Magento Chatbot Builder

Magento (Adobe Commerce and Magento Open Source) is a leading open-source e-commerce platform, trusted especially by mid-market and B2B merchants who need catalogs, pricing rules, and multi-store setups that hosted platforms cannot express. That sophistication comes with an expectation: extensions must be clean, scoped, cache-aware, and auditable.

The Conferbot Magento 2 module (Conferbot_Chat) meets that bar. It embeds the AI chat widget on every storefront page and passes the logged-in customer's identity plus live store and cart context to the bot - so when a shopper with a loaded cart asks a question, your bot and your agents already see the cart total, item count, and currency.

Configuration is two steps in the admin: paste one Chatbot ID under Stores > Configuration > Services > Conferbot, set Enable Chat to Yes. All widget appearance and behavior stays in the Conferbot dashboard, delivered at runtime. The bot itself is built once in the visual builder and can simultaneously serve your other sites, WhatsApp, and Messenger.

This guide covers how the module works, real Composer and manual installation steps, configuration including Magento's default/website/store scopes, exactly what context is passed, commerce use cases, theming, AI and live agent workflows, performance with full page cache, the module's architecture and test coverage, and an honest comparison with generic live chat extensions in 2026.

For broader strategy, see the e-commerce chatbot guide and how chatbots drive e-commerce sales.

Source code & documentation: conferbot-magento on GitHub | Developer guide

The module follows standard Magento 2 frontend architecture: a layout update adds a block to the before.body.end container on every storefront page, and that block's template renders the official Conferbot loader.

What the Module Injects

<script src="https://cdn.conferbot.com/dist/v1/widget.min.js" async></script>

window.ConferbotWidget("YOUR_BOT_ID", "live_chat", { user: { id, name, email }, customData: { platform: "magento", storeName, storeUrl, storeLanguage, storeCurrency, cartTotal, cartItemCount } })

The user object is present only for logged-in customers; guests get store and cart context without identity. Everything else - flows, AI, appearance - the widget resolves server-side from the Chatbot ID over Socket.IO.

One Credential, No Secrets

The Chatbot ID is the only credential. No API key, no workspace ID, nothing secret in your page source - the ID is the same public identifier used by the plain website embed. The module's job is to place it correctly on every page and enrich it with live Magento data.

Standard Blocks, Not Hacks

Injection uses a layout XML update, a view model, and a .phtml template - the exact pattern Magento documents for frontend additions. No core overrides, no plugin interceptors on checkout, no JavaScript mixins to conflict with your theme.

Try Before You Sign Up

The public demo bot (691c970890527a0468f9b2c9) works without a Conferbot account: paste it as the Chatbot ID, enable, flush cache, and a live AI chat runs on your storefront. It is also the fastest diagnostic - if the demo bot renders and yours does not, the problem is bot configuration, not the module. Build your own at app.conferbot.com.

There are two ways to get Conferbot onto a Magento 2 store: the official module (recommended) or pasting the widget snippet into the theme's Miscellaneous HTML slot. Both load the exact same widget; the module adds customer identity and live cart context automatically. Module requirements: Magento 2.4.x (Open Source or Adobe Commerce) and PHP 8.1+.

Option A: The Conferbot_Chat Module (Recommended)

Installs like any Magento 2 extension and automatically passes the logged-in customer's identity plus store and cart context (total, item count, currency) to the bot.

Via Composer:

1. composer require conferbot/module-chat
2. bin/magento module:enable Conferbot_Chat
3. bin/magento setup:upgrade
4. bin/magento cache:flush

Manual (app/code): copy the module repository to app/code/Conferbot/Chat/ in your Magento root, then run the same module:enable, setup:upgrade, and cache:flush commands.

Configure in the admin:

1. Go to Stores > Configuration > Services > Conferbot
2. Paste your Chatbot ID (in the Conferbot dashboard: open your bot, go to Share > Embed into website, copy the 24-character Bot ID). You can also paste the full embed snippet or a dashboard URL - the module extracts the ID on save
3. Set Enable Chat to Yes
4. Save and flush the cache - the chat bubble is live on your storefront

No account yet? Paste the demo bot ID 691c970890527a0468f9b2c9 - it works out of the box.

Option B: Direct Embed via Miscellaneous HTML

The pre-module method still works: in the admin go to Content > Design > Configuration, edit your theme, and under Footer > Miscellaneous HTML paste the canonical widget snippet:

<script src="https://cdn.conferbot.com/dist/v1/widget.min.js" async></script>

window.ConferbotWidget("YOUR_BOT_ID", "live_chat")

Note that the direct embed does not pass the logged-in customer or cart context automatically - that is exactly what the module adds - unless you hand-write the options argument yourself, and hand-written context will not track Magento sessions or scopes. For any store that cares about identity or cart-aware conversations, Option A is the right choice.

Verify

Open the storefront: the bubble appears on every page. Log in as a customer, add something to the cart, and start a chat - the conversation in your Conferbot inbox shows the customer's name, email, and cart context.

Troubleshooting

• Module enabled? bin/magento module:status Conferbot_Chat
• Enable Chat = Yes in the current configuration scope (check the scope switcher)
• Chatbot ID correct, bot published, URL not in the bot's disabled list
• Flush cache: bin/magento cache:flush - with full page cache enabled the block renders into cached pages, so flush after config changes

Full guide: developers.conferbot.com/docs/integrations/magento.

The admin configuration under Stores > Configuration > Services > Conferbot is intentionally small - four fields - because widget behavior belongs in the Conferbot dashboard, not duplicated in Magento.

The Four Settings

• Enable Chat - master on/off switch, scoped (see the multi-store section)
• Chatbot ID - required. The field accepts the bare 24-character hex ID, the full embed snippet, or a dashboard URL; a backend model extracts and validates the ID on save, so a paste-happy admin cannot ship a malformed widget call
• How the Chat Appears - floating chat bubble (live_chat, default) or popup window (popup_chat)
• Widget Script URL - global setting; leave as the Conferbot CDN. Useful for staging environments or self-hosted widget builds

ACL Protected

The configuration section is guarded by its own ACL resource, so you can grant or deny access to the Conferbot settings per admin role like any other Magento configuration.

Everything Else Lives in the Dashboard

Position, colors, theme, language, auto-open, proactive prompts, welcome messages, and disabled URLs are configured in the Conferbot dashboard under Customize and delivered to the widget at runtime. Restyle the chat at any time without touching Magento, without a deploy, and without a cache flush - runtime settings bypass the page cache entirely.

Per-Page Control

The module renders storewide by design. To keep the chat off specific URLs (checkout success, CMS landing pages), add them to the disabled URLs list in the dashboard - the widget checks the list at runtime. Merchants who want widget-level page-type toggles inside the admin can compare the PrestaShop module, which ships per-page-type visibility; on Magento the dashboard list covers the same need without cache interactions.

Multi-store is where Magento earns its keep, and the Conferbot module is scoped accordingly: settings support Magento's default / website / store view configuration hierarchy (Widget Script URL is global).

What Scoping Enables

• Different bots per website: a B2C website runs a retail support bot while the B2B website runs a quote-and-account bot - same Magento instance, different Chatbot IDs at the website scope
• Different bots per store view: language-specific store views each point at a bot with flows written in that language, or one language-aware bot branches on the passed storeLanguage
• Selective rollout: enable the chat on one store view first, watch analytics, then flip Enable Chat to Yes at the default scope when proven
• Brand separation: multi-brand instances give each brand its own bot identity, welcome flow, and styling while agents work one shared inbox

How It Works Mechanically

The module reads its configuration through Magento's standard scoped config, so the effective Chatbot ID and embed type resolve exactly like any other setting: store view first, then website, then default. The widget context always carries the active store's name, URL, language, and currency, so even a single shared bot knows which storefront each conversation came from.

Scope-Aware Troubleshooting

The most common "widget missing on one store" cause is scope: Enable Chat set to Yes at default but overridden to No at the store view, or a Chatbot ID overridden with an unpublished bot. Check the scope switcher in the configuration page before anything else, then flush the cache.

One Inbox Across Everything

However you scope the bots, conversations land in Conferbot's shared inbox tagged with store context - and the same workspace can also serve your Shopify, PrestaShop, or WordPress properties if you run a mixed estate.

Context is the module's real value over a pasted script. Here is the complete payload - nothing more is collected.

Logged-In Customer Identity

When a customer is logged in, the module passes:

• user.id - the customer ID
• user.name - the customer's name
• user.email - the customer's email

Guests get no identity payload - anonymous shoppers stay anonymous unless your flow asks.

Store and Cart Context (All Visitors)

• customData.platform - always "magento"
• customData.storeName, storeUrl - the active store's name and URL
• customData.storeLanguage, storeCurrency - locale and currency of the active store view
• customData.cartTotal, cartItemCount - the visitor's live quote total and item count

Why Cart Context Changes the Conversation

When a shopper opens the chat from a cart worth real money, your bot and agents see it immediately:

• Flows can branch on cart value - offer a shipping threshold nudge to a $80 cart, route a $5,000 B2B cart straight to a human
• Agents triage by value: the conversation with cartTotal: 4900 gets picked up first
• "Do I qualify for free shipping?" gets an exact answer, not a generic policy paste

Injection-Safe by Construction

All values are serialized through a script-safe JSON encoder before being inlined - a customer name containing markup or a </script> sequence cannot break out of the script context. Type casting (cart totals as numbers, counts as integers) and guest/customer/cart variants are covered by the module's PHPUnit suite.

What Is Not Included

The module passes no order history, no addresses, no payment data, and sends nothing anywhere except the widget initialization on your own pages. Outbound webhooks (order status, abandoned cart pushes) are not part of this version - see the architecture section.

Magento merchants skew mid-market and B2B, where a single saved sale can pay for the chatbot for a year. The highest-leverage patterns:

B2B Quote and Account Support

B2B buyers ask precise questions - tier pricing, MOQs, net terms, account setup. A bot answers the structured ones instantly and routes qualified quote requests, with the buyer's identity and cart already attached, to your sales team. No form, no phone tag.

Cart-Aware Pre-Sale Rescue

The moment of doubt happens on the cart page: shipping cost, delivery time, returns. Because the widget carries live cart context, the bot answers against the shopper's actual cart - "your order ships free" beats "orders over $100 ship free". See chatbots for e-commerce sales for flow patterns.

High-Value Conversation Triage

Agents see cart totals in the inbox. When five conversations arrive at once, the one attached to a four-figure cart gets a human first - a prioritization signal generic chat tools simply do not have.

Product Discovery in Deep Catalogs

Magento catalogs run to tens of thousands of SKUs. A guided conversation ("what machine? what spec? what budget?") gets buyers to the right product family faster than layered navigation, and AI answers trained on your catalog content handle the long tail.

Order and Returns Deflection

"Where is my order?" and "how do I return this?" dominate support volume. Flows resolve the policy questions instantly and collect the details agents need for the rest, cutting ticket volume without cutting service.

Multi-Store, Multi-Language Service

With per-scope bots or language-aware flows keyed to storeLanguage, one team serves every storefront - and the same bot can meet customers on WhatsApp where B2B buyers increasingly expect to transact.

Magento themes are heavily customized, and merchants are rightly wary of extensions that fight the theme. The Conferbot widget stays out of that fight entirely.

Dashboard-Driven Appearance

Everything visual is configured in the Conferbot dashboard under Customize:

• Colors and theme - primary color, bubble, header, light or dark, matched to your brand palette
• Position - left or right, with offsets to clear sticky carts, cookie bars, and back-to-top buttons
• Branding - bot name, avatar, welcome screen
• Language - widget UI strings per audience
• Behavior - auto-open rules, load delay, proactive prompts, disabled URLs

Runtime Delivery Beats the Page Cache

Because appearance settings are fetched by the widget at runtime rather than baked into the page, restyling requires no Magento deploy and no cache flush. Change the bubble color in the dashboard and the storefront reflects it on the next page load, even with full page cache and Varnish in front. (Only the module's own admin settings - bot ID, embed type, enable flag - are cache-relevant.)

No Theme Conflicts

The widget renders in a web component with shadow DOM, so its styles and your theme's styles cannot interfere with each other in either direction. Luma, Hyva-style themes, and fully custom frontends all coexist with it the same way.

Conversation Design

The visual flow builder is where merchants differentiate: cart-value branches, B2B routing, product finders, and AI knowledge answers. Start from an e-commerce template and adapt - the same bot design works across every channel you later add.

Magento support teams deal with a mix of instant-answer questions and genuinely complex commerce issues. Conferbot layers three response modes so each conversation gets the right one.

Flows for Policy and Process

Shipping thresholds, return windows, payment options, store hours - scripted flows in the builder answer these instantly and identically every time. Flows can read the passed context: a cart-value branch, a logged-in-customer branch, a per-store-view greeting.

AI for the Long Tail

Train AI answers on your catalog pages, policies, and documentation. Free-form questions ("will this fit a 2019 model?", "is this food-safe?") get grounded answers with guardrails you define - and you choose which parts of the conversation AI may handle versus where the script stays authoritative.

Handover with Commerce Context

When a human is needed - a quote request, an angry customer, a four-figure cart hesitating - the bot hands over to the shared inbox. Agents open the conversation and already see the customer's name, email, store, currency, cart total, and item count alongside the full transcript. They answer the actual question instead of interrogating for context, then hand back to the bot.

Coverage Outside Business Hours

Commerce does not keep office hours. Overnight and weekend conversations get flow and AI answers immediately, and anything needing a human is captured with contact details and full cart context for morning follow-up. For most Magento merchants this converts the quietest support hours into the most measurable ones - analytics show exactly what the bot resolved alone.

Magento performance engineering is unforgiving - merchants invest heavily in full page cache, Varnish, and Core Web Vitals, and no chat widget is allowed to undo that.

Async and Off the Critical Path

The loader script carries the async attribute and the block renders at before.body.end - after your content. It downloads in parallel, never blocks parsing or rendering, and the widget UI initializes lazily after the page is interactive. First Contentful Paint and Largest Contentful Paint are unaffected; there is no cumulative layout shift because the bubble overlays rather than reflows.

Full Page Cache Behavior

The module is FPC-friendly by design, with one operational rule: the injected snippet is part of the cached page, so flush the cache after changing the module's admin settings (bin/magento cache:flush). Widget appearance changes made in the Conferbot dashboard need no flush - they are fetched at runtime. Customer identity and cart context are resolved appropriately for cached pages, following the same patterns Magento uses for customer-specific data.

No Theme or Core Modifications

Layout XML, view model, template - nothing else. Theme upgrades, Magento patches, and other extensions do not collide with the module, and disabling it removes every trace from the page.

SEO Neutral, Conversion Positive

The widget adds no crawlable duplicate content, no interstitial, and no layout shift, so search engines index your catalog exactly as before. The measurable upside is behavioral: answered pre-sale questions convert and reduce bounces. Our website chatbot guide covers how to measure it.

Magento merchants audit extensions before installing them - as they should. The Conferbot module is small, open source, and structured to be reviewed in one sitting.

Thin Wiring, Pure Logic

The Magento wiring (system.xml, ACL, layout XML, view model, template) stays thin. All real logic lives in one pure, dependency-free class - Model/Loader.php - which handles bot ID extraction and validation, widget option building, and script-safe JSON encoding. A backend config model extracts the 24-hex ID from whatever was pasted on save, and a source model provides the embed type options.

Tested Without a Magento Installation

Because the loader class is pure, the PHPUnit suite runs with no Magento installation: bot ID parsing (bare ID, snippet, URL, garbage), strict validation, the demo bot ID, option building across guest/customer/cart variants with correct type casting, handling of missing store fields, and script-injection safety of the inlined JSON. You can run the suite in any PHP 8.1+ environment in seconds.

Clean Admin Integration

• Configuration under Stores > Configuration > Services > Conferbot with its own ACL resource
• Defaults ship disabled - installing the module changes nothing until you enable it
• Settings respect default/website/store scopes; adding a context field is a one-file change in the loader plus a getter in the view model

Honest Scope

What the module deliberately does not do: outbound webhooks (order status pushes, product updates, abandoned cart events) are not part of this version - the module is a storefront widget integration, not a data pipeline. No API key is required and no store data leaves Magento except what renders in the widget call on your own pages.

Licensing

Open Software License (OSL 3.0) - the same license family as Magento Open Source itself - with full source at github.com/Conferbot/conferbot-magento.

The Magento extension ecosystem offers many chat options: pure live chat modules, embedded SaaS chat scripts, and helpdesk suites with chat bolted on. An honest comparison by category:

Capability	Conferbot module	Typical live chat extension	Helpdesk suite chat
Answers when agents are offline	Yes - flows + AI, 24/7	No - offline form	Ticket form
Passes customer identity to the conversation	Yes (name, email)	Sometimes	Sometimes
Passes live cart total and item count	Yes	Rarely	Rarely
Multi-store config scopes (default/website/store)	Yes	Varies	Varies
Visual no-code flow builder + AI answers	Yes	Canned responses	Macros
Same bot on WhatsApp/Messenger/other stores	Yes	Rarely	Sometimes
Open source module, public test suite	Yes - OSL 3.0	Rarely	Rarely
Module cost	Free, included in all plans	Often paid	Per-seat pricing

The structural difference: live chat extensions assume a staffed desk, and helpdesk suites optimize for tickets. Conferbot puts automation first - the bot resolves, qualifies, and captures with full commerce context, and humans handle the conversations where they add value. If your team answers every chat live within seconds all day, a plain live chat module may be all you need. If you want the storefront to sell and support around the clock, that is what this module is for. The module is free on every plan, including the free tier - paid plans start at $19/mo, see pricing - and the e-commerce chatbot guide covers evaluation criteria in depth.

From zero to a cart-aware AI chatbot on your Magento store:

1. Install the module: composer require conferbot/module-chat, then bin/magento module:enable Conferbot_Chat, bin/magento setup:upgrade, bin/magento cache:flush
2. Try the demo bot: under Stores > Configuration > Services > Conferbot, paste 691c970890527a0468f9b2c9, set Enable Chat to Yes, save, flush - a live AI chat runs on your storefront with no account at all
3. Create your account: sign up free at app.conferbot.com - no credit card required
4. Build or pick a bot: start from an e-commerce template (support deflection, product finder, lead capture) or build in the visual editor, then publish
5. Swap the ID and scope it: copy the Bot ID from Share > Embed into website, paste it into the configuration - at the default scope, or per website/store view for multi-store setups - save, flush

The free plan includes the full builder and enough conversations to validate on real traffic; paid plans start at $19/mo - see pricing.

Resources:

• Module source on GitHub (OSL 3.0)
• Developer integration guide
• E-commerce chatbot guide and chatbots for e-commerce sales
• Mixed platform estate? The same one-ID setup exists for Shopify, PrestaShop, OpenCart, BigCommerce, WordPress/WooCommerce, Ecwid, Wix, and Joomla