Tests

Tests is where opportunities become real experiments. Each test has its own detail page with setup, content (variant code), and analytics.

Test status

The Tests page is a kanban board. Tests move through these statuses:

• Idea — the opportunity has been promoted to a test, but no variant has been generated yet.
• Generating — Spar is producing the variant code in the background.
• Draft — variant is generated and being edited.
• Ready — variant is finalized and ready to deploy.
• Running — live on the store, accumulating traffic.
• Paused — temporarily stopped (allocation is frozen but the test isn't finished).
• Completed — finished, with a result attached.
• Cancelled — abandoned before completing.
• Failed — something went wrong (deploy error, missing tracking, etc.).

Drag a card between lanes to update its status.

A page can only have one running test at a time. Launching a test on a page that already has one live is blocked until the other test finishes or is paused.

The three tabs on a test

• Setup — audience, allocation, hypothesis, metric.
• Content — the variant code (HTML, CSS, JS) and any content changes.
• Analytics — sample size, conversion deltas, and statistical significance over time.

The Analytics tab shows a tab for each metric that's measurable for your store, so you can check secondary metrics alongside the test's primary one. Revenue metrics like average order value are shown in your store's currency rather than a percentage, and metrics where lower is better (like bounce rate or cart abandonment) are colored accordingly.

How Spar decides when a test is "done"

By default Spar reads a chance to win (a Bayesian probability that a variation is best) rather than a fixed-horizon p-value. Spar doesn't declare a winner on the first good-looking reading: a test has to run at least seven days, clear minimum visitor and conversion floors, show positive lift, and pass your confidence threshold. You don't need to pick a fixed duration upfront, and a test can be ended manually any time.

The confidence threshold (Aggressive, Recommended, or Conservative) and the decision method are configurable under Settings → Workflow. Teams that want a formal always-valid statistical guarantee can switch the decision method to Sequential.

Where tests run

Spar can deploy tests through a few channels depending on what you've connected:

• Shopify (theme app extension) — for Shopify stores.
• GrowthBook — for teams already standardized on GrowthBook.
• Intelligems — for teams using Intelligems for pricing tests.

Configure your default execution target under Settings → Tests.

Orphan suggestions

If you have running experiments on your platform (e.g. GrowthBook) that don't have a matching opportunity in Spar, the Tests page will surface them as orphan suggestions so you can wire them up.