Skip to main content

When to flag something

Three flags, three actions — thumbs-up for notable wins, thumbs-down for not-quite-right, and a direct call for revenue-loss or guest-impacting issues.

Three different signals, three different responses. Knowing which is which saves your team and ours from chasing nothing.

Three flags, three actions

"Something happened that I want to remember."

A call that was funny, kind, or noteworthy. Something a teammate should see.

Action: thumbs-up the card. Optionally share the card URL internally.

"This call wasn't quite right."

A detail Foodie got wrong, an off-brand response, an unexpected hedge.

Action: thumbs-down the card. Fill in the form. Move on — we see it within hours.

"This is going to be a problem."

A pattern across multiple calls. A complaint that's escalating. A booking system mismatch causing real revenue loss.

Action: call or text your onboarding contact directly. Don't wait for the debrief.

What "red" looks like

  • Same misconfiguration costing you bookings (Tabule Midtown's $3K/month phone-order drop is the canonical example).

  • Multiple guests in a week complaining about the AI specifically.

  • A reservation conflict (two guests booked into one slot) that traces back to Foodie.

  • A safety issue — a guest reported feeling pressured by Foodie. (Vanishingly rare. Always treat as red.)

What's NOT a red

  • One guest complaining "I hate AI." That's a yellow at most.

  • Foodie misheard a single name. Yellow.

  • A guest tried to order takeout and Foodie transferred them. That's working as designed (assuming you've set transfer for takeout).

  • A booking that seemed off until you cross-checked Guest Center and it was fine. Move on.

How to triage internally

If you're a manager or owner:

  • Ask staff to thumbs-down anything yellow. Don't ask them to escalate every call to you.

  • Once a week, scan the thumbs-down feedback for patterns. Patterns are reds.

  • Trust the AI to handle most things; trust your team to flag what doesn't fit.

A note on AI behavior in general

Even with the same configuration, AI systems will occasionally produce a response that surprises you. Most are one-offs. Some are patterns. The signal vs. noise question is the right one. Trust the thumbs-down + pattern matching to separate them.

When to call vs. message

  • Phone call: anything that's costing you revenue right now or affecting a guest right now.

  • Text: anything you noticed, with no time pressure. We'll respond within hours.

  • Email or thumbs-down: documentation, pattern flags, observations, suggestions.

Did this answer your question?