Method & limits

Data collection

Booking.com search pages, rendered in a browser and captured in full, one page per stay-night. A city search was collected through the rendered browser for 16–30 September 2026, every night: a full search-results page for each night.

Basis: 2 adults · 1 room · 1 night · lowest displayed rate · AUD incl. taxes and charges.

Competitive set

47 Canberra hotels and serviced-apartment operators rated 3.5–4.5 stars within 7 km of the centre

Direct peers

11 hotels rated 4–4.5 stars within 3 km — the direct comparison set a guest sees beside Rex

Data-quality controls

An initial automated pull mis-resolved the search destination and returned unusable results. All figures were re-collected via the rendered-browser method, and Rex’s availability was cross-checked on the property page.

Rex’s review count moved from 1,992 on 8 September 2026 to 2,007 as of 16 September 2026. The movement confirms that the property is trading; it does not measure occupancy.

Definitions

Price percentile

The share of the competitive set priced below Rex that night. Guest-score percentile is the share with a lower Booking.com review score than Rex’s 8.6 — a separate quality comparison.

Median and quartiles

The median is the middle competitive rate. The lower and upper quartiles mark the middle half of returned rates.

Target (65th)

The comp-set rate at Rex’s 65th quality percentile.

Gap to target

The quality-aligned target minus the current displayed Rex rate, in AUD. A positive gap means the target is higher.

Unavailable share

Each night’s returned competitive rates compared with the window’s fullest night (33 rates), not the 47-property roster — an availability index, not an occupancy measure. It can include closed inventory or unavailable offers, not just rooms sold.

Near and far windows

The far window has lead time greater than 8 days; the near window covers the earlier bookable nights.

Scenario tails and midpoint

P10, P50, P90 are the lower tail, midpoint and upper tail of the supplied simulated outcomes. The annual change distribution is supplied separately; subtracting the reported revenue percentiles does not reproduce it.

Limits

• Advertised rates are not rooms sold. Displayed prices cannot establish realised average room revenue, occupancy, conversion or channel mix. Rooms-left messages describe displayed inventory, not the hotel’s total unsold rooms.
• Price elasticity is assumed, not measured. It describes how bookings might react to a price change. The simulation is a scenario, not a forecast; cancellation terms, room mix, events and competitor responses can change the result.
• 15 consecutive nights, 0–14 days out; close-in rates describe the booking window, not the season. This snapshot cannot establish a seasonal or annual pattern. The far window is only the later part of this observed period.
• Room count is disputed: 121 / 157. Both counts are modelled; the operating inventory must be verified before adopting revenue estimates.
• The lowest displayed room and cancellation term can differ by hotel and night. The comparison is the offer visible to a guest, not a like-for-like room valuation.
• Unavailable listings may reflect closed inventory, channel restrictions or missing rates. The sold-out label is not proof of demand.
• The comparison roster has 47 properties, but at most 33 returned a rate on any night; the unavailable share is measured against that 33, so fourteen properties that never returned a rate are not counted as unavailable.
• Annual model distributions are supplied results. The source does not contain the sampling process, correlations or a reproducible annual simulation; those outputs cannot be independently validated here.

Prior sample: context only

8 sampled stay-nights between 9 Sep and 12 Dec 2026, collected on 8 September 2026. The earlier sample placed Rex at an average 44th price percentile, spanning 5th to 83rd. It provides context for erratic positioning; it is not the basis of the fortnight recommendation.

Scroll horizontally for further columns.