The storefront chatbot: answers from your own store

The storefront chatbot is the assistant that greets every visitor, on every theme, in your brand colors. It answers shopping questions from your actual store content — products, FAQs, and docs — offers quick replies, captures leads from real conversations, and hands anything it cannot answer to a human. Configuration lives in the dashboard; the work happens on the storefront. ## What it answers from The chatbot's knowledge is your store, not the open internet: 1. **Products** — names, prices, availability, and details from your catalog, so "is the black abaya in stock in medium?" gets a catalog-grounded answer. 2. **FAQs** — every entry you publish under **FAQs** becomes a ready answer: shipping times, return windows, payment methods. 3. **Docs** — help center articles at `/<storeSlug>/docs`, so longer questions ("how do I track my order?") get pointed at the right doc or summarized from it. That sourcing is the design principle: the chatbot is accurate because it recites your content, and it stays accurate because updating an FAQ updates the bot. When a question falls outside the content, the chatbot says so and captures the conversation for you instead of improvising. ## Quick replies Quick replies are the tappable suggestion chips that appear when the chat opens — things like "Track my order," "Shipping times," "Return policy." They do two jobs: they answer the common questions in one tap, and they teach visitors what kind of questions work. The chip labels come from your popular FAQs and can be tuned, so keep the four or five that match what customers actually ask. ## Lead capture When a visitor asks something the content does not cover — or asks to be contacted — the chatbot captures the lead: their contact details and the conversation transcript land in your dashboard. Leads surface in **Messages** and become **CRM contacts**, so a night-owl browser who wanted a custom size wakes up in your pipeline with their question attached. You reply from the dashboard; the conversation continues wherever the customer left it. ## Configure the chatbot 1. Open **Chatbot** in the dashboard (also reachable from **Settings**). 2. Set the **name** — the bot's display name in the chat header. "Safa" beats "Chat Assistant" for a store called Safa Clothing. 3. Pick the **color** — the widget's accent color, matching your brand palette. 4. Choose the **icon** for the launcher. 5. Write the **welcome message** — the first sentence customers read. Name the store and offer the top action: "Hi, I'm Safa — ask me about sizes, shipping, or your order." 6. Save. Changes go live immediately. ## Every theme renders it natively The chatbot is built into the theme system, not injected by a script you bolt on: **every theme renders the chatbot natively**, and it draws the merchant's configured colors and icon automatically. Switch themes, and the chatbot moves with the store — same name, same colors, same welcome message, same knowledge. No reconfiguration, no snippet reinstallation. ## What happens next For the first week, read the chatbot conversations in **Messages** every day. Two patterns pay for the time immediately: - Questions asked repeatedly with bad answers → write the FAQ or doc that fixes them. - Questions asked repeatedly with good answers → add them as quick replies. That loop — real questions into FAQs, FAQs into the bot — is how the chatbot deflects more tickets every month without extra work. ## FAQ **Can the chatbot answer questions about anything on the web?** No, and that is deliberate. It answers from your products, FAQs, and docs only, which keeps answers consistent with your policies and prices. **Does it place orders or check payment status?** It can point customers to the right information and capture their request; transactions still happen in checkout, and edge cases get handed to you as captured conversations. **Can I turn the chatbot off?** Yes — disable it from the chatbot configuration, and the widget stops rendering on the storefront. Re-enable it anytime; the configuration and knowledge persist.