WhatsApp Sells, but Nobody Knows Why

Conversations close orders every day. Which ones, and what made the difference, usually stays inside the head of whoever answered.

Published 2026-10-08 ยท 5 min read

WhatsApp Sells, but Nobody Knows Why

A Channel That Works And Cannot Be Explained

In Brazil, a large share of buying decisions pass through a conversation. A customer sees a product, opens WhatsApp, asks two questions, and either buys or disappears.

Stores know this channel sells. What they usually cannot say is which conversations became orders and what made the difference. The knowledge lives with the attendants: one of them "just knows" when a customer is ready, and another loses the same kind of customer without understanding why.

That makes the channel hard to improve. You cannot scale a result you cannot describe.

What Makes A Conversation Different From A Page

On a website, behaviour is scroll, clicks, and pauses. In a conversation, the signals are different and just as readable:

  • Response rhythm. A customer who answers in seconds is in a different state from one who replies after forty minutes.
  • Who is leading. Some customers ask question after question. Others wait to be guided.
  • Where the conversation stalls. Silence after the price is a different event from silence after the delivery date.
  • Returning to a topic. Asking about returns twice, in different words, shows an unresolved doubt.
  • Change of pace. Short, fast messages that turn long and careful often mark the moment a casual question became a real decision.

    • None of these require reading the content of the messages. They are patterns in how the conversation moves. That boundary matters: Fluence works with behavioural patterns only and never with what a person wrote.

    Uncertainty Looks Different In A Chat

    The same doubts that appear on a product page appear in a conversation, with their own signature.

    A customer who is unsure about price tends to go quiet immediately after hearing it, and may come back hours later with a question about payment terms. Shipping doubt shows as repeated questions about the date. A trust question shows in requests for proof: a photo of the real product, a store address, a confirmation that the exchange is free.

    For the attendant, or for an automated assistant, the practical question is the same as on the site: what is unresolved, and what would resolve it?

    The Handoff Is A Decision Too

    Most stores now combine automated answers with human attendants. The moment of handoff between them is one of the highest-value decisions in the channel, and it is usually governed by a fixed rule: after three messages, or when a keyword appears.

    A better handoff reads the state of the conversation:

  • A customer moving quickly with simple questions is well served by automation. Sending them to a queue slows down a purchase.
  • A customer whose rhythm has changed, who repeats a question or goes quiet after a key answer, is showing a doubt that a person resolves better.
  • A customer who has stopped responding after the price does not need a third automatic follow-up. They may need a different answer, or nothing for now.

    • Start By Making The Channel Legible

    Before optimising anything, a store can ask three questions about last month's conversations:

    1. Which conversations ended in an order, and at which point did the others stop?

    2. What happened right before the stop: price, date, payment, or trust?

    3. What did the attendants with the best results do differently at that point?

    The answers turn individual intuition into something the whole team can use. They also give an automated assistant a rule better than "follow up after 24 hours."

    WhatsApp already sells. The work is finding out why, so that it stops depending on who happened to answer.

    See the WhatsApp Commerce use case.