Ecommerce Live Chat: The Four Jobs It Has to Do
A chat widget on a store is not one tool. It is four: it answers the question blocking the sale, it soaks up the where-is-my-order tickets, it catches the shopper who stalls at the cart, and it has to survive November. Get the first two right and the other two get much easier.
The short version
- Automate facts, escalate money. That one rule decides most of your setup.
- Order status is the biggest single win. It is repetitive, factual and endless.
- Cart chat should offer help, not a discount. Discount prompts train shoppers to stall.
- Size your plan for November, not for June. Peak volume is the only volume that matters.
Job one: answer the question blocking the sale
Almost every abandoned product page has an unanswered question behind it. Will this fit. Is it really in stock. Will it get here before the weekend. Does it work with the thing I already own. None of these are hard questions. They are just questions your product page did not happen to answer, and the shopper is not going to email you to find out.
This is the highest-value thing chat does in a store, and it is almost entirely an AI job, because the answers are factual and the shopper wants them in seconds rather than in the morning.
| Shopper asks | Who answers | What it needs |
|---|---|---|
| Will it fit me? | AI, from the size guide | Link the chart, do not describe it |
| Is it actually in stock? | AI, from live store data | Never guess. Silence beats a wrong yes. |
| When will it arrive? | AI, from your shipping table | Country, plus the daily cut-off time |
| Does it work with my other thing? | AI first, human if unsure | A compatibility list it can quote |
| Can I have a discount? | A human, or a fixed rule | Never let AI invent money |
One warning worth taking seriously. An AI that guesses at stock or delivery dates will cost you more than it earns, because a confident wrong answer turns into a refund and a bad review. Wire it to real store data or tell it to say it will check. Both are fine. Improvising is not.
Job two: absorb shipping and returns
Every store has the same top ticket, and it is not a complaint. It is where is my order. It arrives dozens of times a day, it has a definite answer, and a human typing it out is the least useful hour in your week. This is the part of chat that pays for itself in saved time rather than in extra revenue.
| Request | Automate? | What it takes |
|---|---|---|
| Where is my order? | Automate fully | Lookup by email or order number |
| What is the returns window? | Automate fully | One policy page, kept current |
| I want to start a return | Automate the first half | AI takes order and reason, human approves |
| Change my delivery address | Send to a human | Time-critical and easy to get wrong |
| My refund has not arrived | Send to a human | Money plus frustration |
The pattern in that table is the whole policy. Facts get automated. Money and emotion get a person. A customer whose parcel is missing does not want an efficient bot, and a customer asking about the returns window does not want to wait four hours for a human to paste a link.
Job three: catch the shopper who stalls at the cart
Cart recovery through chat is real but smaller than the marketing suggests. It recovers one specific shopper: the one who stopped because something was unclear. It does nothing for the shopper who was browsing, or who saw the total with shipping and decided against it. Expect it to help a slice, not a third.
Three rules keep it from being annoying. Trigger on hesitation, meaning idle time on the cart or a second visit to checkout, not on arrival. Cap it at one prompt per session. And offer help rather than a discount, because a store that reliably hands out a code to anyone who waits has taught its customers to wait. We go deeper into the timing and the message itself in our guide to cart abandonment chat.
Worth remembering: the fastest way to hurt cart conversion is a chat window that covers the checkout button on a phone. Test the widget on a small screen before you test the copy.
Job four: survive peak season
A store that gets 200 chats in June can get several times that in the two weeks around Black Friday, and the mix changes too. More delivery questions, more first-time buyers, more urgency, and a team that is already busy packing. This is the moment chat either saves your season or becomes another inbox nobody is watching.
The preparation is boring and it works: write the seasonal answers into the knowledge base in October, give the AI a delivery cut-off date it can quote with confidence, and agree the short list of questions it must always pass to a person. Then check your allowance against a peak month rather than a typical one.
| Plan | Price | Messages a month |
|---|---|---|
| Free | $0 forever | 500 |
| Starter | $25/mo | 5,000 |
| Professional | $95/mo | 25,000 |
| Enterprise | Talk to us | Sized to your volume |
Seats come bundled, 3 agent seats plus the owner on Starter and up to 10 on Professional, which matters in December when you pull in temporary help and do not want to pay per person for six weeks of cover. Annual billing takes 20% off. The full comparison sits on the pricing page. It is also worth knowing how other tools charge before you commit: some price per seat, some meter AI separately from the plan, and some sell AI as a credit balance that simply stops when it runs out. Any of those can be the right choice, but each behaves very differently in a peak month.
Setting it up on a real store
Chat is only fast if the person answering can see the order. EzyConn goes onto a Shopify storefront as a single script tag, and its Shopify integration runs through Zapier or a signed webhook, so conversations reach the tools where your order data lives. Your team can reply from Slack, which for a small store usually means the owner answers from their phone between packing runs. WhatsApp works on every plan, the free one included.
That last point deserves emphasis. When we scanned 4,278 small business websites, only 7.5% ran any chat at all, and among those that did, a plain WhatsApp button was 53.7%, more than every dedicated chat vendor put together. The sample was weighted toward local service businesses rather than online stores, but the signal holds: customers message on the channel they already use. Our full live chat adoption study has the method and the country breakdowns. The lesson for a store is not to replace WhatsApp. It is to stop running it out of a personal phone with no history and no handover.
Frequently asked questions
Does live chat work for ecommerce stores?
It works when it answers the question that is blocking the purchase. Sizing, stock, delivery dates and compatibility are the four that stop most carts, and each has a factual answer that chat can give in seconds. It works far less well as a pop-up that interrupts someone who is already reading the product page. Judge it on the pages where people hesitate, not on the whole site average.
What questions should an ecommerce chatbot answer automatically?
Anything with one correct answer that does not move money. Order status, delivery windows, returns policy, sizing, stock and product specifications are all safe to automate. Address changes, refunds that have not arrived, damaged goods and any discount decision should go to a person. The rule is simple: automate facts, escalate money and emotion.
Can live chat recover abandoned carts?
Some of them. A chat prompt on the cart or checkout page catches the shopper who stalled because of an unanswered question, and that is a real recovery. It does nothing for the shopper who was never going to buy, or who left because of the total price. Trigger on hesitation rather than on arrival, cap it at one prompt per session, and offer help rather than a discount.
How do I handle Black Friday chat volume?
Do the work in October. Write the peak-season answers into the knowledge base while you have time, add a delivery cut-off date the AI can quote, and pre-agree which questions it must never answer alone. Then check your conversation allowance against a realistic peak month rather than an average one, because November is not an average month.
Does EzyConn work with Shopify?
Yes. The chat widget is a single script tag that goes into your Shopify theme, and conversations reach your other store tools through Zapier or a signed webhook. WhatsApp works on every plan, the free one included, if you also want to reach shoppers on the channels they already use.
Test it before your busy season
The free plan is $0 forever: 2 seats, 500 messages a month and one chatbot on your storefront. Enough to see which questions your product pages are not answering.
Start freeAdoption figures are from our own August 2026 scan of 4,278 small business websites across seven countries. EzyConn plan details are current as of August 2026.