Skip to content
Otobiz
Sign in

Sell and support in the same thread.

The message that decides a sale and the message that saves a refund arrive in the same place. Answer both from one inbox.

Conversational commerce is the practice of selling and supporting customers inside messaging apps. For an online brand it means answering sizing, stock, shipping, and returns questions in the channel the customer already uses, then following up on the same thread with order updates and repeat-purchase prompts.

What usually goes wrong.

Pre-purchase questions go unanswered for hours
A shopper asking whether something fits, ships in time, or is back in stock is minutes from buying or leaving. When that message sits overnight in a phone somebody left at the office, the sale is decided by the delay, not by the answer.
Where is my order eats the support day
Order status is the single highest-volume question in retail, it has one correct answer, and that answer is already in a system. It should never be typed by a person, and in most small teams it is typed dozens of times a day.
The customer is five different people
The same buyer messages Instagram about a product, WhatsApp about the delivery, and email about the refund. Three tools, three histories, three people apologising for asking again.
Marketing sends to a list nobody consented to
Uploading a customer list and broadcasting to it is the fastest route to a blocked number and a damaged sender reputation. The tooling usually makes it easy to do and stays silent on whether it is allowed.

What Otobiz does about it.

One thread per customer, across every channel

WhatsApp, Instagram, Messenger, and Telegram land in one shared inbox with one contact record. A buyer who moves between channels stays one person with one history, so nobody starts from scratch.

  • Assignment, statuses, snooze, and internal notes
  • Contact record, tags, and past conversations beside the message
  • Delivery and read status on every outbound message

Answers from your own information

The AI drafts from what you connect: uploaded files, your crawled website, question and answer pairs you write, and the contact's own history. Every draft names the sources it used, and says so plainly when it has none.

  • Files, your own website pages, written notes, and Q&A pairs as sources
  • Sources shown on the draft before anyone approves it
  • Replies written in the language the customer used

Product questions from your own catalog

Load your products by hand or by spreadsheet and the AI answers from them, so sizing, materials, and availability come from your data. WhatsApp can show a single product card or your whole catalog inside the conversation.

  • Manual product entry and bulk spreadsheet import
  • Single product and catalog messages on WhatsApp
  • Card carousels on Instagram and Messenger

Order status answered from the real order

Connect your store, or push your orders in from your own systems, and the AI answers status questions from the real order. It never invents one. When there is no matching order the question goes to a person instead of getting a confident guess.

  • The customer's recent orders used to answer
  • Status changes sent as an update on the same thread
  • No order record means no answer, and a handoff instead

Broadcasts that check consent before they send

Build a segment from orders, behaviour, tags, and conversation history, and the send screens every recipient for consent and suppression first. Contacts with no opt-in basis are excluded with a reason you can read.

  • Segments from order recency, value, channel, tags, and engagement
  • Quiet hours, pacing, and per-contact frequency caps
  • Opt-out keywords handled automatically, with a confirmation

Recovery flows on the thread that already exists

Cart recovery, delivery confirmation, address checks before dispatch, and win-back for lapsed buyers all run as multi-step sequences on the channel the customer is already using, with the same consent screening as any other send.

  • Sequences triggered by a tag, a lifecycle change, or an event
  • Address and intent confirmation before a parcel goes out
  • Win-back segments built from time since last order

A person on anything that matters

Campaigns, scheduled sends, and anything that writes to another system queue for approval. The approval is bound to the exact message you approved and re-checked at send time, so what the approver saw is what goes out.

  • Approve, edit, or reject before anything leaves
  • The message you approved is the message that sends
  • Every send logged with who approved it

Common questions

Do you connect to our store platform?

Connect your store and the catalogue and order feed follow it. Products can also be loaded by hand or by spreadsheet, and order records pushed in from your own systems.

How does it answer where is my order?

From the order records in your workspace, and it never invents one. If there is no matching order, the question goes to a person.

Can we broadcast to our whole customer list?

Only to contacts with a consent basis. An imported list with no opt-in is excluded from a marketing audience with the reason shown. This protects your number more than it limits your reach.

Will the AI make up stock or delivery information?

It answers only from what you connected, and every draft names its sources. When it has none, the draft shows that and the case goes to a person.

Can we sell inside WhatsApp?

You can send product and catalog messages, reply buttons, option lists, and a call-to-action link to checkout. The conversation carries the customer to the point of purchase without a tab switch.

How do we handle returns and refunds?

Refunds and payment-related actions do not send themselves. They queue as approvals, so your team decides, and the decision is logged with the name of whoever made it.

What does WhatsApp cost per message?

Meta prices each delivered template message by category and destination country, charged to your own WhatsApp account. Otobiz shows an estimate before a send and the reported actual afterwards, with no margin added.

Can we track which ads produced which conversations?

Yes for click-to-WhatsApp ads. The click identifier is captured on the first message, stored on the contact, usable as a segment condition, and reportable back to Meta as a lead or a purchase.

Do we need a developer to set this up?

Not for channels, knowledge, contacts, segments, or broadcasts. Pushing order records in is the one step that touches your systems.