Case study • Welcomia
Welcomia: modernising a stock image and footage marketplace
A rebuild of a product created more than a decade earlier: product workshop, database and AWS file migration, preserved SEO routes and a public release after ten sprints.
9 min read
6 hours
Product workshop
10
Weekly sprints
2 weeks
Beta to public release
Returning to a product after more than ten years
Virrage Images returned to GMI with a plan to modernise Welcomia.com. The platform sold stock photography and footage to small companies and individuals, with a longer-term vision of allowing partners and freelancers to share their libraries.
The challenge was not a greenfield build. It was the combination of a legacy database, thousands of files, existing SEO routes and expectations for a modern public interface.
Deciding where pixel-perfect mattered
During a six-hour workshop, the team separated public areas that needed a polished interface from back-office functions where reliable operation mattered more than visual finish. That protected the customer experience without spending the same design budget everywhere.
Guest, user and administrator journeys were mapped in Miro. The output became the roadmap, delivery scope and team plan.
Migrating the database, files and watermarks
The legacy database did not match the new model, while the media library had to move to Amazon S3. The team built a proof of concept that transferred files and generated new watermarks and thumbnails.
Some source files were corrupted, so the migration was split into small resumable batches. A limited real photo set reached the test environment early, allowing the team to validate the live application rather than work only with mock data.
- Resumable migration batches.
- Automated watermark and thumbnail generation.
- Early testing with real media.
SEO was part of the migration, not an afterthought
Welcomia depended on organic traffic generated by a large content library. The new application retained the old slugs and routing structure so existing Google visits would not end in errors.
The beta launched at the end of March 2020 and the public version followed two weeks later. The project demonstrated that a product migration must cover code, data, files and historical URLs together.