← work

skillsoft

Skillsoft

A catalog that updates without getting out of hand · 2017 to 2020

It was the most complex project I'd taken on up to that point, at Networks That Matter, now EpicDevs. The screenshots are from that era: Skillsoft's current site isn't this work anymore.

My role
Team lead and main dev. Architect of the importer.
Agency
Networks That Matter, now EpicDevs (opens in a new tab)
Stack
WordPress, PHP
Screenshot of Skillsoft: on a screen, on a desk (enlarge)

The starting point

The catalog lived in giant XML files generated by internal systems, and each team used them for its own purpose, so changing their structure wasn't an option. Migrating once wasn't enough: the XML came back every cycle.

What we did

Accepting the source as it was

We considered moving to spreadsheets, but the product team already generated XML. We decided to adapt to that format and not ask anyone to change anything.

An identity that survives changes

At first I built the slugs from the title, and any editorial tweak caused conflicts. So I combined the XML ID with the title: readable URLs and a stable key.

Screenshot of Skillsoft: wayback catalog 2019 (enlarge)

One course, one record

The same course appeared in several series. The parser unifies it by its code and accumulates its parents, instead of creating copies.

Screenshot of Skillsoft: wayback certifications 2018 (enlarge)

Testing before publishing

Each cycle started from a copy of production: we imported, marketing validated on staging, and only then did we go live, coordinating an editing window.

Screenshot of Skillsoft: on desktop and mobile (enlarge)

By the numbers

  • Between ~4,900 and ~5,900 courses per export
  • More than 4,000 redirects from old URLs
  • 4 XML files and 1 API as sources