Free tool · Indicative signal · Nothing stored
Your guests already told you where the leaks are.
Paste your public reviews. The Outside-In reader classifies what your guests actually said — speed, consistency, value, staff, cleanliness, repeat demand — and maps each signal onto the operating-health domain it belongs to. Evidence-first: no theme without a verbatim quote from your own reviews.
Processed transiently on request — the text is never stored, and the read appears only on your screen.
How the read works
Reviews are the one operational dataset every restaurant already publishes. Guests narrate your speed of service, your consistency between visits, how your pricing lands, how your team carries the room, and whether they intend to come back — years of floor truth, in their words. The Outside-In reader classifies that text against a fixed six-theme taxonomy and groups the findings under the five operating-health domains the GGB Command Matrix measures.
The discipline is in what it refuses to do: it reports no theme without verbatim evidence from your own corpus, it never invents a number, and its findings are labelled exactly what they are — an indicative signal from public reviews. The classifier reads; the mapping from theme to domain is fixed and ours. Pair the signal with the five-domain Command Report for the measured half of the picture.
The honest print
- What happens to the text I paste?
- It is read once, transiently, to produce your classification — then discarded. Nothing you paste is stored, attached to you, or used for anything else. The read appears only on your screen.
- Where should I get the review text?
- Open your Google Maps, TripAdvisor or Zomato listing and bulk-copy the recent reviews — the raw text is fine, formatting does not matter. More recent text gives a sharper current read.
- How does it decide the themes?
- A classifier reads the corpus against a fixed six-theme taxonomy (speed of service, consistency, value perception, staff, cleanliness, demand signals). The mapping from theme to operating-health domain is ours, fixed and published — the model classifies; it never invents the read. Every theme must carry verbatim evidence from your own reviews, or it is not reported.
- Is this a score?
- No. It is an indicative signal from public reviews — classification, not measurement. It never enters your Command Report score. What it does brilliantly is tell you which domain your guests are already complaining about, before your P&L shows it.
- Why is it capped?
- Three reads per hour per visitor, sixty per day sitewide — honest capacity limits on a free AI tool so it stays fast and free. Nothing you paste is queued or retained when capacity is reached.
The reviews show where guests feel it. The books show what it costs. The founder reads both — 28+ years of launch, turnaround and multi-outlet control.