Social Commerce Support Playbook
How to run AI-assisted support across comments, DMs, live commerce, marketplaces, messaging apps, human review, and back-office action automation.
Purpose
This playbook helps teams configure Aijia Customer Service for social commerce support. It is designed for merchants that handle customer questions across short video, live commerce, marketplace stores, messaging apps, and private-domain channels.
The operating goal is not to automate every interaction. The goal is to route each case into the right level of automation, context lookup, review, or human handoff.
Channel Model
Map each channel into one of four support surfaces:
- public comments: short replies, moderation, DM routing, complaint detection;
- private messages: product consultation, coupon questions, order support, after-sales;
- seller centers: order lookup, refund status, logistics state, dispute evidence;
- messaging and private domain: WhatsApp, LINE, WeChat, WeCom, communities, repeat purchase.
Each surface should have its own tone, risk rules, and review requirements.
Recommended Workflow
- Capture message, channel, language, customer context, and campaign context.
- Classify intent: pre-sale, coupon, order, logistics, refund, dispute, complaint, membership, or unknown.
- Retrieve approved knowledge from product, logistics, campaign, and after-sales policies.
- Decide whether the response can be drafted, sent, escalated, or sent to back-office context lookup.
- If back-office action automation is needed, operate only inside authorized sessions and collect evidence.
- Route refund, compensation, address, dispute, and policy-sensitive messages to review mode.
- Store final answer, evidence, approval record, and case summary.
Risk Rules
Start with strict review rules:
- public complaints should be reviewed before sensitive replies;
- refund and compensation promises require approval;
- restricted product claims require approved wording;
- platform policy conflicts should be escalated;
- uncertain identity or order matching should not trigger direct execution.
Relax rules only after the team has enough examples, QA results, and supervisor confidence.
Knowledge Requirements
Prepare knowledge by channel:
- product facts and fit guidance;
- campaign and coupon rules;
- shipping timelines by region;
- return and refund policy;
- marketplace-specific constraints;
- brand tone examples;
- escalation phrases and handoff rules.
Knowledge should be versioned and reviewed. Do not let one-off campaign messages become permanent policy without an owner.
Evidence Requirements
For important cases, keep:
- original customer message;
- retrieved knowledge;
- AI draft and action plan;
- screenshots from seller center or logistics pages;
- human approval record;
- final sent message;
- case tags and resolution summary.
This evidence is what makes AI-assisted execution manageable at scale.
