E-commerce
June 28, 2026
Automation can quickly answer repetitive questions, but it must not absorb all situations. Some requests require judgment, empathy, responsibility, or a commercial decision.
The right question is not "what can we automate?", but "where does automation help without damaging trust?".
This guide helps to decide what must remain human in e-commerce customer support.
Summary
Why would automating everything be risky?
A chatbot can be very useful for tracking an order, explaining a policy, or qualifying a request. But it can also worsen a situation if the customer is angry, if the rule is ambiguous, or if the cost of an error is high.
Human involvement remains necessary when the response engages the relationship.
Automation should remove friction, not withdraw judgment where it is needed.
Convert over 2,000 customers on average per month with Qstomy.
The world’s 1st Shopify AI dedicated to customer conversion
Empowering 200+ e-commerce merchants
Which cases should be kept human?
Keep disputes, exceptional refunds, fraud, security, major defects, highly dissatisfied customers, VIPs, B2B, legal requests, negotiations, emotional cases, and unforeseen situations human.
Risk must guide the level of automation.
Which cases can be automated without any problems?
Automate verifiable and repetitive information: order tracking, library hours, return policy, availability, fees, size guides, refund status, product FAQs, and proof collection.
These answers must remain up to date and transfer if the context goes out of scope.
How to set the thresholds?
Set thresholds based on amount, customer tenure, sentiment, legal risk, dispute history, urgency, sensitive product, and confidence in the data. A clear threshold prevents the bot from continuing for too long.
The transfer must be predictable.
How to keep a smooth experience?
The handoff from bot to human must convey the context: order, question, proof, sentiment, answers already given, and expected action. The customer should not have to repeat their entire story.
A well-prepared transfer is often more appreciated than a poor instant answer.
Successful automation makes humans more available for the moments that matter.
It is also useful to explain to the customer why an agent is taking over the conversation. A simple phrase like “I am transferring you to a specialist to check this exception” turns the transfer into a sign of attentiveness rather than a bot failure.
The transition must be visible and reassuring.
Which flow to follow?
The flow must decide based on risk and clarity.
Identify request, customer, order, amount, emotion, risk, data, and channel.
Verify if the response is factual, repetitive, policy-based, or requires judgment.
Automate if the framework is clear and transfer if the risk exceeds the threshold.
Transmit context, evidence, history, sentiment, and recommendation to the agent.
Measure successful automation, escalations, satisfaction, errors, and reopenings.
Which examples should be used?
A refund status can be automated. The decision to grant an out-of-policy refund after a bad experience must remain human.
The difference lies in the judgment required.
When to transfer?
Transfer as soon as there is strong emotion, ambiguity, dispute, high amount, security, compliance, strategic client, public risk, missing data, or unforeseen request.
The bot must know when to stop.
Which KPIs should be monitored?
Track automation rates, escalations, errors, satisfaction by request type, reopens, time saved, cost avoided, and bot-related complaints.
These KPIs show if the boundary is well placed.
Which mistakes should be avoided?
Avoid automating exceptions, hiding the transfer, forcing the customer to repeat, letting the bot argue with an angry customer, or measuring only the automated volume.
Trust is worth more than a high automation rate.
How can Qstomy help?
Qstomy can connect the chatbot to support conversations, contact reasons, tags, orders, returns, refunds, reviews, WhatsApp channels, website chat, knowledge bases, escalation rules, and dashboards to respond with precision.
The chatbot helps the team identify pain points, improve quality, choose the right channel, and know when a human should take over without automating a sensitive or misunderstood situation.
Explore AI support, the AI sales agent or request a demo.
Key takeaways
Takeaways
Human support must remain present for judgment, emotion, exceptions, disputes, security, compliance, VIPs, and ambiguous situations.
What the customer must understand
The customer must feel that their problem is being handled at the right level, not blocked by a bot.
The right limit of the chatbot
The chatbot can automate factual questions and prepare the transfer, but it must stop when risk or emotion increases.
Enzo
June 28, 2026
