Make the recommendation reviewable
Show the current rate, proposed change, market context, rationale and guest-price consequence in one decision packet.
← Selected work
Live operating-system reconstruction
A pricing recommendation is not yet a decision.
Revenue Decision Lab is the closed-loop pricing system I built inside the short-term-rental business I run. It connects daily evidence to human approval, execution tracking and the learning that follows.
Public edition
The demo uses fictional properties and synthetic prices. Decisions stay in your browser and cannot touch a live pricing or booking system.
Interactive artifact
Decide. Track. Learn.
Approve or decline a synthetic recommendation, inspect its decision record and close the loop with a booking or expiry outcome.
Synthetic, browser-local demo
The decision lifecycle
Require an explicit decision
Approve, decline or hold. Recording a choice does not silently change a live system, and every write remains human-controlled.
Carry the hypothesis forward
Preserve what should happen, the success condition and the counterfactual before the outcome is known.
Turn the outcome into evidence
Review after 24 and 72 hours, then at booking or expiry. Record a confidence-rated learning instead of relying on memory.
What it demonstrates
The useful system is not the dashboard. It is the operating memory.
A clean recommendation can still produce a poor decision if it ignores booking pace, gap shape, guest-facing discounts or the cost of waiting. The lab brings that context together before action.
More importantly, it retains the original hypothesis. Each approved, declined or held recommendation becomes a small experiment that can be reviewed against its own success condition and counterfactual.
Over time, the decision ledger becomes a reusable evidence base by property, lead time, gap type and intervention size. That is what makes the operating rhythm improve.
Built from a real operating need, reconstructed with synthetic data and published as reusable code.