Selling in Chat Changes Two Numbers, Not One

Chat automation is sold in two ways. To a support manager it is a cost saving, because fewer messages reach a person. To a founder it is a sales channel, because more conversations end in an order. Both are true. Each on its own is misleading, because the two numbers move together and sometimes pull against each other.

This piece sets out the arithmetic, including the parts that do not flatter the idea.

The figures this piece uses

Sources: Baymard Institute, 2025. e-Conomy SEA 2026. Meta WhatsApp pricing, checked 24 October 2026.

The sales side

Three things produce revenue in a chat that a website page does not.

The question blocking the sale gets answered while the customer is still there. Will it fit? Will it arrive before Saturday? Can I return it? Is there one in black? A shopper with an unanswered question at the point of purchase either asks or leaves. An answer in seconds wins some of them back. An answer the next morning wins back very few.

Recommendations happen in context. A customer who describes what they need can be pointed to the right item, which a product grid cannot do. One Shopify App Store review, from a watch retailer who runs the shop alone, says: “Love the fact that it can recommend the right watch to my customer”.

The sale can complete in the chat. There is no separate checkout, no account to create and no delivery charge appearing at the last step. Those are the steps where the 70.22% of documented carts are abandoned.

How much is that worth to you? Nobody outside your business can say, and be wary of anyone who offers a figure. The e-Conomy SEA report shows the direction: most shoppers in the region now compare with AI and trust what it says. It also reports order volume uplifts “as large as 75%” for some commerce players. That is the top of what some companies saw, so treat it as a ceiling and not a forecast for you.

The support side

Support is easier to measure, because the work can be counted.

Take a month of messages and sort them into three piles.

1. Answerable from a document. Hours, delivery areas, return window, parking, materials.
2. Answerable from a system. Where is my order, is this in stock, which slots are free on Friday, what is my balance.
3. Needs judgment. Complaints, exceptions, negotiations, anything medical or legal.

Most chatbots handle pile one. Pile two is usually larger, and tools that only search documents fail on it, because the answer lives in your order records, calendar or stock. Pile three should stay with people, and businesses that automate it lose customers.

Your realistic saving is piles one and two. A vendor who quotes a single resolution percentage is quoting their best customer. Ours, with its scope attached: for one bakery merchant, AeroChat handles 87 to 94% of conversations with no person, through festive peaks. That is one merchant, one business type and one period, and we do not present it as a platform-wide figure. Ask the same scope of anyone else’s number. See resolution rate.

Where the two numbers interact

Both sales pitches leave this part out.

Selling in chat creates support volume. Every order taken in a conversation produces a “when is it coming?” message in the same conversation, usually within a day. Add chat selling without connecting delivery updates and you have created a new support queue. See WISMO.

Support quality is a sales input. The question answered at 10pm is sometimes the sale. Keeping sales and support in separate budgets hides that trade-off.

Automation that cannot hand over costs you on both sides. A customer stuck with a bot that will not fetch a person does not buy and does not come back. In our review study, support quality was the largest complaint theme at 19%, while control over what the AI says ranked fifth at 8%. Customers complained most about not getting help. See live chat handover.

A new cost line on WhatsApp

Since 1 October 2026 Meta charges for WhatsApp service messages, the free-form replies you send inside the 24-hour window a customer’s message opens. Each business phone number gets 1,000 free service messages a month, and unused ones do not roll over. Order and delivery update templates sent inside that window are now charged too.

So the number of messages it takes to finish a conversation now has a price once you pass the free tier.

• Three short messages where one clear one would do cost three times as much.
• Sending three delivery updates to prevent one “where is my order” question is a calculation. Work it out on your own volumes. See conversation-based pricing.

A worked example, with the unflattering parts left in

Take a store that gets 600 customer messages a month. The example figures below are for illustration.

Support side. About 400 messages a month no longer wait for a person. At three minutes each, including the interruption, that is about 20 hours. What 20 hours is worth depends on what the person would otherwise do. For an owner who runs the shop alone it is 20 hours back. For a team of four it will not reduce headcount. It shows up as faster replies and a shorter backlog. Be clear with yourself about which one you are paying for.

Sales side. Count last month’s conversations where a customer asked something, you answered late, and they never replied. Leave out the ones who said no. Count only the ones who went quiet. Suppose there were 25, at an average order of, for example, US$60. That puts about US$1,500 a month at stake in late replies, of which you would win back some share. Many owners are surprised the first time they run this count.

Costs to subtract. Meta’s per-message charges once you pass 1,000 replies a month on a number. The time to get your product and policy data into one consistent place. And the first month, when you will spend time adjusting the workflow before you save any.

Plan. AeroChat plans are priced on AI responses. A store like this one, with about 400 automated messages a month, needs roughly 400 to 500 AI responses. That sits at the top of Starter, which includes 500 a month. Growth includes 2,000, leaves room for busy months, and is the first plan that adds payment links so the sale can complete inside the chat.

How to work out what it is worth to you

Four numbers from your own data. None requires buying anything.

1. Messages per month, by pile. From the sort above.
2. Minutes per message. Time yourself on ten. Include the time it takes to switch back to what you were doing.
3. Conversations that went quiet after a late answer. Count them for one week. This is your visible lost-sales number.
4. Replies per finished conversation. You need it to compare any vendor’s pricing and to estimate your WhatsApp message bill.

With those four you can see whether, for your business, this is mainly a support project, mainly a sales project, or both.

Read next

• The four pieces a chat sale needs
• Meta’s Q3 2026 results and the new price of a WhatsApp reply
• Southeast Asia already buys in chat
• Resolution rate
• Ticket deflection

Measure both numbers on your own messages

Connect your channels to AeroChat and run a week of real conversations through it. You will see how many it answers without you, how many replies each one takes, and how many end in an order.

Start your 7-day free trial No credit card needed.

On Shopify? Install AeroChat from the Shopify App Store.