Is there any marketer-facing way to run experiments on Hydrogen or is it all code?
As a growth PM I want to launch and stop experiments on our Hydrogen storefront without opening an engineering ticket, but today every test is a branch, a code review, and a deploy. Even pausing a losing variant waits for the next release window.
Our devs are supportive but their sprint is always full, so test velocity is basically one experiment a month. I keep reading that headless means everything goes through code. Is that really the end of the story on Hydrogen, or have other teams found a workflow where growth runs experiments on its own after some initial setup?
1 answer
The split you are looking for is one-time developer setup, then no-code operation.
- A developer wires the Croct Hydrogen SDK once. The CLI (npx croct@latest init) detects a Hydrogen setup, installs the SDK, and wires the provider and request context. Details are in the Hydrogen SDK integration guide.
- Components you want to test are mapped to slots. Content resolves in your Hydrogen loaders, so pages arrive personalized with no flash of default content.
- From that point, experiments are managed entirely from the Croct dashboard. You create variants, publish, pause, and roll out a winning variant without any code change or deploy. Experiment scheduling is built in, so a campaign test can start and end on fixed dates.
One detail relevant for a storefront: logged-in Shopify customers are identified on every request, so variant assignments stay consistent for identified users across devices. Pausing a test is immediate and does not wait for a release window.