← All Insights

Catalog & SEO Systems

The Best Automation Isn’t Always the Fanciest

A transparent relevance model made category relationships repeatable at scale.

Charles Montgomery · Founder, Alter eTail

From developing a Python-based category internal-linking workflow

Imagine asking a merchandising team to hand-pick related categories across a large product catalog. That may work for a few dozen pages. At enterprise scale, it becomes inconsistent, repetitive work.

To make the decision more repeatable, Alter eTail’s founder developed a Python workflow that compared category titles using TF-IDF and cosine similarity. It produced text-relevance scores and suggested relationships between categories.

The output was imported into a PIM to populate a related-categories module. The model wasn’t pretending to understand every product the way an experienced merchandiser might. It was doing a defined, inspectable job.

That distinction matters. The value wasn’t in using the newest AI model. It was in converting a recurring manual decision into a process a team could reproduce, question, and improve.

At Alter eTail, we judge automation by the bottleneck it removes and the control it preserves—not how impressive the technology sounds.

These perspectives draw on our founder's experience in previous roles, not Alter eTail client engagements.

Facing a Similar Challenge?

Let's talk about the opportunities, decisions, and challenges facing your eCommerce business.

Start a Conversation