Case study
Drupal 7 to 8 Modernisation
Planned and delivered the migration of a healthcare marketing platform spanning more than 6,000 pages from Drupal 7 to Drupal 8, adapting templates, automating visual regression, and improving performance through caching.
- My role
- Migration planning, test automation, and performance
- Context
- Healthcare digital platform modernisation, July 2019 to February 2020, covering more than 6,000 content and search pages.
- Tech stack
- Drupal 7, Drupal 8, PHP, JavaScript, Node.js, Visual Regression
Problem
A large healthcare marketing platform had to move from Drupal 7 to Drupal 8 without losing thousands of content and search pages, breaking established templates, or turning visual comparison into an unbounded manual exercise.
Constraints
- More than 6,000 content and search pages had to remain available through the migration.
- Templates had to be adapted to a new Drupal generation while preserving customer-facing behavior.
- Visual coverage had to scale beyond practical page-by-page manual testing.
- Performance work had to account for caching behavior as well as frontend output.
Approach
- Planned the data migration and the sequence for adapting templates to Drupal 8.
- Separated reusable template work from content migration so each stream could be reviewed clearly.
- Created a sitemap-driven visual regression workflow to compare the large public surface systematically.
- Used caching strategies and repeatable performance checks as part of migration acceptance.
Architecture decisions
- Used the sitemap as the coverage inventory for automated visual comparison.
- Kept template adaptation explicit instead of masking platform differences behind brittle compatibility code.
- Treated caching as an architectural performance layer with its own validation needs.
Trade-offs
- Broad visual automation made regressions visible at scale, but meaningful differences still required human triage.
- Aggressive caching improved delivery speed while making invalidation behavior an important part of correctness.
Impact
- Migrated more than 6,000 content and search pages within the documented July 2019 to February 2020 project window.
- Automated visual regression reduced the manual comparison burden across the public site.
- The LinkedIn project record reports a 300% performance improvement attributed to the caching strategy.
What I learned
- Large platform migrations need an inventory that can drive both delivery and verification.
- Performance, templates, content, and test automation have to move as one modernisation programme.