The tables everyone fights over.

In the first letter I made a claim that fits in five words: the seat is the product.

Then I spent three weeks talking about chatbots and confessions.

This week, the receipts.

Here is what a booking form knows about a guest: a date, a time, a number.

Here is what a room knows about them, if you let them walk it before they book: what they looked at first. Where they lingered. Which table they spun the camera round to check the view from. The one they nearly chose and then didn’t.

Multiply that by every guest who has opened a venue on RAYN since February and you get a number that still surprises me when I say it aloud:

Thousands of signals of intent, captured before a single booking was made.

The industry has never had this kind of behavioural data at this point in the booking journey, for the simple reason that the booking flow never gave the guest a way to reveal it.

A time slot has no opinions.

A room is nothing but opinions.

So what do the opinions say?

01 — Where guests look vs where they book

Which zone gets the most attention?

Which zone gets the most bookings?

And are they actually the same place?

02 — The table nobody picks

Which table gets repeatedly viewed, rotated around and inspected, but almost never selected?

And what’s sitting next to it?

The pass? The toilets? A pillar? A draught?

03 — The time before choice

How long does someone explore before choosing?

And does spending longer in the room mean higher intent or higher abandonment?

Every operator I’ve shown these numbers to has done the same thing: nodded, and then named the table.

They already knew.

The floor manager has always known which two-top gets asked to move, which corner the regulars angle for, which terrace table is a fight on Fridays.

(Ask the manager. He’s off on Tuesdays, but ask him.)

What they’ve never had is that knowledge in a form that can be counted, priced, or acted on at 11pm when nobody’s on the floor.

Instinct doesn’t scale. And it doesn’t stay when the manager leaves.

I’ll add the admission this letter owes.

For the first four months, our own recommendation engine ranked tables using my assumptions about what made a good one: view, quiet, the usual operator’s prejudices.

Guests had been telling us otherwise the entire time, in clicks and hovers and abandoned choices.

We simply weren’t listening to them yet.

In July, we let the data overrule me.

The engine now learns which seats guests actually want from what guests actually do.

It is a humbling thing to be out-argued by thousands of signals.

And the recommendations got better the moment I lost.

Why does any of this matter beyond a founder’s ego?

Because the industry prices time and guests value place, and the distance between those two facts is money left on the table.

Literally, on specific tables.

If you know which eight seats are fought over at eight on a Friday, you know something worth pricing.

If you know which four seats are quietly avoided, you know something worth fixing before you build a menu around covers you’ll never seat there.

And if you know a guest looked at the terrace three times before settling for indoors, you know what to offer them next time before they ask.

The seat was always the product.

What’s new is that the product has finally started talking.

Next week: your reviews are a dataset, not a reputation problem. What 2,000 words of guest complaints know about your Thursday service that your GM doesn’t.

— B

Disagree with any of this? Message me. I answer.

Bhrij Patel is the founder of RAYN, an AI-native hospitality intelligence platform, built after years operating restaurants from quick service to fine dining. For an honest, vendor-independent read on where AI fits your operation, ask about the AI Readiness Audit