Skip to main content
When to use which Attribution Model

Different Attribution Models Explained + Which one you should use when

Valentine Strunz-Happe avatar
Written by Valentine Strunz-Happe
Updated over a year ago

This video answers:

  • How the different Attribution models in Klar work & how they attribute orders to touchpoints/channels

  • Which Attribution model to use when

Attribution Models Explained

We're looking at this specific customer journey to go through the attribution models:

Static Attribution Models

  • Last Touch - last touch gets all the value -> 100% Email Automation in this case

  • First Touch - first touchpoint gets all the value -> 100% TikTok in this case

  • Linear - every touchpoint gets the same ratio of the full value -> 20% for each of the 5 touchpoints

  • U-Shape - first and last touchpoint get a bit more, other touchpoints in between get less but equally distributed part of the value -> 30% TikTok, 13% each Facebook Paid, Influencer and Branded Paid Search, 30% Email Automation

  • Unique - full value for each touchpoint -> 100% TikTok, 100% Facebook, 100% Influencer, 100% Branded Paid Search, 100% Email Automation

Dynamic Attribution Models (Klar unique)

  • Data-Driven - reallocating the value of touchpoints dynamically based on time spent on site, user intent, channel type, order of channels and time lags and also Zero-Party-Data like discount codes and post-purchase-survey (PPS) answers

    • If the touchpoint that's mentioned in the Zero-Party-Data (in a discount code or PPS answer) does not exist already in the journey, we'll inject a new touchpoint, adding it to the journey as last or first touchpoint

  • Marketing Mix Model (built on top of Data-Driven model) - factoring in correlations between channels and reallocating branded/direct traffic to the original source where the demand was generated, based on Machine Learning algorithm

    • PPS answers, correlations of spend impact, impression impact and engagement impact on branded traffic and conversions

When to use which Attribution Model

  • Unique: might be useful for day-to-day optimization, "which creatives have some sort of impact on a conversion"

  • We wouldn't recommend relying on the other static models as the reality is way more dynamic and blurry

  • Data-Driven stacked with Marketing Mix Model: giving you an understanding of what marketing channels are actually driving conversions -> to allocate the marketing budget to the channels that are most relevant & effective in your user's journey

    • MMM especially useful for short user journeys that seem to be mainly driven by branded/direct channels (based on clicks observable)

    • Factoring in things that don't generate a "touchpoint" like Word of Mouth as a new channel (mainly via PPS)

    • Reallocating direct / branded traffic to the original source the demand was generated

    -> Understanding what the impact of my marketing channels really is and which of them are actually driving conversions

    -> Best option to allocate marketing budget because it factors everything in

Did this answer your question?