シェアしてもらえるとうれしいです！

Beyond ChatGPT: Using Manus for 24/7 AI Customer Support and Earning $1,200/Month Passive Income

2026/6/25 bot-development, Python Automation

シェアしてもらえるとうれしいです！

This article contains affiliate links.


Beyond ChatGPT: Using Manus for 24/7 AI Customer Support and Earning $1,200/Month

Disclosure: All earnings, deflection rates, and KPI figures in this article are illustrative simulations based on typical SMB e-commerce support economics, not guaranteed results. This post contains a referral link to Manus; if you sign up through it, I may receive a small credit. “Mostly passive” below refers to ongoing maintenance after setup — initial deployment per client still takes real, hands-on hours.

Executive Summary

  • Offer: build and maintain a 24/7 AI support bot for e-commerce stores using Manus, on a $400/month retainer
  • Outcome target: 60–80% ticket deflection, sub-5-second first response, 24/7 coverage, consistent tone, and compliant PII handling
  • Income target: 3 clients × $400/month = $1,200/month, with the workload front-loaded into setup and tapering into light weekly maintenance
  • Time budget: roughly 6–10 hours for initial deployment per client, then ~30–60 minutes/week for maintenance and reporting

Why Manus Over “Just ChatGPT”

  • Persistent deployment: a real bot living on the client’s storefront and channels — not a one-off chat session — typically embeddable as web chat with email and messaging channels connected
  • Knowledge integration: ingest sitemaps, FAQs, policy pages, and product feeds, with changes synced automatically
  • Guardrails and escalation: enforce strict boundaries, integrate order-lookup APIs, and auto-escalate to a human when needed
  • Analytics: track deflection rate, CSAT, containment, first response time, and fallbacks — the numbers that actually prove ROI to a client
  • Maintainability: publish updates, version prompts, and keep a test suite so improvements don’t quietly break existing flows

Create your Manus account here to start building.

The $400/Month Productized Offer (What Clients Actually Buy)

Deliverables included

  • 24/7 support bot matching the client’s brand voice
  • Coverage for the top 8–12 intents: order tracking, shipping times/costs, returns/exchanges, product availability, sizing/fit, warranty/defects, discount/gift card issues, account/login problems, wholesale inquiries, order changes/cancellations, subscription basics, and contact info
  • A clear handoff and escalation path (email or ticketing system) for edge cases
  • Weekly reporting: deflection rate, unresolved intents, CSAT, top queries
  • Up to 2 small updates per week — a new FAQ, a refined intent, a guardrail tweak

Scope and limits

  • Target deflection: 60–80% for a typical SMB store, where order-tracking questions alone often represent 35–55% of inbound volume
  • Answer coverage: every policy and shipping answer must be grounded in client-provided sources — never invented
  • Escalation threshold: escalate after 2 failed attempts to answer, or if sentiment drops below a defined threshold
  • Data privacy: no PII beyond order email and order number for lookups, with a hard cap on stored logs per the client’s retention policy

Pricing and margin

  • Monthly retainer: $400/store, no long-term contract, 30-day cancel — friction-free signup tends to close faster
  • Platform costs: bundle Manus and channel fees inside the $400; aim for a 60–80% margin after platform costs and your time; raise the retainer or add a tier if usage grows
  • Setup fee: $0 for faster closes, or $250 one-time if you need immediate cash flow — this guide assumes $0 to optimize for closing speed

The ROI Math You Can Show Prospects

This is the model that actually sells the retainer — most owners won’t act on “AI support bot,” but they will act on a savings number that’s bigger than your fee.

  • Baseline: 200 tickets/month at $3–$6 per ticket fully loaded (wage + overhead) = $600–$1,200/month in support cost
  • With the bot: 60% deflection automates 120 tickets at near-zero marginal cost
  • New support cost: 80 human tickets × $3–$6 = $240–$480/month
  • Savings: $360–$720/month versus baseline — your $400 retainer sits comfortably inside that savings band, making it close to net-zero or net-positive for the client, on top of 24/7 coverage
  • Upside: faster pre-sale answers can plausibly lift conversion by 1–3% on visitor segments who engage support before purchase, though this varies considerably by store

Who to Target (and Who to Skip)

Best fit:

  • Shopify/WooCommerce stores with 5,000–100,000 monthly sessions
  • 30–300 inbound tickets/month across email, chat, and social
  • Clear policy pages and documented shipping/returns processes
  • 20+ SKUs with recurring pre-sales questions

Avoid, at least initially:

  • Highly regulated verticals requiring approval on every response
  • Stores with no documented policies, or offers that change weekly without records
  • Very low ticket volume (under 20/month) — the ROI case gets hard to make

Prospecting That Converts

Fast validation outreach email

Subject: Cut your support volume by 60% this month (no new headcount)

Hi {FirstName}, I build 24/7 Shopify/WooCommerce support bots that handle order tracking, shipping, returns, and presales. Target deflection is 60–80% with sub-5s first response. For stores handling ~200 tickets/month, that’s $360–$720/month in savings. My retainer is $400/month, no setup fee, cancel anytime. If I can’t show at least 40% deflection in 30 days, you don’t pay. Want a 10-minute Loom walking through how this would work on {Brand}? — {Your Name}

LinkedIn DM opener

Saw your recent sale on {Product}. Quick win: a 24/7 bot that handles “Where’s my order?”, returns, and sizing. My clients deflect 60–80% of tickets within 30 days. $400/mo. Want a 5-min rundown?

Loom pitch outline (5 minutes)

  • 0:00–0:45 — state the problem with their actual numbers (tickets/month, current response time)
  • 0:45–2:30 — demo the bot live, answering shipping, returns, and order lookups
  • 2:30–3:30 — walk through guardrails and escalation to a human
  • 3:30–4:30 — show the savings math on screen; make clear $400 fits inside their current cost
  • 4:30–5:00 — present the risk-free trial structure and next steps

Setting Up Your Manus Workspace, Step by Step

1. Create your account

Use this invitation link and create a separate workspace per client — “{ClientName} Support.”

2. Collect source materials (client checklist)

  • Sitemap URL (including /policies, /shipping, /returns, /faq)
  • Returns policy, warranty policy, size guides
  • Shipping carriers and SLAs, domestic and international
  • Discount policy rules — what’s allowed and what isn’t
  • Product feed or catalog (CSV, API, or sitemap)
  • Escalation email or ticketing inbox (support@brand.com)
  • Order lookup API access — Shopify private app credentials or an order-status endpoint; if unavailable, fall back to carrier tracking links plus order email/number

3. Load the knowledge base

  • Add the sitemap as a knowledge source to crawl policy and FAQ pages, with daily auto-sync enabled
  • Upload PDFs/Docs for any policies not indexed by the sitemap
  • Connect the product catalog if supported, or ingest a curated list of the top 50–200 product page URLs
  • Tag content (Policy:Shipping, Policy:Returns, Product:Sizing) to sharpen retrieval

4. Define actions for operational tasks

  • Order status lookup: inputs — order_id, email; outputs — status, carrier, ETA, tracking_url
  • Return eligibility check: inputs — order_id, item_sku, purchase_date; outputs — eligible (true/false), window_end_date, instructions_url

If a client can’t provide API access, implement “soft” actions instead: ask for order email + order number and guide the customer to carrier tracking via standard URL conventions, or link to the returns portal with fields pre-filled where possible.

5. Guardrails and privacy

  • Collect only order email and order number — never raw card data, never a full address unless strictly required for a return label
  • Set hard refusal boundaries: no medical, legal, or unapproved pricing claims; no discount promises beyond approved rules
  • Escalate when the user provides conflicting info, shows strong negative sentiment, or repeats an unresolved question twice

6. Build intents and sample utterances

  • Shipping: “How long does shipping take?”, “Do you ship to Canada?”, “What’s the shipping fee?”
  • Order tracking: “Track my order”, “Where is my package?”, “Order #1234 status?”
  • Returns/exchanges: “How do I return?”, “Is this eligible to return?”, “Exchange size M for L.”
  • Sizing/fit: “Does this run large?”, “I’m 5’10\” 160 lbs — what size?”
  • Stock: “When will black medium be back?”, “Is this in stock?”
  • Discounts/gift cards: “Promo code not working”, “How to use gift card?”
  • Warranty: “My zipper broke”, “Coverage for defects?”
  • Wholesale, account/login, cancellations, and “talk to a human” round out the remaining intents

7. Draft and install the system prompt

Paste the template below into Manus as the bot’s system/instruction prompt, then parameterize it with the client’s actual brand details and links.

8. Test flows before going live

  • Positive path: “Where is my order?” → ask for email + order number → call the lookup action → reply with status and ETA
  • Negative path: invalid order number → ask to confirm → escalate after 2 failed attempts
  • Returns path: item purchased 40 days ago → reply with policy and eligibility status
  • Sizing path: map to the size guide; never speculate about body shape
  • Discount path: expired coupon → explain policy without making unauthorized exceptions

Aim for under 5 seconds to first response and under 20 seconds to resolution on simple cases.

9. Deploy across channels

  • Web widget: grab the embed snippet from Manus; on Shopify, paste it into theme.liquid before the closing </body> tag, or use the theme customizer’s Custom HTML section; on WooCommerce/WordPress, add it via a site-wide footer HTML block
  • Email: connect the support@ inbox via forwarding or IMAP, with an autoresponder routing common questions to the bot and handing off unresolved threads to a human
  • Messaging (optional): connect Facebook Messenger/WhatsApp where supported, with proper opt-in/consent and channel-compliant templates

10. Go-live criteria

  • Bot answers at least 80% of FAQs accurately in dry runs
  • Escalation works end-to-end, including ticket creation or email handoff
  • Order lookup tested against 3–5 real sample orders
  • Logging enabled for reviewing unknown questions, with PII minimized
  • Client has reviewed tone and compliance language

A Battle-Tested System Prompt for Manus

Copy the template below and replace every placeholder in braces with the client’s actual details.

System prompt title: {BRAND_NAME} Support Assistant (Manus)

Purpose
- You are the official, 24/7 customer support assistant for {BRAND_NAME}, an e-commerce 
  store at {SITE_URL}. Your goals:
  1) Resolve common inquiries instantly using approved knowledge sources.
  2) Automate operational tasks via allowed actions.
  3) Protect customer data and adhere to policy.
  4) Escalate to a human agent when you cannot confidently help.

Tone and style
- Be concise, helpful, and friendly.
- Use the brand's voice: {BRAND_VOICE_GUIDELINES}, but never invent claims.

Knowledge boundaries
- Only answer using ingested policy/FAQ pages, product pages and size guides, 
  and approved internal notes.
- If information is missing or uncertain, say: "I don't have that information yet," 
  and offer to escalate.

PII and compliance
- Only request: order email and order number for order status; never ask for full 
  address or card details.
- Mask order numbers when echoing back (show last 2-3 digits).
- Do not store sensitive data in free-text logs.
- Comply with: {PRIVACY_POLICY_LINK} and {DATA_RETENTION_RULES}.

Allowed actions (call only if relevant)
- action_order_status_lookup(order_id, email) -> {status, carrier, tracking_url, eta}
- action_return_eligibility(order_id, item_sku, purchase_date) -> 
  {eligible, window_end_date, instructions_url}
- If an action fails or returns uncertain data, explain the situation and offer escalation.

Top intents and rules
1) Where is my order? -> ask for email + order number, call lookup action, give status/
   ETA/tracking link. Never promise delivery dates beyond the quoted ETA.
2) Shipping times/costs -> use {SHIPPING_POLICY_URL}; for unlisted regions, say we 
   don't currently ship there.
3) Returns/exchanges -> use {RETURNS_POLICY_URL} and {RETURNS_PORTAL_URL}; beyond the 
   window, explain policy and offer store credit only if permitted.
4) Sizing/fit -> refer to {SIZE_GUIDE_URL}; never speculate about body shape or health.
5) Discounts/gift cards -> only mention publicly available offers; never create 
   custom discounts.
6) Warranty/defects -> follow {WARRANTY_POLICY_URL}; provide RMA steps; escalate with 
   photos if needed.
7) Order changes/cancellations -> allowed only within {CANCEL_WINDOW_HOURS} hours if 
   unfulfilled; otherwise route to returns policy.

Escalation
- Escalate immediately if: two failed attempts on the same issue, strong negative 
  sentiment, or requests outside allowed scope.
- Escalation path: create a ticket at {ESCALATION_EMAIL_OR_TICKET_SYSTEM} with the 
  conversation transcript; tell the customer "I'm connecting you to a human agent."

Prohibited content
- No legal, medical, or financial advice.
- No price promises beyond published policy.
- No derogatory or offensive language.

Response format
- One to three short paragraphs or bullet points.
- Always include relevant link(s) and a next step.

End of system prompt.

KPIs to Track From Day One

  • Containment/deflection rate: target 60–80% within the first 30 days
  • First response time: under 5 seconds
  • Resolution time: under 60 seconds for standard shipping/return queries
  • Escalation rate: ≤20% after initial tuning, falling to ≤10% after 30 days
  • CSAT/thumbs-up rate: ≥80% on resolved interactions
  • Order status lookups automated: 50–100/month is a reasonable target for a small store

Weekly Maintenance SOP (30–60 minutes)

  • Review analytics — identify the top 10 queries and anything with low CSAT
  • Check “unknowns” and add 3–5 new canonical Q&A entries
  • Update policies (holiday shipping dates, new launches)
  • Extend intents or add synonyms; confirm retrieval points to the correct policy anchors
  • Run a 10-case regression test: shipping, returns, order lookup, discount, sizing, and two edge cases
  • Send the client a 3-number summary — deflection %, escalation %, CSAT — plus one recommended improvement

Client Onboarding Checklist

  • Confirm brand voice guidelines (3–5 bullets)
  • Confirm “no-go” answers (e.g., never quote custom discounts)
  • Get the escalation inbox and SLA expectations (e.g., human response within 24 business hours)
  • Verify order lookup connectivity (API or manual fallback)
  • Decide the return window and exceptions (holiday grace period, final sale items)
  • Confirm the data retention window for chat transcripts (e.g., 30 or 90 days)

Live Deployment Playbook (Shopify Example)

  • Add the Manus chat widget via the embed code under Channels > Web Widget
  • Set the chat to appear on all pages, with a proactive greeter on high-exit pages (cart, shipping policy)
  • Greeting copy — cart page: “Questions about shipping or returns? I can help instantly.” Product page: “Not sure about size or stock? Ask me.”
  • Add an “Order status” quick-reply that triggers the lookup flow
  • Create a Help Center page and embed the bot there too, linked from the header/footer

Email and Messaging Channel Configuration

Email autoresponder: “Thanks for reaching out — most questions about orders, shipping times, and returns can be answered instantly here: {Help Center link with bot}. If you still need help, reply to this email and a human will respond within 24 business hours.” Route emails containing order numbers to the bot first; escalate if unresolved after one exchange.

WhatsApp/Messenger (optional): use opt-in templates for order updates and FAQs, keep responses concise, link to the help center for anything longer-form, and confirm channel policy compliance for promotional messages.

Proof and Reporting That Closes Renewals

30-day report template:

Metric Modeled value
Tickets handled 220
Contained by bot 144 (65.5%)
Escalated 76 (34.5%)
First response time 2.1 seconds (median)
Bot CSAT 84%
Order lookups automated 92
Estimated cost saved 144 × $3.50 = $504
Retainer $400
Net monthly savings ~$104, plus 24/7 coverage

Round out the report with 2–3 anonymized customer examples and one recommended improvement with a projected impact — for example, adding an international shipping flow for an expected +5% containment.

How to Keep This Mostly Passive at $1,200/Month

  • Onboard 3 clients to reach the $1,200/month target
  • Monday: 30 minutes reviewing all client dashboards and drafting mini-reports
  • Wednesday: 30 minutes adding content updates across all clients (copy/paste variations)
  • Friday: 30 minutes of regression testing and sanity checks
  • Automate reminders for recurring policy updates (holidays, new product drops)
  • Template everything — system prompt, weekly report, and test suite — and reuse the same core intents and actions across clients, changing only brand voice, policies, and links

Quality Bar and SLAs You Can Promise (and Meet)

  • First 14 days: ≥50% containment, no hallucinated policies, always link to source
  • Day 30: ≥60% containment, ≤10% failed answers, ≥80% CSAT on bot-resolved threads
  • Client SLA: 99% bot uptime (platform dependent), under 48-hour turnaround on manual policy updates, immediate escalation forwarding with transcript

Common Pitfalls (and How to Avoid Them)

  • Bot invents answers for missing policies: add a strict refusal path plus a link to the real policy — “I don’t have that info yet” should always be available as a fallback
  • Order lookups fail: validate API credentials, catch errors gracefully, and escalate to a human instantly on failure
  • Tone mismatch: give the system prompt concrete brand-tone bullets and examples, and review 10 conversations weekly
  • Escalation spikes after promotions: pre-load promo FAQs and validity windows, and add a “Promo issues” intent before each campaign launches

Legal and Data Protection Checklist

  • Offer a standard data processing agreement covering chat logs and retention if a client requests one
  • Minimize PII — only request email + order number, and mask identifiers when echoing them back
  • Agree on a retention/deletion window (e.g., 30 or 90 days) and document it in the contract
  • Confirm channel-specific consent requirements (e.g., WhatsApp opt-ins)
  • Make sure the web widget supports keyboard navigation and ARIA labels for accessibility

Contract Template Essentials

  • Scope: intents covered, channels supported, weekly updates included, and what’s explicitly out of scope
  • KPIs: containment target, FRT target, CSAT target
  • Fees: $400/month, cancel anytime with 30 days’ notice
  • Platform costs: included in the retainer unless a usage threshold triggers a tier upgrade (e.g., over 1,000 automated conversations/month)
  • Data handling: retention and deletion commitments
  • Liability: clear limitations — no guarantees on conversion lift

Realistic Rollout Timeline (Per Client)

  • Day 0–1: intake and data collection
  • Day 1–2: knowledge ingestion and action wiring (order lookup, returns)
  • Day 2–3: intent buildout, system prompt, and tests
  • Day 3–4: staging and client review
  • Day 5: go live on the web widget
  • Day 6–7: connect the email channel, finalize escalation rules
  • Day 8–30: weekly improvements and KPI reporting

How to Package Higher Tiers

  • $600/month: add WhatsApp/Messenger, product recommendation flows, and one monthly A/B experiment
  • $1,000/month: add proactive cart-saver flows, email/CRM integration, bi-weekly optimization, and a 24-hour SLA on updates

Your Action Plan for This Week

  • Day 1: set up your Manus account, build a demo bot for a sample store using public policies (no PII), and record a 3-minute Loom showing order tracking, returns, and sizing flows
  • Day 2: send 30 cold emails plus 10 LinkedIn DMs, and add a Calendly link for 15-minute discovery calls
  • Day 3–4: close one pilot at $400/month with a 30-day “no pay if under 40% deflection” guarantee
  • Day 5–7: deliver the build using the steps above, and report KPIs at day 7

Frequently Asked Questions (for Clients)

Will this replace my support team?
No — it reduces repetitive tickets so the team can focus on complex cases, and anything nuanced or sensitive gets escalated.

What if the bot gives a wrong answer?
It’s restricted to official policies and sources, with refusals built in for anything undocumented. Logs get reviewed weekly and adjusted accordingly.

Can it track orders?
Yes, if an order lookup endpoint or e-commerce backend connection is provided. Otherwise, it guides customers to carrier tracking with clear instructions.

Is customer data safe?
Only order email and order number are collected, card details are never stored, and the client’s data retention policy is followed throughout.

Closing the Loop

The promise stays simple: instant answers for repetitive questions, strict boundaries for safety, and measurable savings that offset the fee. Build the system once, iterate lightly, and replicate it across clients using your own template.

Get Started With Manus Today

Create your Manus workspace now, use the system prompt in this guide, and aim to land 3 clients at $400/month to reach the $1,200/month target.

Appendix: Discovery Call Checklist (10 Minutes Max)

  • Ticket volume over the last 3 months (average/month)
  • Top 5 inquiry types, with estimated % per type
  • Current first response time and CSAT
  • Policy source-of-truth URLs
  • Escalation inbox and human SLA
  • Access to order lookup or returns portal
  • Go-live page(s) for the chat widget
  • Sign-off on refusal boundaries (what the bot should never say)

Appendix: Day-30 Optimization Ideas

  • Add proactive prompts on cart and product pages (“Ask about shipping times”)
  • Expand the international shipping matrix with clear timeframes
  • Clarify return exceptions (final sale, hygiene items)
  • Add synonyms and locale variants (“shipping” vs. “delivery”)
  • Load seasonal content — holiday cutoffs, pre-order ETAs

Final Note

The math works because most e-commerce support volume clusters around a handful of repeatable questions. With Manus handling 60–80% of those with grounded, non-hallucinated answers, a $400/month retainer becomes an easy “yes” for a small store — and three of those retainers gets you to the $1,200/month target.