Use Case 01 · The Variance Investigator

GOP missed. The question is why — and which causes will repeat.

A weekly flash posts a profit miss. By the time anyone can assemble the story across four systems, the week is already over. Aria does that assembly the moment the actuals land — and tells you which problems are recurring and which were noise.

The premise · Why it gets this far

The whole information flow is retrospective.

The P&L closes, the flash report posts, and then someone investigates — after the money has already moved. Every system reports its own slice after the fact, and no layer holds the cross-system history needed to see the whole picture. A variance reaches $88K because by the time anyone can assemble the story across rate, labor, parity, and the P&L, the week is gone.

Why the stakes are high

GOP margins are compressing across every property type — the squeeze is on cost, not revenue. And GOP is contractually live: it gates operator incentive fees and feeds management-agreement performance tests. A GOP miss isn't a scorecard entry — it's money, and eventually leverage.

What Aria is

It sits above your systems and reads them as databases.

Aria does not replace the PMS, the revenue system, the rate shopper, the labor system, or the people who run them — nor the asset manager whose accountability function it serves. It connects to what you already run, in read-only mode, and does the synthesis none of those systems does alone. Nothing is modified; nothing is actuated.

What the agent does

From a $88K mystery to a routed diagnosis.

The investigation runs the moment the actuals land — no one has to ask it to look. Each step is something Aria does, on this property, and reports back.

1
The trigger

A weekly flash posts: GOP is $88K under budget at one property. Aria sees it the moment the actuals land. Before a human opens a tab, the decomposition has started.

2
The first fork — rate or volume

Occupancy ran ahead of budget (78% vs 76%), so the rooms miss isn't demand — it's rate. ADR came in $14 under, concentrated on specific nights and one booking channel, because Aria reads realized rate by day and channel, not the blended number. The 'soft demand' branch is ruled out in seconds.

3
The rate sub-investigation

Aria reaches across systems: intended rate from the revenue system (was the discount deliberate?) and competitors' displayed rates from the rate shopper (did the market move, or just us?). Competitors held, the discount wasn't strategy, and one channel — Booking.com — is booking below BAR on weekend nights. Joined to realized bookings, the cost is ≈$47K.

4
The labor sub-investigation, in parallel

F&B labor ran 420 bps over. Aria pulls scheduled-versus-worked hours from the labor system and the banquet calendar from sales & catering. A light banquet week, but the schedule wasn't pulled down to match — labor high relative to the demand that actually materialized, worth ≈$13K.

5
The one-time charge

$28K in maintenance hits appears in work-order data. Aria checks the maintenance history, finds no prior occurrence, and classifies it: unplanned but genuinely one-off — separated from the two real findings.

6
The pattern test — what a busy team misses

Aria tests each finding against the property's history. The rate softness has recurred 4 of the last 5 weeks — a process problem, not luck. The labor overrun repeats on light-banquet weeks. The maintenance charge has not. Two recurring problems and one piece of noise.

7
The synthesis and the hand-off

Aria assembles a narrative readable in thirty seconds, every number traceable, then routes the next action to the person who owns it — parity to the revenue manager, scheduling to the F&B director, the recurring-vs-noise story to the asset manager. The experts still decide; the assembly is done.

What the asset manager reads
GOP missed $88K against a $640K target. Not occupancy — you ran 78% vs 76%. The miss is rate: ADR $14 under on weekend nights, isolated to Booking.com booking below BAR while the comp set held and intended rate was correct — a parity leak worth ≈$47K. F&B labor ran 420 bps over on a light banquet week. $28K in maintenance charges hit, but they're one-off. The rate leak has recurred 4 of the last 5 weeks — a pattern, not luck.
Where it goes

After a year of running, it stops waiting for the close — and starts forecasting it.

Reactive
Predictive
Explanation
Prevention
Per-finding
Learned model
What a year of joined history earns

From investigator to forecaster.

The Variance Investigator cleans up after the period faster and more completely than any human can. Its value compounds: a year of retaining and joining every source creates something none of the systems has — a learned model of the property and the portfolio.

1
The asset of record is the history, not the period

Twelve months of every source — rate, labor, banquet calendars, parity signals, maintenance, P&L — normalized into one timeline. Not better access; accumulated, joined history that can't be shortcut by buying the same connectors.

2
It forecasts the close instead of waiting for it

Aria reads the leading indicators it spent a year learning to connect and projects where GOP lands before the period closes. The 45-minute investigation becomes a 7-day-ahead forecast.

3
Explanation becomes prevention

“GOP is tracking $40K under — same rate pattern as prior periods, labor scheduled heavy against a light banquet week again. Here's what to change now, before the weekend.” The leak is caught before it happens.

4
The humans move up the value chain

None replaced — but the revenue manager, F&B director, and asset manager move from assembling and explaining the past to acting on a forecast. The backward-looking labor is absorbed so completely the work itself shifts forward.

See the Variance Investigator work.

Watch Aria decompose a live $88K miss across seven systems, separate the recurring problems from the noise, and route a diagnosis — in seconds.

View the live demo →Browse all use cases