Parse.ly Traffic BoostNo. 015

ML-powered link management for publishers

Parse.ly is a content analytics product, acquired by Automattic, used by major publishers to understand what’s working on their sites. Traffic Boost is a feature I designed that uses machine learning to help editors improve content recirculation through internal linking.

The problem: publishers know internal links drive traffic, but managing them manually is tedious. Which posts should link to which? Where should links go? Are they actually working? Most internal linking is ad-hoc and never revisited.

Traffic Boost approaches this as a system, not a series of individual decisions:

Visualization

A directed graph showing how traffic flows between posts through links — which content is sending traffic, which is receiving, where the dead ends are.

AI recommendations

ML models suggest where to add links based on content relevance, traffic patterns, and editorial context.

Automated insertion

For publishers who want it, the system can place links automatically based on rules and thresholds.

Performance monitoring

Track whether links are actually being clicked, adjust recommendations based on outcomes.

The design challenge was balancing automation with editorial control. Publishers have strong opinions about their content. They don’t want a black box making decisions. I designed the system so AI suggestions are always visible, explainable and overridable — but the defaults are smart enough that accepting them is usually the right call.

The interface handles enterprise publishing workflows: draft states, scheduled posts, editorial roles, approval chains. Links suggested in a draft behave differently than links in published content.