Statsig

@statsig check the 'new_checkout_flow' feature gate for user ID 'usr_42' with country set to 'US'

The feature gate 'new_checkout_flow' evaluates to **true** for user 'usr_42' in the US context. This gate is currently enabled for 15% of US users as part of the checkout-redesign experiment running through March 2025. The evaluation used rule set version 1.2.4 and matched the 'us_early_adopters' targeting segment. No exposure was logged by this check.

This read-only check is ideal for debugging gate behavior before committing to log an exposure. Note that a 'false' result is ambiguous—it could mean the user failed targeting rules or that the gate name doesn't exist in your Statsig project. Always verify gate names in the Statsig console first.

@statsig log a custom event named 'checkout_completed' for user 'usr_42' with value 149.99 and metadata including cart size and payment method

Successfully logged the 'checkout_completed' event to Statsig. The event includes user ID 'usr_42', numeric value $149.99, and metadata fields 'cart_size: 3' and 'payment_method: credit_card'. This event is now permanently recorded and will appear in your Statsig metrics dashboard within 2-5 minutes. It can be used to calculate conversion rates, revenue metrics, and experiment impact analysis.

Logged events are **permanent and cannot be deleted** via the runtime API. Use this for production analytics only—test events will pollute your metrics. Ensure your API key has write permissions and that event names match your metric definitions exactly, as Statsig is case-sensitive.

@statsig get the 'recommendation_algorithm' dynamic config for user 'usr_42', then explain which ranking strategy I should use based on the returned parameters

The 'recommendation_algorithm' config for user 'usr_42' returns: `{"strategy": "collaborative_filtering", "min_score": 0.65, "max_results": 20}`. Based on these parameters, you should use **collaborative filtering** as your ranking strategy, filtering recommendations to those with a similarity score of at least 0.65 and capping the result set at 20 items. This configuration is part of the 'rec_algo_test' experiment, currently in the treatment group 'cf_optimized'.

This example combines Statsig's config retrieval with AI reasoning to interpret the parameters. Dynamic configs return empty objects if the name doesn't exist, so always validate the response structure. No exposure is logged here—call Log Custom Exposures separately if you need attribution for experiment analysis.

When Statsig beats homegrown flags for gradual feature launches

A 6-person SaaS team shipping a redesigned billing flow needs to gate the feature behind a flag and roll it out to 10% of paid users first. The Check Feature Gate tool evaluates the flag without logging an exposure, so the team can test the gate logic in CI without polluting analytics. Once live, they use Log Custom Exposures to track which users actually saw the new flow, feeding Statsig's experiment dashboard. The dynamic config tool lets them tweak copy or thresholds without redeploying. This setup works when you have distinct user segments and want to measure conversion lift. If your product has fewer than 500 active users or you only ship once a quarter, a boolean environment variable is simpler. Use Statsig when you need per-user targeting and real-time rollback.

How layers route support tickets to the right help article

A 12-person support org uses Statsig layers to route incoming tickets to different knowledge-base articles based on user plan, region, and issue type. The Get Layer tool evaluates the routing rules and returns the article ID and metadata without logging an exposure until the agent actually opens the article. This keeps analytics clean: only real views count. The team updates layer parameters in Statsig's console to adjust routing weights as new articles go live. The ranked-variants tool could handle A/B tests on article headlines, but the support team hasn't needed that yet. This approach wins when you have more than 50 articles and routing logic changes weekly. If your knowledge base is under 20 articles or routing is static, a hardcoded lookup table is faster. Choose Statsig when routing rules are a product surface, not a config file.

When dynamic configs beat hardcoded onboarding for mobile teams

A 4-person mobile team runs onboarding experiments on their iOS app and needs to change screen order, button copy, and illustration assets without waiting for App Store review. The Get Dynamic Config tool fetches the current experiment variant and returns a JSON blob with screen definitions. The team logs events with Log Events to track drop-off at each step, feeding Statsig's funnel reports. The API-key auth means the mobile client can call Statsig directly without a backend proxy. This setup works when you ship experiments faster than you ship app updates. If your onboarding flow is stable or you only test once per quarter, baking the flow into the app binary is simpler. Use Statsig when experiment velocity matters more than client-side bundle size.