Abandoned Cart Lite for WooCommerce (by Tyche Softwares) stores tracked carts in wp_ac_abandoned_cart_history_lite (logged-in shoppers) and wp_ac_guest_abandoned_cart_history_lite (guests captured by email on the checkout page). Email templates live in wp_ac_email_templates_lite, sent records in wp_ac_sent_history_lite. SleekAI exposes those tables as named variables on the bot's system message so a shopper clicking the recovery link in their inbox lands on the cart-restored page with a bot that already knows their items.

That changes the recovery flow from email-and-checkout to email-and-conversation. A shopper returns and asks why they bounced, the bot reads the abandoned cart contents and the merge code data, and answers questions about size, shipping, or coupon directly. If the recovery email includes a coupon merge code via the upgrade path, the bot can confirm the discount is active and explain how it applies to the restored cart.

Display conditions scope the bot to the recovery URL pattern (the link in the email), so shoppers who land via a normal session do not get a misaligned message. Conversation logs capture each recovery ask with model name and token usage, useful when operators want to A/B test whether changing the recovery email template or the bot's first message clears more carts.

Abandoned cart recovery is a math problem hiding inside a marketing problem. Out of every 100 carts, about 70 abandon. The plugin recovers some of them through a well-timed email, often with a coupon.

But the recovery email is one-way: it shows up, the shopper clicks, and they land on the same checkout page that they bounced off the first time. If the original objection was about shipping or sizing or a coupon they were not sure about, the recovery click does not address it. They bounce again.

A chatbot scoped to the recovery URL turns the click into a conversation. The shopper lands and the bot greets them with their actual cart, the assigned coupon, and an open question. Sizing, shipping, returns, anything that made them hesitate the first time can be answered in seconds, with the cart restored on the same page.

The math improves because the second bounce stops being inevitable. Generic chatbots cannot do this. They have no idea what was in the abandoned cart, no access to the recovery token, no awareness of which coupon was assigned to which row in the history table.

They will either greet the shopper as new (which is jarring after a personalized email) or send them to a FAQ (which is the inbox-to-FAQ loop the recovery email was trying to escape). The variable mapping between the bot and Abandoned Cart's history tables closes that gap. It also makes the recovery program testable: the operator can split a template across two coupon levels and see in the logs which conversations actually convert, not just which emails get opened.