Experiments

Experiments let you run A/B tests against an existing paywall rule. You define two-to-four variants that override the rule’s template, meter limit, message, or product list, and Latch splits incoming traffic across them and tracks conversions.

Open in Dashboard

Jump directly to the matching dashboard area while you follow this guide.

Open Experiments

Experiments list

Each card shows:

• experiment name and target paywall rule
• status: draft, running, paused, completed
• traffic share — the percentage of paywall impressions included in the test
• variant summaries (template, meter limit, message snippet, product count)
• the winning variant once one has been declared

Creating an experiment

Click Create Experiment and provide:

• Name — internal label, e.g. “Modal vs Sticky Bar Test”
• Paywall Rule — the rule to vary
• Traffic % — share of impressions enrolled (the rest see the rule’s default config)
• Variants — at least two, up to four. Any field left blank inherits from the underlying rule.

New experiments start in draft. Click Start to enroll traffic, Stop to halt it.

Reading results

Click Results on a running or completed experiment to see assignment counts, conversion counts, and conversion rate per variant.

API reference

• GET /api/v1/admin/experiments
• POST /api/v1/admin/experiments
• POST /api/v1/admin/experiments/:id/start
• POST /api/v1/admin/experiments/:id/stop

Webhook events experiment.assigned and experiment.converted fire as users are bucketed and convert — see Webhooks.