Free e-Book:The Modern Data Stack:A Technical Roadmap.Download for free now!
Unlocking the Power of lift based Multi-Touch Attribution

Unlocking the Power of lift based Multi-Touch Attribution

A Case Study with MercadoLibre
Gian Franco Lancioni

Posted by Gian Franco Lancioni

on April 23, 2024 · 3 mins read

About Company

MercadoLibre is the largest online commerce and payments ecosystem in Latin America. Through a suite of technology solutions including Mercado Pago, Mercado Ads, Mercado Envios, and Mercado Crédito they enable customers in 18 countries to carry out their commerce offering solutions across the entire value chain.


Online attribution systems typically try to measure the contribution of each advertisement touchpoint (usually ad clicks) in conversion journeys. This means trying to accurately measure the incremental lift brought by each touchpoint. Correctly measuring this is a critical step to be able to properly understand the true impact of each marketing campaign and optimize marketing budget allocation. Most online attribution systems suffer from several limitations:

  • They only contemplate a fixed set of touchpoints (last-click, 7-day time decay, etc.)
  • They weigh the contribution according to some fixed rule (first click 100%, even distribution, etc.) that regularly overestimates the true contribution of bottom-of-the-funnel channels and doesn’t align with lift test measurements.
  • They don’t factor in the order of the touchpoints or user-level features that might cause the same set of ad touchpoints to influence conversions differently for different users.

In short, most online attribution systems are not truly Multi-Touch (MTA) and do not properly measure the contribution of each touchpoint. Additionally, some SaaS providers might provide closed-source MTA systems but not being able to truly understand, access and tune the underlying models severely limits the usefulness of such an important system.

MercadoLibre’s team sought to design and implement a new MTA data-driven approach that could mitigate most of the usual shortcomings. The new solution had to apply advanced ML models that could accurately account for business variables such as channel interaction, conversion prediction, time dependency, user characteristics & incremental contribution in a single solution.

By improving their attribution models, the client hoped to achieve greater real-time measurement of what were the ad touchpoints driving true incremental marketing results that would help them better optimize their marketing spend in a daily or hourly cadence.

Read more about MTA and different types of allocations here.



Our teams collaborated to design and implement a new multi-touch attribution platform using advanced deep-learning models for understanding true click-level impact.

This involved a thorough exploration of the client's industry drivers and variables, research of the latest scientific advancements in the field, quick iteration of PoCs with validation, and incorporating realistic user-context information like customer behavior, categories, and control variables. This approach helps control and reduce misleading estimations and biases, ensuring more accurate results.


Our recently adopted multi-touch attribution model has achieved a real-time and campaign-level measurement much closer to the real lift estimations of each channel. It has also enabled marketing teams to analyze the influence of each touchpoint across the customer journeys, identifying successful strategies and areas for improvement in resource allocation. Consequently, this has led to enhanced campaign performance and more informed decision-making.


Wrapping Up

Interested in diving deeper into marketing solutions? Check some of our other content:

Marketing Content:

It’s not our first time working with Mercado Libre and other industry leaders, read more stories here: