Meridian canonicalises hotel data into a typed, traceable operating layer — so every brief, signal, and comparison carries its basis, lineage, and limits.
Most hotel groups already have dashboards. The problem is not that the numbers are absent. The problem is that the numbers do not always mean the same thing across properties, systems, regimes, and reporting bases.
A "channel mix shift" means little unless Meridian knows what it knows. Every comparison Meridian renders carries its conditions explicitly.
Meridian does not produce figures. It produces typed objects. Each value carries the metric definition, canonical line, mapped local accounts, cohort frame, regime, window, and declared limits that determine whether it is safe to use in a comparison.
When a figure cannot be rendered with a complete basis, Meridian withholds it and names the reason. The withholding is not a warning. It is the product.
A number is only useful to an AI system if its basis, lineage, comparison frame, and limits travel with it.
Meridian starts at portfolio level. It positions each property against its closest operating cohort — not ranked by KPI, but placed by operating pattern. Properties that move away from cohort consistently are flagged.
Persistent movements enter a governed signal queue. Signals are not alerts. They are typed claims with evidence, confidence, persistence, and declared limits. The signal queue is ranked by persistence, materiality, and inspectability — not by algorithm confidence alone.
Meridian does not ask which property is good or bad. It asks which property is behaving unlike its operating cohort.
Every signal in Meridian expands into its recipe, evidence, comparison basis, persistence record, and noise checks. The visitor sees why the signal fired, what evidence supports it, and what it explicitly does not infer.
Meridian does not claim campaign intent, pricing cause, guest behaviour, or profitability unless the recipe supports that claim. What it does not infer is declared in the signal, not omitted.
No guest intent, campaign performance, or channel profitability. Meridian identifies the movement and routes it. It does not explain it.
Meridian uses USALI 12th as the canonical operating structure. Every metric has a home: a canonical schedule line, a declared basis, a set of local accounts that map into it, and a set of comparison cuts that are valid at that line.
When a metric is derived rather than native — Food Cost %, for example, is not a USALI line but a ratio computed from two lines — Meridian declares that. The basis is always in the object. The number is never bare.
If outlet mapping, account mapping, currency basis, or feed completeness is insufficient, Meridian withholds the comparison and names the reason. When the analyst resolves a mapping, the decision is recorded with full lineage. Every resolution is traceable in Audit.
Resolving a mapping closes the gate and makes withheld comparisons renderable. Every decision is recorded with lineage in Audit.
When mapping is incomplete, Meridian withholds comparison and names the reason. The withholding is surfaced in signals, reports, and the USALI explorer.
It is the semantic and intelligence layer that makes hotel operating data comparable, inspectable, and safe to use — for the humans and AI systems that need to reason on top of it.
These five posts make the argument in sequence. Together they describe the problem Meridian is the answer to.
No pitch deck. No sales process.
A guided walkthrough of the layer, applied to your operating context.