How Australian Retail and Hospitality Businesses Should Handle Customer Enquiries with AI

Ask any café owner or retail manager what eats their time and the answer is almost always the same: answering the same questions, over and over. “Do you have tables this Saturday?” “Is the blue one available in a size 10?” “What’s your returns policy?” “Can I get a refund?” These aren’t hard questions. They’re just relentless.

The problem isn’t that the questions are complicated, it’s that they arrive across five different channels simultaneously. Phone. Email. Instagram. Google. The website chat widget you installed six months ago and forgot about. Keeping up with all of them during a busy trade is genuinely difficult, and letting any one channel go quiet for a few hours means lost customers.

AI can handle most of these enquiries without you touching them. But “set up AI for customer service” is too vague to act on. Each channel has different tools, different constraints, and a different kind of customer. This guide goes channel by channel, what to say, how to set it up, and which tools actually work for Australian retail and hospitality.

The channels that matter

Before getting into scripts, it helps to be clear about what you’re actually managing. Most retail and hospitality businesses deal with five main enquiry channels:

• Phone (missed calls), especially relevant for hospitality. People call to book or check availability, hit voicemail, and never call back.
• Email, booking confirmations, stock questions, returns. Lower urgency than phone but higher stakes if ignored for days.
• Instagram DMs, product availability, event info, general curiosity. Very high volume for retail. Response time expectations are short.
• Google Reviews, not strictly enquiries, but they need timely, considered responses. Left unattended, they do real brand damage.
• Website chat, the easiest channel to fully automate. Most visitors just want quick answers before they decide whether to buy or book.

Not every channel needs the same level of automation. Phone and Google Reviews benefit most from AI-assisted responses. Instagram DMs and website chat can be largely automated. Email sits in the middle, AI drafts save time, but you’ll still review before sending.

Phone: missed call text-back scripts

A missed call from a potential customer is usually a lost customer. Most people won’t leave a voicemail and won’t call back. But they will read a text.

The fix is a missed call text-back: an automated SMS that goes out within a minute or two of any missed call. It doesn’t need to be elaborate. It just needs to arrive fast and give the caller somewhere to go.

Tools that do this:

• OpenPhone (from ~$15 USD/month), has built-in auto-reply for missed calls. You can write a custom SMS template. Works well for small teams.
• Google Voice, free for personal use, limited for business. Good if you just want a basic setup without paying.
• Chime Labs, Australian-built. Designed specifically for local service businesses. Has AI-powered follow-up flows on top of the basic text-back.

Script: missed call text-back (café or restaurant)

Hi, it’s [CAFÉ NAME], sorry we missed your call! If you’re looking to book a table, you can grab a time at [BOOKING LINK] or reply to this message and we’ll get back to you shortly. If it’s urgent, call back during service (Mon–Fri 7am–3pm).

Script: missed call text-back (retail store)

Hi! You’ve reached [STORE NAME]. Sorry we missed you, we’re probably serving customers in-store. Text us your question here and we’ll reply ASAP, or visit us at [ADDRESS] during opening hours ([HOURS]).

Keep it to two sentences. Long SMSs don’t get read.

Email: AI-drafted responses for common enquiries

Email enquiries fall into a few predictable categories: booking requests, stock questions, returns and refunds, and general “how do I…” questions. For each one, you can have ChatGPT or Claude draft a professional reply in about 30 seconds, then review and send.

The trick is giving the AI enough context about your business so it doesn’t produce something generic.

ChatGPT prompt template for booking enquiries:

You are the customer service assistant for [BUSINESS NAME], a [café/restaurant/retail store] in [SUBURB], [STATE]. Write a friendly, professional email reply to this customer booking enquiry: “[PASTE CUSTOMER EMAIL]”. Include: confirmation that we’ve received their request, our availability for their requested date/time, how to confirm the booking (online: [BOOKING LINK] or by calling [PHONE]). Keep it to 3 short paragraphs. Sign off as [YOUR NAME], [BUSINESS NAME]. Australian spelling throughout.

ChatGPT prompt template for stock/product enquiries:

You are customer service for [BUSINESS NAME]. A customer has asked: “[PASTE THEIR QUESTION]”. Our current stock situation is: [DESCRIBE, e.g. “we have 3 left in navy, none in red, restock expected 15 July”]. Write a clear, helpful reply. If we don’t have what they want, suggest the closest alternative or offer to notify them when stock arrives. Keep it brief. Sign off as [YOUR NAME], [BUSINESS NAME].

For returns and refunds, don’t automate the full response. Use AI to draft it, then review before sending. Returns can involve ACL (Australian Consumer Law) obligations, and a poorly worded response creates legal exposure. More on this in the “what not to automate” section below.

Instagram DMs: auto-reply setup with ManyChat

Instagram DMs are where retail businesses get buried. The volume can be high, the questions are repetitive (“do you have this in store?” “what’s the price?” “do you ship to Perth?”), and customers expect a reply within the hour.

ManyChat connects to your Instagram account and lets you set up automated DM flows. You define trigger keywords, and it sends a configured response. Free plan covers basic auto-replies; the Pro plan (from ~$15 USD/month) adds more flows and Instagram DM automation.

How to set up a basic flow in ManyChat:

1. Connect your Instagram Business account under Settings → Connected Accounts.
2. Go to Flows → New Flow → Instagram DM.
3. Set a keyword trigger (e.g. “stock”, “available”, “in store”, “price”).
4. Write your auto-reply message (see template below).
5. Add a “Send to human” step if the conversation continues beyond the initial reply.

Script: Instagram DM auto-reply for “do you have X in stock?”

Hey! Thanks for reaching out 👋 We don’t have live stock visibility on Instagram, for the most accurate info, check our website at [URL] or pop us a DM with the specific product and size and we’ll check for you. We aim to reply within a few hours during business hours (Mon–Sat 9am–5pm AEST).

Script: Instagram DM auto-reply for shipping enquiries:

Hi! We ship Australia-wide via Australia Post. Orders over $75 get free standard shipping; under that it’s a flat $9.95. Express is available at checkout. We usually dispatch within 1–2 business days. Any other questions? Just ask!

Keep the tone casual, Instagram DMs are a conversational channel. Overly formal replies feel strange here.

Google Reviews: AI-written response templates

Responding to Google Reviews matters more than most business owners realise. Google’s own guidance notes that businesses that respond to reviews are seen as more trustworthy, and reviews with owner responses get higher placement in local search results. Responding also gives you a chance to correct misinformation before it calcifies.

Use this prompt in ChatGPT or Claude to generate your responses:

Prompt for responding to a positive review:

Write a warm, genuine response to this Google Review from [REVIEWER NAME]: “[PASTE REVIEW TEXT]”. Mention one specific thing they said to show you actually read it. Don’t be sycophantic. Keep it to 2–3 sentences. Sign off as [YOUR NAME], owner/manager of [BUSINESS NAME].

Example output:

Thanks so much for taking the time to leave a review, Marcus, really glad the brisket hit the mark. We’ll pass on your kind words to the kitchen team. Hope to see you in again soon., Jess, The Smokehouse

Prompt for responding to a negative review:

Write a professional, non-defensive response to this negative Google Review from [REVIEWER NAME]: “[PASTE REVIEW TEXT]”. Acknowledge their experience without admitting legal liability. Offer to discuss it further offline (include our email: [EMAIL]). Keep the tone calm and constructive. 2–3 sentences maximum. Sign off as [YOUR NAME], [BUSINESS NAME].

Example output:

Hi Trish, thank you for sharing this, and we’re sorry your visit didn’t meet expectations. We’d like to understand what happened so we can do better. Please reach out directly at hello@thesmokehouse.com.au and we’ll make it right., Jess, The Smokehouse

One firm rule: never argue in a Google Review response. Even if the review is factually wrong, a terse public reply damages you more than the review does. Use the offline redirection.

Website chat: FAQ bot with Tidio or Crisp

A basic website chat bot is one of the highest-value AI implementations for retail and hospitality, and one of the quickest to set up. Most visitors to your website want the answer to one of five questions: opening hours, location, price range, how to book, and whether you have a specific product or dish. A bot can answer all five without any human involvement.

Tidio has a free plan that covers basic FAQ chatbots for small sites. The paid plan (from $29 USD/month) adds AI responses powered by your business data. Crisp is similar, with a generous free tier and a clean interface.

The five FAQ responses to configure first:

• Opening hours trigger: “hours”, “open”, “close”, “trading” → “We’re open Monday to Saturday, 9am–5pm, and Sunday 10am–3pm. Public holidays may vary, check our Facebook page for updates.”
• Location trigger: “where”, “address”, “parking”, “find you” → “We’re at [ADDRESS]. Street parking is available on [STREET], and the nearest train station is [STATION] (5-minute walk).”
• Booking trigger: “book”, “reservation”, “table”, “appointment” → “You can book online at [LINK] or call us on [PHONE]. Same-day bookings are best made by phone.”
• Price trigger: “price”, “cost”, “how much”, “expensive” → “Our [main items] start from $[X]. Full menu/price list at [URL].”
• Returns trigger: “return”, “refund”, “exchange”, “broken” → “We accept returns within 30 days with proof of purchase. For details on our returns policy, visit [URL] or email [EMAIL].”

Set a fallback for anything outside these triggers: “Happy to help! That one’s best answered by our team, leave your name and email and we’ll get back to you.”

What not to automate

AI handles volume well. It handles nuance badly. There are specific situations where automating the response creates more problems than it solves:

• Formal complaints. A customer who is genuinely angry wants to feel heard by a person. An automated response to a serious complaint, especially one involving a bad experience, an injury, or a significant financial loss, will escalate the situation. Route these to a human immediately.
• Refund disputes. Under the Australian Consumer Law, consumers have specific rights for major failures (goods that don’t work as described, services not fit for purpose). If a customer invokes their ACL rights, a scripted response could expose you legally. Handle these manually.
• Anything involving a name, address, or payment detail. Don’t let a chatbot collect sensitive personal information. Route payment questions to your booking or checkout system.
• Medical or safety questions. If you’re in food and a customer is asking about allergens for a health reason, not just curiosity, make sure a human confirms. Incorrect allergen information can have serious consequences.

The clearest trigger for human handoff: if the customer is upset, if money is in dispute, or if the answer could create legal liability. Everything else is fair game for AI.

Putting it together: a channel decision map

Here’s a simple framework for deciding how each enquiry should be handled:

You don’t need all of this running on day one. The highest-value starting point for most hospitality businesses is the missed call text-back, it recovers bookings that would otherwise be lost and takes about 30 minutes to set up. For retail, Instagram DM auto-replies come first, the volume alone justifies it.

Add channels as you go. The goal isn’t to automate everything; it’s to make sure no enquiry goes unanswered.

Sources

• OpenPhone, business phone and auto-reply: openphone.com
• ManyChat, Instagram DM automation: manychat.com
• Tidio, website chat and FAQ bot: tidio.com
• Crisp, business messaging and chat: crisp.chat
• Chime Labs, Australian missed call automation: chimelabs.com.au
• Google Business Profile Help, “Interact with customers”: support.google.com/business/answer/3474050
• Australian Competition and Consumer Commission, ACL consumer guarantees: accc.gov.au/consumers/consumer-rights-guarantees

This post is part of the AI Customer Ops series, AI-powered scripts and systems for Australian small business customer communications.

More practical guides: How-To Hub →

More on retail AI: AI for Australian Retailers & eCommerce Hub →

📊 Compare AI tools side by side | 💼 Free resources & AI prompt packs

More on AI for hospitality: Hospitality & Food Service Hub →

📊 Compare AI tools side by side | 💼 Free resources & AI prompt packs