top of page
DrawKit-Vector-Illustration-ecommerce-04.png

Thesis

This study investigates how marketers can accurately assign credit for conversions to multiple touchpoints and channels that facilitate customer journeys. The objective is to utilize attribution models, specifically by utilizing Markov Chain analysis, to uncover similarities and differences in conversion paths among different industries. The traditional heuristic model and the Shapely Value are applied for comparison purposes. 

Jun 2022 - Jan 2023

DrawKit-Vector-Illustration-ecommerce-01.png
DrawKit-Vector-Illustration-ecommerce-02.png

Multi-Channel Attribution Model in Customer Journey with Markov Chain Analysis

  • Identified similarities  and differences in the results of the Markov Chain model across channels

  • Determine which channels appear most frequently in customer journeys

  • Identified channels with a high probability of conversion

Description of data

The statistical analysis was performed using Python and Excel.

Description of data

The study gathered online marketing data from 17 companies in Google Analytics and then classified them into three industries.

Channel: Referral, Social Media, Organic Search, Organic Video, Organic Social, Display, Direct, Paid Search, Other 

Result:  Heuristic Model

heuristic model
heuristic model
heuristic model

  • Source organic search has more conversions in the retail and service industries, while Social is the next highest

  • The number of conversions on Social was lower in the food industry compared to source Direct

  • Credit assigned to the first interaction was the most significant in each source of the three industries, except source Direct

  • The number of conversions in display sources was very low in the food and service industry

Result: Shapely Value

shapely value
shapely value
shapely value

  • The source labelled as ‘Other’ has the highest attribution values of 127.0 and 297.0 in the food and servive industry

  • Organic search has the highest attribution value in the retail industry

  • Paid search, on the other hand, has the lowest attribution values in the retail and service industries​

Result: Markov Chain

SERVICE INDUSTRY

RETAIL 

FOOD INDUSTRY

Markov Chain
Markoc Chain
Markov Chain

  • Thicker edges indicate higher transition probability weights between each source and target.

  • The Display channel has the highest probability of leading to conversion in the food and service industry. 

  • Paid search has the highest likelihood of leading to conversion in Retail. 

  • Referral to Direct has the highest transition probability in the food industry, excluding direct self-transitions

  • The Direct channel has a high likelihood of transitioning from both Paid search and Display in the service industry

  • The Direct channel has a high likelihood of transitioning from Referral and Organic search.

bottom of page