Halving page-load times for an 8-figure e-learning store

Challenge

The Great Courses ran a heavily customized Adobe Commerce store with years of accumulated modules and a large digital catalog. Its category and product pages had become slow enough that the conversion funnel visibly suffered.

Slow pages on a store like this rarely have one cause. Custom modules, third-party extensions, integrations and hosting had all grown over time, and each change had made sense on its own. Nobody had a single view of where the time went, so performance work tended to stall on guesses.

Approach

I measured real-user performance first, so every later change could be compared against the same baseline. Then I worked through caching, indexing, database queries, frontend delivery and infrastructure, in order of impact rather than in order of ease.

Each fix shipped as a small, reviewable change and was measured again after release. Changes that did not move the numbers were not kept. The work ran alongside the merchant’s regular feature roadmap, so the store kept shipping while it got faster.

At Guidance I also maintained a ScandiPWA headless storefront in production, including the GraphQL work behind it.

Result

Typical page-load times fell by roughly half. At Guidance I also built the ERP integrations, custom shipping modules and subscription billing for stores like this one, and I reviewed code and mentored the engineers who worked on them.

The habits from this program still shape how I approach performance: a measured baseline, a ranked fix list, small releases and a before-and-after comparison. They are the core of the performance optimization service I offer today.