WooCommerce Checkout Field Editor stores the customized field set in three wp_options rows: wc_fields_billing, wc_fields_shipping, and wc_fields_additional. Each row is a serialized array of field definitions with a name, label, type, required flag, position, and class. SleekAI exposes those arrays as named variables on the bot's system message, so a shopper asking what the "VAT number" field is for gets the actual field's purpose, not a generic checkout FAQ.

That turns abandoned-on-checkout-because-of-a-confusing-field into a conversation. A shopper hits an unexpected "Tax ID" field and asks what it is, the bot reads the label and the field's description (often configured in the editor) and explains why the store collects it. A wholesale buyer asks why a billing field is required for their account type, the bot checks the required flag and the display conditions and explains the rule.

Display conditions scope the bot to the cart and checkout pages so the standard support bot stays on the rest of the site. Conversation logs capture each checkout-field ask with model name and token usage, which makes it easy to spot fields that confuse real customers (high ask rate on the same field name) and either rewrite the label or drop the field, instead of inventing an explainer page no one reads.

Customized checkouts are useful and dangerous in the same gesture. They collect the data the store needs (VAT, account number, delivery notes), but every added field is another reason a first-time shopper might bounce. The classic pattern is a small red asterisk next to a field name the shopper has never seen, no explanation in sight, and a checkout that fails validation when they try to skip it.

The shopper either fills in a wrong value and the store deals with bad data, or they leave. WooCommerce Checkout Field Editor already structures the data. Every custom field has a label, a type, a required flag, and (often) a description set right in the editor.

A chatbot that reads the same option arrays explains each field in plain language the moment the shopper hovers and asks. The store does not have to maintain a separate FAQ page; the bot pulls from the same source of truth that drives the form itself. Generic chatbots cannot do this.

They have no idea what wc_fields_billing is. They will either tell the shopper to contact support (a real-time question becomes a 24-hour wait) or guess at what the field is for, which is worse than guessing at a tracking number because the wrong answer breaks the order. The variable mapping between the bot and the option arrays closes that gap.

It also lets the operator iterate on the form without breaking the bot: change a label in the editor, the next conversation reflects it. That makes the checkout a living document, not a forced ad-hoc PHP-and-CSS project every time a regulation changes.