Douyin’s E‑commerce Tracking Journey: From Log 1.0 to a Unified Attribution Platform
This article examines Douyin Group’s e‑commerce data‑tracking evolution, detailing the transition from early log‑free collection through Log 2.0’s failed overhaul to the streamlined Log 3.0 framework, and explains the resulting SDK, BTM/BCM management, and attribution platform that solve quality, efficiency, and analysis challenges for data engineers.
1. E‑commerce Business Status and Issues
The article focuses on the e‑commerce scenario, introducing Douyin Group’s tracking journey, solution, attribution practice, and benefits, aiming to provide data engineers with useful ideas for post‑tracking data processing.
Log Evolution
Log 1.0 : Before 2013, Douyin did not collect traffic logs internally and relied on external tools.
Log 2.0 (unreleased) : 2015‑2016 attempted a framework upgrade with unified page identifiers, but high implementation cost and lack of cross‑team cooperation prevented rollout.
Log 3.0 : Since 2017, the data team has used a simplified event‑parameter model, compatible with Log 1.0, and managed via a dedicated tracking platform.
2. Scenario Iteration
As the app evolved from a recommendation feed to a comprehensive ecosystem, user paths deepened and became more complex, creating new challenges for attribution and analysis.
Key Problems
Quality issues: missing or incorrect logs cause online incidents.
Increasing attribution blind spots make business analysis difficult.
High maintenance cost and low efficiency of tracking implementation.
3. Solution Overview
The data team built a framework consisting of tracking management (BTM & BCM), an SDK, an attribution platform, and analysis products.
Tracking Management (BTM/BCM)
BTM adopts the SPM concept (business, page, block, slot). Example:
btm:a5425.b8692.c2154.d8060encodes an e‑commerce point, personal page, order module, and review button.
BCM unifies key dimension fields (product_id, shop_id, author_id, promotion_id) to avoid inconsistent naming.
SDK Capabilities
Key‑event handling : automatically reports events such as product exposure with configurable rules.
Chain linking and parameter aggregation : BTM mode passes parameters across pages without manual forwarding.
Path reconstruction and attribution : builds a clean user path, resolves overlapping markings, and outputs attribution results.
Attribution Platform
Provides tag‑based point classification, strategy configuration, and data‑processing task generation to replace fragile document‑only attribution rules.
4. Benefits
Improved tracking quality: the “other” category is reduced below 0.X%.
Higher development efficiency: reduced effort for cross‑team tracking implementation.
Enhanced attribution capability: stable rule‑based attribution and flexible custom analysis.
ByteDance Data Platform
The ByteDance Data Platform team empowers all ByteDance business lines by lowering data‑application barriers, aiming to build data‑driven intelligent enterprises, enable digital transformation across industries, and create greater social value. Internally it supports most ByteDance units; externally it delivers data‑intelligence products under the Volcano Engine brand to enterprise customers.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.