch.
  • Assisted → overall channel contribution.
  • Linear → balanced collaboration.
  • U-Shaped → balances start and end of journey.
  • Markov → real impact across the entire journey.