10 chatbot flows every online store should set up

It's late August, and your inbox already has a rhythm. "Where is my order?" "My code doesn't work." "Can I swap this for a bigger size?" The same handful of requests, every few minutes, each needing the same few steps. In three months, Black Friday and the holiday rush will multiply every one of them, and add "Will it arrive before Christmas?" on top.

Those steps are exactly what a chatbot flow is for. A flow asks the right question, fetches the right data and shows the right answer, the same way every time, at 2pm or 2am. This guide gives you ten flows that earn their place in almost any online store, with the trigger, the steps, sample copy and what to measure for each.

Why most store chatbots don't help

If you've tried a chatbot before and switched it off, it probably had one of these problems:

• It fired on every page. A pop-up three seconds after landing, on every visit, trains shoppers to close it.
• It asked too much. Five questions before any answer. Most people give up at the second.
• It tried to do everything. One giant decision tree for all questions, which broke the moment someone typed something unexpected.
• It hid the human. No way out except closing the window.
• Nobody measured it. It ran for months without anyone knowing if it helped.

Good flows are the opposite: narrow, short, well timed and measured.

How a good flow is built

Every flow below follows the same anatomy. Use it as a checklist when you build your own.

Part	What to decide	Example
Trigger	When the flow starts	Visitor opens chat and clicks "Track my order"
Audience	Who sees it	All visitors / returning customers / cart value over a threshold
Messages	What you say, in as few words as possible	"Sure. What's your order number or email?"
Action	What happens with the answer	Look up the order in the store and show status and tracking
Exit	How to reach a person at any point	"Talk to the team" button on every step
Goal	The one number this flow should move	Order status tickets per week

Keep messages short, ask one thing at a time, and always tell the shopper what happens next.

The ten flows

1. Order status lookup

The flow that removes the most tickets in almost every store.

• Trigger: shopper picks "Where is my order?" in chat, or types something similar.
• Steps: ask for order number or email, verify, show status, carrier and tracking link. If the parcel is late, apologise and offer a person.
• Measure: order status tickets reaching the team, week over week.

"Here's your order #10482: shipped on August 14 with DHL. Track it here. Anything else I can help with?"

2. Holiday shipping cut-off

In December, "Will it arrive before Christmas?" can outnumber everything else.

• Trigger: visitor on a product or cart page, from early December until your last delivery date.
• Steps: a short proactive message with the order-by dates for each region, plus a link to express shipping.
• Measure: shipping-date questions, and conversion of visitors who saw the message.

"Ordering a gift? Order by December 17 for standard delivery in the US, or December 21 with express."

Change the dates to your own carrier deadlines and switch the flow off the day after the last one, or replace it with a message about delivery after the holidays.

3. Cart recovery on exit

• Trigger: visitor with items in the cart moves to leave the page, or sits on checkout without acting.
• Steps: ask if anything is stopping them, with quick replies like "Shipping costs", "Delivery time", "Sizing" and "Just browsing". Answer the concern, or connect them with a person.
• Measure: recovered carts and revenue from conversations started by this flow.

Avoid offering a discount by default. You'll train shoppers to wait for it. Answer the real objection first.

4. Discount code helper

• Trigger: shopper writes about a code, or clicks "My code doesn't work".
• Steps: ask for the code, explain the common reasons (expired, minimum spend, excluded products, one per order), and offer a person if none apply.
• Measure: discount-related tickets during promotions.

Update this flow before every sale with the exact rules of the current promotion.

5. Product finder

• Trigger: visitor on a category page for more than 30 seconds, or a "Help me choose" button.
• Steps: two or three questions ("Who is it for?", "Budget?", "Indoor or outdoor?"), then show three matching in-stock products.
• Measure: add-to-cart rate and orders from the flow.

Three questions is the limit. If you need more, your catalogue needs better filters.

6. Out of stock and back in stock

• Trigger: visitor on a product page for a sold-out item.
• Steps: offer a back-in-stock alert by email, and suggest similar products that are available now.
• Measure: emails collected and sales from alternatives.

7. Return request

Returns peak in January, after the holiday orders. Build this flow well before Christmas.

• Trigger: shopper picks "Return or exchange", or mentions a return.
• Steps: ask for order number, check the order date against your window, collect the reason and photos for damaged items, then create a ticket or send the return link.
• Measure: time spent per return, and the share of returns turned into exchanges.

8. Out of hours

• Trigger: a chat starts outside your support hours and the AI agent can't fully answer.
• Steps: say clearly when the team is back, collect the email and question, and create a ticket.
• Measure: next-morning backlog and replies sent within your promised time.

"We're offline until 9am CET. Leave your email and question and we'll reply first thing."

9. First-visit welcome and email capture

• Trigger: a first-time visitor who has viewed at least two pages.
• Steps: one friendly line, an offer (free shipping threshold, a guide or a small first-order discount if that's part of your strategy), and an email field with clear consent.
• Measure: emails collected per thousand visitors, and unsubscribe rate later.

10. Post-purchase check-in

• Trigger: a few days after delivery, when the customer next visits your site or opens chat.
• Steps: ask if everything arrived as expected. A "yes" leads to a review request; a "no" goes straight to a person.
• Measure: problems caught early, and reviews collected.

Which flows to build first

Start where the volume is, then add revenue:

Order	Flow	Why first
1	Order status lookup	Biggest single source of tickets
2	Holiday shipping cut-off	Seasonal peak in December
3	Return request	January returns wave
4	Out of hours	Protects response times overnight
5	Cart recovery	First flow that directly adds revenue
6 to 10	The rest	Once the first five run smoothly

Common mistakes

• Too many triggers on the same page. One proactive message per page view, at most.
• No exit to a person. Every step needs one.
• Stale content. Promotion rules and shipping dates change; flows don't update themselves.
• Building a flow for a question the AI agent could answer. If the answer is in your help content, let the AI handle it and save flows for tasks with steps.
• Never looking at the numbers. Review each flow's goal monthly and switch off the ones that don't help.

How to roll them out before the holidays

1. This week: build the order status and out-of-hours flows, and test them with real order numbers.
2. In September: add the return request and cart recovery flows while traffic is steady, so you can measure what they do before the peak distorts the numbers.
3. By late October: set up the shipping cut-off flow with this year's carrier deadlines, ready to switch on in early December, and add the product finder for gift shoppers.
4. Every month: check each flow's goal, update copy and dates, and remove what doesn't work.

Flows handle the structured tasks; an AI agent handles the rest. To see how the two fit together, read what AI chatbots can and can't do for online stores, and for more on recovering sales, our guide to reducing cart abandonment with live chat.