E-commerce
July 1, 2026
A customer may be faced with an automated decision without always recognizing it: cancelled order, refused payment, unapplied promotion, blocked account, personalized recommendation, or support routing. They want to know why and if they can contest it.
The chatbot must explain what can be shared, direct to the available recourses, and transfer cases where an authorized person must verify. It must avoid revealing sensitive internal rules, particularly regarding fraud or scoring.
This guide shows how to handle customer questions about automated decisions with clarity, caution, and respect.
Summary
Why must automated decisions be explained?
An automated decision can seem arbitrary if the customer does not understand the general reason or the next step. The feeling of unfairness increases when the decision blocks a purchase, a payment, a benefit, or account access.
The chatbot must provide a helpful explanation without exposing rules that could be bypassed. It must also offer a path for verification or dispute when the policy allows for it.
Transparency does not mean revealing the entire algorithm; it means explaining the decision, its limitations, and the possible recourse.
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 to recognize?
Automated decisions may concern payment, anti-fraud, order cancellation, eligibility for a promotion, personalized recommendations, account limitation, support prioritization, or automatic action denial.
The bot must recognize the category before responding. An anti-fraud cancellation and a non-cumulative promotion do not require the same level of explanation.
level of explanation.
How to explain without exposing internal rules?
The chatbot can provide a general reason: incomplete information, unmet eligibility criteria, verification required, unconfirmed payment, or security rules. It must not detail thresholds, signals, or methods that could be abused.
If the customer asks for proof or disputes the decision, the bot must direct them to the appropriate support team rather than debating the decision.
How do I submit an appeal?
The recourse can be a human review, a ticket, an information update, a payment retry, an account correction, or a privacy procedure depending on the subject.
The chatbot must explain what the customer can provide and what outcome is possible, without guaranteeing that the decision will be reversed.
How do you keep a respectful tone?
An automated decision can be perceived as an accusation, especially in risk control situations. The bot must avoid guilt-inducing phrasing and prefer neutral language: “the order could not be validated automatically”.
The customer must feel that a review remains possible when the situation justifies it.
Which flow to follow?
The flow must explain without weakening the rule.
Identify the decision concerned: payment, order, promo, account, recommendation or support.
Give the general explanation available without revealing internal thresholds.
Check simple actions: information, cart, payment, eligibility or preferences.
Offer recourse, human verification or official procedure depending on the policy.
Transfer disputes, fraud, personal data, account blockages and sensitive decisions.
Which messages should be used?
To explain: "This action could not be automatically validated according to the rules applied to your file."
For appeal: "I can forward your request for human review if policy allows."
For limitation: "I cannot detail the internal security criteria, but I can guide you through the next steps."
When to transfer?
Escalation is necessary if the customer disputes the decision, requests a human review, brings up their data rights, reports an account block, or challenges a cancellation related to security.
The bot must convey the decision, context, account, order, general rule, action already attempted, appeal request, and urgency level.
Which KPIs should be monitored?
Track contested decisions, human reviews, reversed decisions, dropouts, complaints, privacy requests, account blocks, and processing times.
These indicators show whether automated decisions are understood and reasonably challengeable.
Which mistakes should be avoided?
Avoid revealing anti-fraud rules, accusing the customer, refusing any recourse, promising a reversal, or using technical vocabulary that is difficult to understand.
The chatbot must provide useful transparency without weakening necessary controls.
How can Qstomy help?
Qstomy can connect the chatbot to authenticity certificates, orders, payment statuses, cancellation policies, warranties, automated decisions, and support procedures to respond clearly, and then transfer sensitive cases with an actionable summary.
The chatbot helps the customer understand what is happening without inventing a certificate, a payment capture, a cancellation reason, a warranty decision, or an automated rule that has yet to be confirmed by a reliable source.
Explore AI support, the AI sales agent or request a demo.
Key takeaways
Key Takeaways
An automated decision must be explained with its category, general reason, possible action, and available recourse.
What the Customer Needs to Understand
The customer must understand what is blocking, what they can correct, and when a human review is possible.
The Right Boundary for the Chatbot
The chatbot can explain simple cases, but it must hand over disputes, personal data, anti-fraud, and sensitive blocks.
Enzo
July 1, 2026
