Why AI must never guess a supplier rate

Ask a piece of software for a hotel rate and it will give you a number. Ask a travel operator and you will get questions first. Which season? Which room category? Which market is the guest from? Is this a weekday or a festival weekend? Who is the agency asking?

A rate is not a number. It is a season, a market, a relationship. An operator knows this without thinking. A model does not, unless someone puts that knowledge in front of it.

Some destinations are simply tricky to operate. Singapore is one: hoteliers change rates several times a day, and the big hotels layer blackout dates on top for the large conferences that run through the year. A rate sheet that was right at breakfast can be wrong by lunch.

And beyond the rates themselves there are the exceptions - innumerable ones - because what we deal with is people, and people challenge the system on a regular basis.

When software guesses a rate, the quote still looks clean. That is the dangerous part. The agent finds out it was wrong when the supplier confirms a different number, or worse, when the customer is already at the hotel. In thirty-one years I have watched agencies live and die on one thing: the person on the other side trusting that their quote is correct.

So NAMA has one rule it will not bend: nothing guessed. Every number in a NAMA quote traces back to the agency's own rate sheet, supplier or margin rule. If a rate is missing, the system says it is missing.

A missing rate is a phone call. A guessed rate is a lost customer.