Travel Chatbot: Turning Enquiries into a Brief

The vagueness problem

In most industries an enquiry is specific and qualification adds context. In travel it is the reverse: the enquiry is a mood and qualification has to add specificity. That inverts the design — the bot is not filtering a list, it is building a specification.

The cost of getting this wrong is measured in consultant hours. An under-specified enquiry produces an offer that misses on price, dates or board basis, which produces a second exchange, a revised offer and often a third. Three rounds is how a travel enquiry dies.

The second factor is timing. Travel research happens in the evening, at weekends and during the long gaps in a working day, and enquiries cluster heavily outside office hours. A question answered at eleven at night is competing against nothing; the same question answered at ten the next morning is competing against three other agencies.

The five fields that make a brief

1. 1Dates, and how fixed they areThe most valuable single field, because flexibility is worth money and travellers do not volunteer it. 'The last week of July' and 'any week in July' produce completely different offers, and asking which one it is takes one message.
2. 2Party compositionAdults, children and their ages. Child ages are not a detail — they determine pricing bands, room configuration and whether a property is even eligible. This is the field most often missing from a brief and the one that most often forces a second round.
3. 3Destination, or the flexibility around itSome travellers name a country, others name a feeling. Both are workable, but they have to be recorded differently: a named destination is a constraint, while 'somewhere warm' is a brief for the consultant to fill.
4. 4Budget, with its basis statedPer person or total, and whether flights are included. An unstated basis is the single most common cause of an offer that lands badly, because a figure that looks generous per person looks impossible as a total.
5. 5The constraint that overrides everythingDirect flights only, a specific board basis, accessibility, a visa situation, travelling with a pet, a fixed departure airport. One of these usually exists and it eliminates most of the options, so it is worth asking directly rather than discovering it after the offer.

What the bot can answer immediately

These are the questions that otherwise wait until morning, and they are most of the volume. All are factual and safely indexable.

• Visa requirements as published, quoted from official material rather than summarised, with the caveat that rules change.
• What is and is not included in a package: transfers, board basis, baggage, insurance.
• Booking and deposit terms, payment schedule, and what happens if plans change.
• Cancellation and amendment rules, which are the questions travellers worry about most and ask least.
• Documentation: passport validity rules, whether a child needs their own document, consent letters for a child travelling with one parent.
• Practical destination facts: typical weather in a month, time difference, currency, whether a resort suits families or couples.

Where the bot must stop

• Visa outcomes. The bot may state published requirements; it must never assess whether a specific person will be granted a visa.
• Health, vaccination and medical fitness advice, which belongs with a medical professional and is regulated in many markets.
• Confirming a price as held. Travel pricing moves constantly, and a figure a bot confirms will be treated as a quote.
• Any statement that a booking is confirmed before the reservation system says so.
• Safety and political conditions at a destination, which should quote official advisories verbatim or not be answered.
• Insurance advice on coverage, exclusions and claims, which is a regulated conversation in most jurisdictions.

What the consultant should receive

The whole point of the intake is that a consultant can build one offer that lands. That requires structure, not a transcript.

• The five fields as structured values, so the queue can be sorted by departure date and by budget.
• Flexibility flags — dates flexible, destination flexible, airport flexible — which are what the consultant works with.
• A one-sentence summary in the traveller's own words, which carries the tone the offer should match.
• Anything already quoted in the conversation, so the consultant does not contradict it.
• The channel and hour of the enquiry, and whether the traveller has enquired before.
• Any constraint that eliminates options, flagged rather than buried in the summary.

What to measure

1. 1Complete-brief rateEnquiries reaching the consultant with all five fields present. This is the number that determines how many rounds an offer takes, and it is the one the flow exists to move.
2. 2Offers per bookingHow many offers a consultant sends before a booking. A falling number is the clearest evidence the intake is working, and it is measured in consultant hours.
3. 3Out-of-hours enquiry shareTravel skews heavily outside office hours. Measured separately, this usually justifies the project on its own.
4. 4Time to first offerFrom enquiry to the first priced proposal. In travel this competes directly against other agencies working the same enquiry.
5. 5Second-round rate by missing fieldWhich field's absence most often forces another exchange. In most agencies it is child ages or budget basis, and knowing which one lets you fix the flow rather than guess at it.

How Vexvon handles travel enquiries

The bot answers around the clock from your own material — your site crawled to as many as 4,500 pages, PDFs up to 25 MB for tour documents and terms, and bulk entries at up to 200 per request. Entries are typed into thirteen categories including pricing, campaign, policy and branch, so a seasonal promotion can be added and withdrawn without touching the booking terms.

Prices and availability are the values that should not be indexed, and do not have to be: your own system can be registered as an API tool the agent calls mid-conversation. Where the lookup fails, the defined behaviour is to say it could not be checked rather than to confirm a figure — which matters more in travel than almost anywhere, because a quoted price is treated as an offer.

Extraction runs on the conversation rather than a form, which suits an enquiry that arrives as a paragraph: phone numbers are detected through a five-stage process, free-form fields such as dates, party and budget are captured alongside, and a one-sentence summary is written to the record. A buyer-intent filter removes conversations that are not genuine enquiries before a lead is created.

The brief reaches the consultant group as a Telegram notification with a Start button; whoever presses it is assigned, and the record moves through a nine-stage pipeline with reminders checked every minute. Language is detected per message, which matters in a sector where the same agency serves travellers in three languages from one inbox.

Frequently asked questions

1. 1What should a travel chatbot qualify on?Dates and how fixed they are, party composition including child ages, destination or the flexibility around it, budget with its basis stated, and the one constraint that overrides everything.
2. 2Why do child ages matter so much?They determine pricing bands, room configuration and property eligibility. 'Two children' is not a brief; it is two different prices and possibly two different properties, and its absence is the most common cause of a second round.
3. 3Can the bot answer visa questions?It can state published requirements, quoted rather than summarised, with the caveat that rules change. It must never assess whether a specific person will be granted a visa.
4. 4Can the bot quote a price?Only from a live lookup, and never as a held figure. Travel pricing moves constantly and anything a bot confirms will be treated as a quote the agency has to honour.
5. 5Which metric shows the intake is working?Offers per booking. A falling number means the brief is complete enough to price in one pass, and it is measured directly in consultant hours.
6. 6When should a consultant take over?Once the five fields exist, and immediately for visa outcomes, medical questions, insurance coverage, destination safety and anything that reads as a complaint.

Count the rounds on ten recent enquiries

Take ten enquiries that turned into bookings and count how many messages passed between the first contact and the first accurate offer. Then find which missing field caused each extra round. In most agencies the same two fields account for nearly all of it, and asking for them at intake is a change that takes an afternoon.