Notes on running an AI front desk
Practical writing for service businesses evaluating or running AI customer engagement, including the parts vendors usually leave out. Grouped by topic; each post answers one question.
AI receptionists
What an AI receptionist is, what it should and should not answer, and how to judge one before customers meet it.
Seven myths about AI receptionists, and what is actually true
The beliefs that lead businesses to buy the wrong AI receptionist or avoid a useful one, each answered plainly, including the two that cut in opposite directions: it knows nothing until you tell it, and it is further along than you think.
· 6 min readEscalation keywords: the starter list every service business needs
A starter list for any business, additions for nine trades, why you should deliberately tune them to over-fire, and why testing a bare word tells you nothing.
· 6 min readYour AI receptionist's first week, day by day
What to read and what to fix on each of the first seven days, which kind of day-one failure is fine and which is not, and the three thresholds that tell you it is ready for more autonomy.
· 6 min readSetting your AI receptionist's tone of voice, with examples
Tone is a paragraph you change in one place, not a personality. The five things it has to cover, the nine phrases that make any assistant sound like a machine, and the sentence most tone paragraphs forget.
· 7 min readAn AI receptionist setup checklist: 15 things to have ready
Fifteen things to gather before an assistant answers a customer, grouped by what each one protects, with the five that are the real minimum and the three most businesses forget.
· 6 min readAn AI receptionist for a dental office: what it should and shouldn't answer
Insurance, pain and new-patient enquiries are most of a dental front desk's day. Why 'do you take my insurance' has three different answers, the one message that must never get an automated reply, and five things to try in a demo.
· 8 min readWhat is an AI receptionist?
Software that answers calls and messages from facts you approved, books from your real calendar, and refuses what it should not touch. What separates one from a chatbot, the four things it cannot do, and how to tell before you buy.
· 8 min readMissed call text back
Turning the calls you could not answer into conversations, without annoying anyone or breaking a rule.
Missed call text back for auto repair shops
What a ringing phone costs a repair shop, what the first text should ask a driver, and the two rules that matter most: a safety check that outranks booking, and a definition of missed that does not text people who misdialled.
· 6 min readMissed call text back for law firms: what the first text should and should not say
A missed call to a firm may be somebody in trouble. Why asking for the details of the matter by text is actively harmful rather than merely risky, the one case where an identifying text puts somebody at risk, and the script that handles both.
· 6 min readMissed call text back for salons and barbershops
Salon calls peak exactly when every chair is full. The first text, the five replies you will actually get, the calls you must not text back, and how to tell a recovered booking from a sent message.
· 8 min readRewriting your voicemail greeting for a text-first business
Fifteen seconds, three things, and one rule that matters more than the wording: never promise a reply speed you cannot keep. Five examples, what to leave out, and why a long greeting costs you the caller twice.
· 5 min readMissed-call text-back scripts that get replies
Seven copy-ready scripts with their character counts, the four rules behind them, and the one trade whose obvious script does not fit in a single segment and what to do about it.
· 6 min readMissed call text-back: what it is and how to do it legally
A text within seconds of an unanswered call, the five US rules that decide whether you may send it, the four numbers you must never text back, and why counting sends as recoveries is marking your own homework.
· 6 min readSpeed to lead and follow-up
Why the first minutes decide most leads, and how to follow up in a way people reply to.
First-touch vs last-touch attribution, explained with one customer
Two ways to credit the marketing that produced a customer, shown with one journey, with the arithmetic of what each would have you cut and why the honest answer is to keep both numbers side by side.
· 5 min readLead source tracking for a small business, without spyware
Tag the links you control, carry the parameters to the thing that creates the lead, and record nothing about people who never get in touch. The six traps that break attribution, and why cost per customer beats cost per enquiry.
· 5 min readFollowing up with new patient enquiries at a dental office
A new patient enquiry is worth years and arrives at nine in the evening from somebody messaging three practices. What the first reply should answer, what must never be asked by text, and the three-step follow-up that then stops.
· 5 min readSpeed to lead for med spas: the evening DM problem
Med spa enquiries arrive as evening DMs from people messaging three clinics at once, and the price question is a trap because nobody quotes in the same unit. What the first reply should do, and the clinical line that must not move.
· 6 min readSpeed to lead for home services companies
The extreme case: an urgent need, a valuable lead, and every person who could answer up a ladder. The two-question first reply, the number that is usually the real problem, and the three answers that are better than silence.
· 6 min readSpeed to lead: why the first five minutes decide the sale
What speed to lead measures, why the curve is steep, and the measurement decision that makes or breaks it: a clock that runs in opening hours rather than wall-clock time, because a weekend enquiry has not waited fifteen hours by Sunday.
· 7 min readBooking and reminders
Booking by text, keeping calendars honest, and the reminders that actually reduce no-shows.
Timezones and online booking: the daylight-saving bug nobody notices
Four different things go wrong on the two weekends a year when the clocks move, and they look like one bug. What each looks like from the front desk, why a database cannot catch the worst of them, and what to ask a vendor.
· 8 min readStaff rotas and availability: setting them up so online booking stays honest
A slot is a claim that a qualified person is free. The seven things it is computed from, the default that causes the most awkward mornings, and why preventing double booking has to happen in the database rather than in the software.
· 6 min readCancellation policy examples for service businesses
Five policies you can paste, when each one is the wrong choice, the four places the wording has to appear, and why a fee you cannot collect makes every other policy you have look optional.
· 6 min readHow to calculate your no-show rate, and what to do with it
The formula, the denominator most people get wrong, the four splits that turn an average into an action, and what a no-show actually costs once you count the gap as well as the slot.
· 6 min readCan an AI receptionist actually book appointments?
Three different things get sold as booking, and the difference between them is which one leaves the customer with an appointment. The question that separates them, and why a phone booking needs a different dialogue from a text one.
· 7 min readReducing no-shows: what actually works, and what just annoys people
Two reminders rather than four, the wording that turns a reminder into a rebooking, why the good version costs the same to send as the weak one, and the consent rule that decides whether you may send any of them.
· 7 min readBusiness texting
Numbers, templates, segments and the mechanics of texting customers at a service business.
Sending links and photos by text: what works and what gets filtered
Which links carriers distrust, how to write one a customer will tap, why a single curly apostrophe can double what a message costs, and the honest position on photos: Frontiva can receive them and cannot yet send them.
· 6 min readBusiness texting etiquette: 12 rules customers notice
Twelve habits that decide whether a business text reads as helpful or intrusive, why four of them are carrier rules rather than manners, and the one that quietly doubles what you pay per message.
· 8 min readUsing Twilio for small business texting: what to know first
Bringing your own carrier account means the number is yours and the registration is yours, and it also means nobody else will file it for you. What the platform does, what the tool on top must do, and the failure that looks like success.
· 5 min readSMS reminders vs email reminders for appointments
Confirm by email, remind by text, and know that both channels fail silently in different ways. The cost asymmetry that decides your reminder schedule, and the reply problem that catches most businesses.
· 5 min readTexting compliance
Consent, 10DLC, STOP and HELP, quiet hours and the records to keep. Plain English, not legal advice.
An SMS compliance checklist for service businesses
Twenty checks across six areas, the four that account for nearly every failure, what each failure actually looks like when it happens, and the order to do them in. Practical guidance, not legal advice.
· 7 min readConsent revocation: how customers can opt out, and how fast you must honour it
A customer can withdraw consent by any reasonable means, which is more ways than a keyword list covers. What the confirmation text may and may not say, the five places an opt-out has to reach, and why one opt-out must cover the whole business. Not legal advice.
· 6 min readTexting rules by state: why the strictest one is the one to follow
Several states regulate business texting more tightly than federal law, and the state that counts is the recipient's. Why one strict rule beats tracking fifty, what it costs you, and the five settings that implement it. Not legal advice.
· 6 min readSTOP, HELP, and the texts you are required to answer
The keywords US carriers expect every business texting program to handle, what your replies should say, and the mistake that leaves people opted out on one channel and reachable on another.
· 6 min readChannels and the shared inbox
Email, Instagram, WhatsApp, Messenger, web chat and SMS, in one place your team can actually work from.
Tagging conversations: a tag list a small team will actually use
A starter list of ten, the five rules that keep tags meaningful, the synonym problem that quietly destroys most tag lists, and what to do with a label whose count is zero.
· 5 min readSnoozing conversations: how to use it without losing people
Snooze, close and a follow-up task are three different tools and picking the wrong one is how a customer gets forgotten. The one rule that decides which, and the misuse that hides a debt instead of paying it.
· 5 min readShared inbox best practices for a front desk team
The failure is not slowness, it is ambiguity: three people read a message and each assumes another will reply. Ten habits that remove the ambiguity, and the one that only software can enforce.
· 5 min readCustomer records
The few fields a service business needs, keeping records clean, and why the timeline matters more than the form.
A CRM for a team of five: what to use and what to skip
Five people need a customer record, a lead list where every lead has an owner, and follow-up tasks. What to set up in week one, the four things to refuse, and how to give each person the least access that lets them work.
· 5 min readContact notes that help the next person, and ones that do not
Assume the customer will read it, because in some places they can ask to. The four things worth writing, the note that should have been a field, and the note that should have been a task.
· 5 min readCustom field examples by industry
Fields for twelve trades, the two-part test a field has to pass, and the field that is worse than nothing because it looks authoritative and is wrong the moment the customer drives away.
· 6 min readLifecycle stages for a service business: lead, prospect, customer, lapsed
Four stages, and why adding a fifth makes the data worse. What must move a contact mechanically rather than by judgement, how to set the lapsed threshold for your own trade, and why lapsed does not mean marketable.
· 6 min readCRM or shared inbox: which does a service business actually need?
An inbox organises messages. A CRM organises people. Most service businesses buy one, need both, and end up with the customer's history scattered across three tools.
· 6 min readReviews and reputation
Asking for reviews properly, responding to the bad ones, and what you are and are not allowed to send.
Responding to positive reviews: short, specific, human
A reply to a good review is written for the next person reading the profile, not for the reviewer. The two-sentence structure, what a regulated practice must never confirm, and why most five-star ratings need no reply at all.
· 6 min readGoogle Business Profile basics for a service business
Wrong hours on Google cost more customers than a missing blog post. The fields that matter, the one that gets profiles suspended, the section a competitor can write in, and a monthly routine that takes ten minutes.
· 5 min readRunning AI responsibly
Grounding, guardrails, prompt injection and audit trails, explained for the person accountable for what the AI says.
Sensitive topics by industry: what your AI should never answer
The six categories that apply to every business, the additions for twelve kinds of practice, how to phrase a handoff, and why the emergency check has to sit outside the score rather than inside it.
· 6 min readWhat an AI readiness test should cover before customers see it
Five scenarios, two of which must never be part of the score. Why a test that averages everything lets an emergency failure be outvoted, and why the price question has to be built from your own service list.
· 7 min readAutonomous vs approval mode: how to decide for each AI agent
Five pieces of evidence that should move an agent, the one measurement people use that is not evidence, why splitting by job beats choosing a level, and what going back does not mean.
· 5 min readWriting an AI use policy for a small service business
One page, eight questions, written for your team rather than a regulator. The template, the three clauses people write without having the controls to back them, and the review that keeps it true.
· 6 min readBy industry
How an AI front desk fits the specific workflows of one kind of business.
An AI front desk for medical practices
Logistics answered at nine in the evening, every clinical question refused before an answer is attempted, and the two questions that account for most of a medical phone line. Plus the reminder wording that is itself a disclosure.
· 6 min readAn AI front desk for restaurants: reservations, catering and the questions nobody has time for
For a restaurant the value is deflection, not booking. The questions worth answering automatically, why an AI must never tell a guest a dish is safe, and the one enquiry at 6:45pm you cannot afford to miss.
· 7 min readAn AI front desk for gyms and fitness studios
A joining decision has a window measured in hours, and the enquiries arrive when the floor is busiest. What one recovered trial is actually worth, and the clinical question this trade does not think of as clinical.
· 6 min readAn AI front desk for auto repair shops
Wait or drop off is the question that decides the booking. The five vehicle details to capture before anybody picks up, the safety questions that must never get an answer, and why a repair price quoted by text becomes an argument at the counter.
· 6 min read