Updated 2026-07-14 · 3 min read
«I already tried that: I put in a chatbot and my clients hated it.» We hear it every week, and the sentence hides the most expensive misunderstanding in the trade: calling two technologies by the same name when they resemble each other about as much as a photocopier resembles a receptionist. Let's separate them the most honest way there is: by showing you the same conversation with both.
The same client, two experiences
With a menu chatbot:
- Client: hiya, it was about the colour with Sara friday afternoon
- Bot: Welcome to Ana's Salon! Choose an option: 1 Book an appointment · 2 Opening hours · 3 Location
- Client: 1
- Bot: To book, go to this link: ana-salon.bookings.com
- Client: …but do you have a slot on Friday with Sara or not?
- Bot: I didn't understand your message. Choose an option: 1 Book an appointment…
The client wanted an answer; she got an answering machine with buttons. Typical result: she gives up and writes to the next salon.
With an AI assistant:
- Client: hiya, it was about the colour with Sara friday afternoon
- ClaudIA: Hi Marta! Of course: Sara has 4:30 and 6:00 free on Friday afternoon. A root colour is £45. Which shall I book you in for?
- Client: the 6 one
- ClaudIA: Done! Friday at 6:00, colour with Sara. I'll send you a reminder the day before. See you then!
The same scrappy question, no «option 1» — appointment closed in four messages.
The differences that produce that gap
1. Understanding vs. recognising. The chatbot looks for keywords inside a script; it derails on the way people actually write (abbreviations, context, two questions at once). The AI interprets the *intent*: «the colour with Sara friday afternoon» is four pieces of information (service, stylist, day, time of day) and it uses them all at once.
2. Real data vs. canned answers. The chatbot recites fixed text; the assistant checks your live calendar — it only offers slots that exist, it knows Sara doesn't work Mondays and that a balayage needs three hours. Without that, «book» means «off you go to another website», with the client leak you already know about.
3. Resolving vs. deflecting. The definitive measure: how many conversations end without you stepping in? The chatbot deflects (a link, «we'll call you», collecting details); the assistant books, moves and cancels appointments inside the chat, and when the subject is delicate, it hands it to you with the whole conversation.
When a chatbot is enough (honesty requires it)
If all you want is for somebody to give the opening hours and the address out of hours, an autoresponder or the WhatsApp Business away message do the job for little or nothing — we cover it in the WhatsApp guide. The jump to an AI assistant is justified when the goal changes from «reply with something» to «conversations that end in appointments»: at that point the chatbot isn't the cheap version of the solution; it's a different solution to a different problem.
And if you've already been burned by a chatbot: your clients didn't hate the automation — they hated the menus. When the conversation flows and the appointment gets booked, nobody minds who's typing on the other side. What they mind is that at ten at night somebody held Friday's slot for them.