The short answer

A mention names the hotel, a citation links a source, a recommendation suggests the property for a decision, and a booking is a completed transaction recorded by the booking system. These outcomes can overlap but are not interchangeable. A website visit or telephone inquiry is also distinct from a confirmed booking.

Classify the answer before counting it

An answer may mention a hotel while citing an OTA, or cite a hotel policy without recommending a stay. Preserve the destination URL and recommendation language so the classification can be reviewed.

A reference to a hotel chain does not necessarily identify a specific property. Use entity-matching rules and record uncertain cases rather than forcing them into a positive count.

Connect commercial outcomes with consent and evidence

If booking analytics or CRM records are used, define the attribution method and its limits. A referral can provide evidence for the visit it records; it does not capture every earlier interaction in the guest journey.

Phone requests submitted to Arrow’s shortlist pages are audit inquiries for Arrow AI. They are not hotel bookings and should never be reported as guest revenue.

What to check

  1. Define each outcome before calculating a metric.
  2. Keep source destinations and entity matches inspectable.
  3. Separate audit requests from guest inquiries and completed bookings.

A practical example

An illustrative answer mentions a hotel and links an OTA; the reader then submits an Arrow audit request. The report may record those three different events, but none proves a hotel booking.

Sources and further reading

These steps are Arrow AI's implementation guidance. The example is illustrative; it is not a measured customer result. Publication and source access do not guarantee an AI recommendation.

Apply this to your website

Identify the source gaps and decide what to improve with Arrow's free audit. Review what Arrow GEO measures before choosing a reporting scope.