The 20 Questions You Answer Every Day, Answered Once

7:12am, Instagram DM. "Hi! How much is a full groom for a cavoodle? And do you have anything this Saturday?"

8:41am, text message. "Morning, sorry to bother you. What do you charge for a spoodle full groom, is nails and ears included or is that extra? Also are you open Saturdays"

Two people, 89 minutes apart, asking the same question in different words. You will answer both from the bath with one glove off, typing the same three facts twice.

You are not short of time. You are short of somewhere to put the answer.

Client FAQ automation for a pet business is not a chatbot talking to your clients. It is the 20 questions you already answer every week, written properly once in your own words, then put where a client hits them before it occurs to them to message you. The AI part is narrow and real: it turns the replies already in your phone into a clean answer set in an afternoon, rather than you writing 20 answers from a blank page on a Sunday night.

Client FAQ automation for a pet business is not a chatbot

You are good at answering. That is precisely the problem.

A fast reply feels like good service, and the first time it is. Typed for the 36th time this year it is a tax you pay because the answer lives nowhere but your thumbs. Answering is something you do. Having answered is something you keep.

Here is the test. Leave your phone in the car for a working day: how many of those 20 questions answer themselves? Your honest number is probably one, and it was an accident.

The 20 questions, in the words they actually arrive in

Take Steph, a two-groomer salon in Sydney's Inner West. She is a worked example we follow through this series, with numbers you would recognise rather than a real client's books. She scrolled back three months of messages and pulled out the 20 that keep recurring.

Not 20 topics. Twenty questions, phrased the way they landed. "Cancellation policy" is a heading; "sorry, something's come up, can I move it?" is a question. Answer the question and it reads like you. Answer the heading and it reads like terms and conditions.

Booking and availability

• "Do you have anything Saturday?"
• "How far out are you booked?"
• "Can I get in this week at all? He's pretty bad."
• "Do you do late afternoon? I don't finish till four."

Price and what is included

• "How much for a groodle?"
• "Is that with nails and ears, or is that extra?"
• "Do you do just a bath and blow dry?"
• "Why is it more than last time?"

What to bring and how to prepare

• "Do I need to bring anything?"
• "Should I feed him first?"
• "Do you need her vaccination card?"

Your dog, and how the day goes

• "She hates the dryer. Is that going to be a problem?"
• "He's 11 and has never been groomed. Is he too old to start?"
• "Please don't shave him. Can you keep him fluffy?"
• "Can I stay and watch?"

Payment

• "Can I pay on the day?"
• "Card or cash?"

Cancelling, moving, and running late

• "Sorry, something's come up. Can I move it?"
• "What happens if I have to cancel?"
• "We're 15 minutes behind. Is that alright?"

Yours will differ by three or four: a daycare gets "does she have to be desexed", a walker "do you go out if it's raining". The six groups hold.

Write each answer once, from the replies you have already sent

You are not starting from nothing. You have written every one of these answers dozens of times, and they are sitting in your sent messages. That pile beats anything a chatbot would invent, for one reason: it is already in your voice.

Teaching a tool your prices, your policies, and the way you write is its own job, earlier in this level. If you built it already, point this task at it.

Before you paste a single message, strip every name, phone number, address, and anything about a dog's health. You want the shape of your answers, not your client list, and what never goes into an AI chat is worth a read first. The Office of the Australian Information Commissioner recommends organisations do not put personal information into publicly available generative AI tools at all, because once it is in there it is very hard to track and potentially impossible to remove.

It knows what you show it and nothing else, which is what context means. Show it your replies, and this instruction turns them into an answer set.

Steps three and four earn their keep. Your sent folder disagrees with itself, because your notice period changed in March and half those messages predate it. Asked to be helpful, a chatbot smooths that over and builds your answer set on a policy you abandoned. Told to stop and ask, it hands you the decisions you have avoided.

If the cancellation answer comes back mostly brackets, the policy is the missing piece: write it, then get it checked properly.

Where your FAQ answers should live, ranked by how much work they remove

Four destinations, nowhere near equal:

1. Your booking page. This is the one that works. It answers the question at the moment the client has it, intent already switched on, and no message gets sent. Two thirds of the 20 belong here: the price beside the service, the "what to bring" line above the confirm button, the notice period in the terms they tick.

2. A pinned story highlight. Second, because it is where the DMs start. Six cards, one question each, at the front of your profile. Somebody about to type "how much for a groodle" sees the answer directly above the message box. Not everyone looks. Enough do.

3. An auto-reply. Third, and doing less than it appears to. It buys you an hour and covers the top two or three, then the conversation carries on anyway. Worth having, not a fix.

4. A printed card at the desk. Last for volume, first for the awkward ones. Almost nobody's phone is out at the desk, so the volume is tiny. What it handles is the late cancellation and the surcharge, and it makes those the salon's rule rather than your preference, which is the difference between a policy and an argument. It also means Deb can point at it on a Saturday without ringing you.

One document is the source; the four are copies, and copies go stale. Change the document, then re-run this.

The line: an auto-reply that answers, not an AI that answers for you

An auto-reply carrying a sentence you wrote and read is fine. It is a sign on a door. What is not fine is anything that generates a fresh answer and sends it before you have read it.

And never let it:

• Send something you have not read.
• Delete something you have not backed up.

Those two lines come from the never list, and this is the post where the first gets tested, because the whole point of an answer set is that it goes out without you. Fine, as long as "without you" means without you retyping it. Write it, read it, then let it repeat. Never let it write and send in the same motion.

The client questions that stop arriving once your confirmations do

A few of the 20 are not knowledge gaps. "Am I still booked in?" and "what time was I again?" are questions your system created. The client is not asking for information; they are asking for reassurance.

Those stop the day confirmations and reminders go out on their own. In Petboost, booking confirmations and reminders are set up once and sent by the automation rules you pick: a rule, not an AI writing something new each time. The evidence on reminders is better than most expect, and this is the cheapest destination of the four: the answer arrives before the question forms.

The honest limit, since somebody always asks: our own assistant, Grazi AI, answers you about your own business. It is read-only, it talks to owners and their teams, and it does not talk to your clients. No AI in Petboost writes a client a reply and sends it, and that is deliberate.

Three of the 20 that should never get a canned answer

"He's 11 and has never been groomed. Is he too old to start?" You can write a good general answer about introducing an older dog. You cannot write one that covers that dog, and a template sounding confident about it is worse than no answer.

Same for "she hates the dryer", and for "please don't shave him", where any template promising something about a matted coat becomes the promise a client holds you to. Mark those three as "answer, then a real conversation", and let the written part do only the first half. Where AI must never go in pet care, meaning anything clinical, behavioural, or incident-related, is a post of its own.

The habit applies to all 20 regardless: read every answer as though a client will hold you to it, because one will. Why it gets things wrong, and the ten-second check.

What this actually reclaims for you

Say 14 of the 20 arrive in a normal week. Three minutes each, once you count reading it, checking the diary, and typing your notice period out again. That is 42 minutes a week, and it never feels like 42 minutes, because it arrives in three-minute pieces with your hands wet.

The answer set is one afternoon: roughly 90 minutes with the two prompts above, plus an hour putting the answers onto the booking page and into a story highlight. The messages will not stop, so say half of them stop arriving because the answer got there first. That is 21 minutes a week, around 17 hours across a working year, and the afternoon pays for itself inside two months.

The 42 minutes was never the real cost, though, and the 21 is not the real win. The win is that the answer stops depending on you being free. Deb points at the card. The 7:12am DM reads it before you are awake. And what they get is your best version, written sitting down, instead of the one you thumbed out with one glove off.

If you want to hear an answer set working out loud, ring ours: Call Emily on 1800 291 005. She is our line rather than something you can point at your salon, and she answers our most-asked questions all day without tiring of them.

Next in this level: voice note to written handover note, the same move applied to the notes your groomer writes in the car park.