Outside In · Public Growth Signals

Dishoom marketing teardown: AI search readiness and booking clarity

21 September 2026 · Evidence-led outside-in review

Dishoom's public content is unusually rich for hospitality, and its llms.txt file shows deliberate thinking about how AI systems represent the brand. The sharper question is whether every guest receives equally clear, current information for the café they intend to visit.

Scope and limitation: public sources only. This review does not assess revenue, campaign performance, customer data or internal strategy. It separates what was visible on 21 September 2026 from hypotheses that require internal validation.

The wider lesson

Structured information creates value only when humans and machines receive the same current answer.

Observed 01

The brand publishes an llms.txt file

Dishoom provides structured guidance for AI tools and explicitly warns against using generic or outdated booking rules. That is an advanced public signal of information governance.

Observed 02

Booking rules vary by café

The estate includes location-specific booking and walk-in arrangements. Variation can be operationally sensible, but it increases the importance of clear location pages.

Possible implication

A hypothesis, not a verdict

Dishoom has already identified the risk of generic answers. Applying that discipline to every human-facing location journey could reduce expectation gaps without changing the underlying operating model.

Alternative explanation

A credible reason the signal may be benign

Guests may already reach the correct location page from maps, booking links or CRM messages. The variation itself may be part of a deliberately flexible service model, not a content failure.

The smallest useful test

Learn before committing to a larger fix

Ask first-time visitors to answer three questions from a location page: can I book, when might I queue, and can I order delivery here? Track confidence and errors by café.

What I would check internally

  1. Does each location page lead with its own booking rule?
  2. Are AI guidance, maps data and on-page copy updated from one source?
  3. Can delivery availability be answered without a redundant postcode journey?
  4. Who owns policy changes when operations vary by café?

Public sources checked

Correction policy: public pages change. If a source has moved or the context is incomplete, email hello@faisalconsulting.co.uk and I will review it.

Download the two-page executive snapshot

The evidence, alternative explanation and smallest useful test in a meeting-friendly format.

Part of Outside In: Public Growth Signals: what customers, search engines and AI platforms can see, what it may mean and what should be tested before drawing conclusions.