Show each segment a different storefront — and prove it converted

No. Search a product in the wizard and choose Create AI-optimised template. Comver AI scans the live product page, generates CRO recommendations and writes a new templates/product.comver-ai-<id>.json into your live theme with benefit bullets, a delivery bar and offer blocks pre-configured. Open it in the Shopify theme editor and change anything. If the AI pipeline is unavailable, Comver clones your base product template so you still have a variant to customise.

Visitors are bucketed deterministically by visitor id, so the same person always sees the same variant, split per your slider (10–90%). The storefront script records experiment_viewed and experiment_converted on add-to-cart and checkout intent. Results use a two-proportion z-test: below 100 sessions per variant the label is “low confidence” and no winner is declared; z ≥ 1.96 is medium (≈95%), z ≥ 2.58 is high (≈99%). Promoting the winner means assigning the winning template to the product.

Yes — by Customer Intelligence segment (19 named segments such as High Intent Visitors, Cart Abandoners, Facebook Ads Traffic, VIP customers, Non-subscription customers), by a saved audience, and by behavioural conditions: time on site, page views, cart value, cart contains product or category, UTM, device, country, language, logged-in state, active subscription. Triggers include timed delay, exit intent, scroll depth, cart inactivity, PDP views and URL contains.

Email (Resend), SMS (Twilio), a Callio voice call with SMS fallback, a unique discount code, an on-site experience (popup, AI Shopping Assistant message or offer, announcement bar, product recommendation, personalised discount, content block or PDP variant), a webhook, a customer tag and an audience move — plus delays, conditional splits on purchase, audience membership, LTV, order count and consent, and random or channel A/B splits.

Customer Activation settings let you require Shopify email and SMS marketing consent, add a signed unsubscribe link to marketing emails and STOP instructions to SMS. Suppression is enforced in the dispatch layer before any provider call, a manual suppression list exists per channel, and newsletter sends apply consent and suppression to every recipient. Callio follow-ups can be limited to people with a phone number who consented to SMS.

The product keeps counted and modelled apart. Attributed revenue, sessions, views, submissions, ATC rate and orders are counted from events and orders. Monthly opportunity, recoverable per month and estimated revenue potential are modelled from your own AOV and conversion rate and carry the estimate mark and the sentence “Always modelled, never counted.” When there is no data the UI shows an em dash or an empty state, not a demo value.

56 premade definitions in four groups: 18 high-intent / visitor recovery (High Intent Visitors, Viewed Product 3+ Times, Added to Cart No Purchase, Checkout Started No Purchase, Visitors From Paid Ads / Google Search / AI Search No Purchase, Exit Intent, Coupon Intent…), 11 traffic sources (Facebook / Google / TikTok Ads, Google Search, AI Search, Instagram Organic, Email, SMS, Direct, Referral, Influencer / UGC), 15 lifecycle (New, Active, VIP, Repeat, Loyal, At-Risk, Churned, Winback, Subscription Candidates, Discount Sensitive…) and 12 commerce behaviour (Cart / Browse / Checkout Abandoners, Bundle Buyers, Category Loyalists, Free Shipping Seekers…). Each carries a recommended action, reasoning and a suggested experience, offer and A/B hypothesis.